Celerra Network Server Release Notes 6.0.70.4
Content
EMC® Celerra® Network Server
Version 6.0.70.4
Release Notes
February 28, 2013
P/N 300-009-958
REV 18
These release notes contain information on these topics:
Product description ................................................................................... 2 New features and changes ........................................................................ 2 Known problems and limitations .......................................................... 27 Problem prevention ................................................................................. 50 Technical notes ......................................................................................... 56 Environment and system requirements................................................ 78 Documentation ......................................................................................... 79 Installation and upgrades ....................................................................... 81 Fixed problems ......................................................................................... 88 Version 6.0.70.4......................................................................................... 89 Version 6.0.65.2......................................................................................... 98 Version 6.0.60.2....................................................................................... 115 Version 6.0.55.3....................................................................................... 129 Version 6.0.51.6....................................................................................... 141 Version 6.0.43.1....................................................................................... 189 Version 6.0.43.0....................................................................................... 189 Version 6.0.42.3....................................................................................... 190 Version 6.0.41.4....................................................................................... 222 Version 6.0.41.3....................................................................................... 225 Version 6.0.40.8....................................................................................... 301 Troubleshooting and getting help ....................................................... 326
1
Product description
Product description
The EMC Celerra Network Server is a network-attached storage (NAS) server. It enables clients on a network to store and retrieve files over the network. These release notes contain information about version 6.0 of Celerra Network Server. Check EMC Online Support at http://Support.EMC.com for updates to this document. Updates are made when a new version or patch is released, or when new information is discovered.
New features and changes
Version 6.0.70.4
NS gateway support for new Symmetrix Enginuity 5876.159.102. • • NS gateway supports Enginuity 5876.159.102 running on any of the VMAX platforms (10K, 20K, and 40K). NS gateway software features such as TimeFinder® and SRDF®, which are supported on the earlier Symmetrix Enginuity, will be supported on 5876.159.102 Enginuity. Software features such as FAST™ and Federated Tiered Storage (FTS) are not supported with 6.0.70.4. Thin BCV (Business Continuance Volume) disks are supported on NS gateway. NS gateway inherits all hardware and software restrictions imposed by VMAX 10K, 20K and 40K, and Enginuity 5876.159.102. VP Compression is supported on NS gateway. It is recommended that you create a user defined pool for the compressed LUNs so that consistent performance can be provided for the file systems. VP compression must be done from the Solution Enabler host. I/O Quota is supported on NS gateway. It is recommended to create a userdefined pool for the LUNs that have the same I/O Quota configuration so that the user-defined pool can provide consistent performance for file systems. I/O Quota must be configured from the Solutions Enabler host. The recommended configuration (best practice) for NS gateway using I/O Quota with masking view (LUN masking feature) is as follows: Ensure that a parent Storage Group is in a masking view with the Data Mover initiator group and Symmetrix FA port group.
® ™
• • • •
•
•
2
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
Create multiple Child Storage Groups with different performance expectations. For example, extra performance, performance, and capacity. Set an appropriate I/O Quota in each Storage Group. Execute diskmark operation. If the storage groups are all from fibre channel, create user defined storage pools for each storage group. Create file systems from the user defined storage pools.
Version 6.0.65.2
For NS-G2, NS-G8, VG2, VG8 gateways, if you plan to upgrade to VNX File OE 7.0 or later in future, do not install this version.
Versions 6.0.61.0, 6.0.60.2, 6.0.51.6, 6.0.43.1, 6.0.43.0, 6.0.42.3, and 6.0.41.4
There were no new features or changes to functionality in these versions. The Fixed problems section contains changes made in this version.
Version 6.0.55.3
NS-G2/NS-G8 6.x support for VNX Block arrays This release supports installation of NS-G2/NS-G8 Gateways and their connection with VNX Block arrays using the Fibre Channel protocol. New features in the upcoming VNX Operating Environment for Block 5.32 are not supported.
Version 6.0.41.3
Checkpoint scheduling relative name convention As of this release, SnapSure supports checkpoint CVFS names that are dynamically assigned based on the relative order of checkpoints (for example, nightly.001, nightly.002, nightly.003, and so on.) To use this option, you must specify the relative CVFS name prefix, the delimiter, and the starting index. The delimiter and starting index values are recorded in the nas_param database and must be manually edited. The complete CVFS name while using the relative prefix option is a concatenation of the prefix, delimiter and starting index. The prefix of a relative checkpoint must be a PFS-wide unique string and can contain up to 20 ASCII characters. It must not include intervening spaces, colons (:), or slashes (/). The delimiter must be either a period (.)
EMC Celerra Network Server Version 6.0.70.4 Release Notes 3
New features and changes
or an underscore (_). The starting index must be either 0 or 1. This feature is managed through the GUI or the CLI, or both. CPU Utilization Threshold Alerting As of this release, you can enable CPU utilization threshold alerting and receive automated SNMP trap or email notifications when a Data Mover CPU utilization crosses a certain threshold for a set period of minutes and persistence. You can also be notified when the condition is no longer applicable and the CPU utilization subsides to a normal level. The Configuring Celerra Events and Notifications technical module provides more detailed information about this feature. File system replication with common base checkpoints This feature is also known as "Incremental Attach Replication". When you replace source or destination file systems, the replications and internal checkpoints are deleted. You will have to perform a full data copy to be able to create future replications with the original configurations. To avoid performing a time-consuming full data copy over the WAN, you can create a local replica of the file system that you want to replace. You can then create one or more user-specified checkpoints on the file system to be replaced, the local replica, and each of the other file systems involved in the replication. After creating user-specified checkpoints, you must refresh each replication by specifying the user checkpoint on the source and destination file system for that replication. Consequently, the user checkpoint of each destination file system (including local replica) has the same view of the file system as the user checkpoint on the source file system that will be replaced. User checkpoints on the source and destination file systems, which have the same content, are called common base checkpoints. You can now use the local replica for replications without performing a full data copy over the WAN. Implementing this technique is useful in disaster recovery scenarios as well. To avoid a full data copy in the event of a disaster, create one or more user-specified checkpoints on the source and destination file systems and refresh their replication by using these user checkpoints. This process transfers data from the source user checkpoint to the destination file system through the existing replication session. Thus, the destination user checkpoint has the same view of the file system as the
4 EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
source user checkpoint. This process is repeated for each destination in the one-to-many configuration. For a cascading replication, the replication must be refreshed for this first hop and then for the second hop. After this process, even if you delete the replication sessions, you can configure new replication sessions between any two destinations because replication automatically identifies the common base checkpoints and avoid full data copy. You can create user checkpoints for file system and VDM replications. For VDM replications, you can create the user checkpoints on the root file system of the VDM. You cannot delete user checkpoints when they are being used by replication either for transfer or as an active common base. Greater than 4 TB Quotas As of this release, the storage limit per user, group, or tree quota has been increased from 4 TB to 256 TB. You can control the usage of the disk resource and limit the expensive disk space usage by various user or group or trees on a file system. Support for 256 TB quotas does not imply any change in the maximum size of the underlying file system. Caution: If you replicate a quota-enabled file system from a system running Celerra Network Server (CNS) version 6.0.41.0 or later to a destination system running a CNS version earlier than 6.0.41.0, the destination file system can become unusable. Upgrade quota database limits for file systems You can upgrade your quota databases to the new 256 TB limit from the existing 4 TB limit. The nas_quotas command has a new option called quotadb that, when run, performs the upgrade. The file system will be unavailable during the upgrade.
Note: Before the upgrade process runs, the Control Station displays the estimated upgrade time on the file system whose quota database will be upgraded, and also displays a warning message to notify you that the file system will be unavailable during the upgrade process. If you are in interactive mode, a dialog displays letting you choose whether you want to continue. Since the file system will be unavailable during the upgrade, the upgrade can be deferred to a later time. If you are in non-interactive mode, after displaying the estimated upgrade time message and warning message, the upgrade process starts immediately.
The Command Reference Manual contains more detailed information about the -quotadb command.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
5
New features and changes
Online quota update tool Once a tree, user, or group quota is enabled on a directory, in some instances, differences may occur between a reported quota usage and the actual usage. The online quota update tool lets you run a quota check on a tree or file system. The tool updates the tree, group, and user usage statistics, while keeping the directory tree and file system online, and keeping data access available. You can also run a quota check in the offline mode, if desired. If you have file system administrative privileges, you can use the Celerra CLI to run the online quota update tool. If an error is encountered, an error message is generated. The status of the online quota check shows the percentage completed.
Note: 512 MB of free space is needed on the file system to run a tree quota check.
Performance problem collection tool This tool enables users to send all relevant information related to the file system to EMC for diagnosis. The performance problem collection tool is a single CLI tool that gathers all the relevant logs (systat, IO trace, etc.) for a specified file system so that it can be easily transmitted to EMC as part of a support escalation. Prior to this release, users had to manually provide this information. Symmetrix Enginuity version 5875 This release supports the new Symmetrix Enginuity version 5875, which runs on the VMAX platform, and includes a new version of Solutions Enabler v7.2. VNX supports tolerance for Enginuity 5875 release only. It does not leverage any new features included in 5875. User-defined pool usage enhanced The -is_greedy attribute of the nas_pool command has been enhanced to include identifying which member volumes of a user-defined storage pool are used to provide space when creating or extending a file system. If the -is_greedy attribute is set to n (default), the system uses space from the user-defined storage pool's existing member volumes in the order that the volumes were added to the pool to create a new file system or extend an existing file system. If the -is_greedy attribute is set to y, the system uses space from the least-used member volume in the user-defined storage pool to create a new file system. When there is more than one least-used member
6
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
volume available, AVM selects the member volume that contains the most disk volumes. For example, if one member volume contains four disk volumes and another member volume contains eight disk volumes, AVM selects the one with eight disk volumes. If there are two or more member volumes that have the same number of disk volumes, AVM selects the one with the lowest ID. When extending a file system, the -is_greedy attribute is ignored unless there is not enough free space on the existing volumes that the file system is using. Windows event log auto archive With Windows operating system, applications can use the event logging mechanism to log their own events. Celerra currently supports three such event logs that are security, system, and applications. The physical format of these logs use a Microsoft format called 'evt' that has a limitation of 4 GB in size because there are some fields stored on 32 bits integers. Windows 2008 has introduced a new format 'evtx' that does not have this limitation. The event log auto archive feature allows you to automatically archive an event log on a particular trigger and to continue the logging on a new event log without losing any events. The archive is triggered on a time and/or on an event log size basis defined by parameters in the Windows registry. You can specify a retention policy to keep the event log archives before they can be recycled based on the duration and/or the total archive disk size. All parameters are stored in the Windows registry of each VDM. Therefore, each VDM will have its own configuration. The parameters can be viewed and edited with tools like regedit. The Configuring and Managing CIFS on Celerra technical module provides more information about this feature.
Version 6.0.40.8
Celerra Upgrade Tool -localization upgrade support The Celerra Upgrade Tool (CUT) allows you to download language packages and manage your Celerra in your primary language. The Download Software wizard helps you download the language packages from the EMC Online Support website based on your Celerra configuration. The Celerra Upgrade wizard then installs the language
EMC Celerra Network Server Version 6.0.70.4 Release Notes 7
New features and changes
packages on your Celerra. Celerra file services will not be disrupted during language package installation. This feature is supported on EMC Celerra 6.0 systems only. Unisphere upgrade procedure EMC Unisphere Service Manager (USM) is a collection of tools that help you update, install, and maintain your CLARiiON system. It also includes the Celerra Upgrade Tool (CUT) which consists of tools that help you perform software upgrades to Celerra. This procedure explains how to use USM to upgrade Celerra Unified Platform systems. 1. 2. Go to EMC Online Support and select Support by Product > Celerra NS> Downloads. Select the USM installation (executable) file (for example, Unisphere Service Manager (USM) (Windows) 1.2.2.1.0020) and download the file to a known location on your PC. In the folder where you save the USM installation file (for example, UnisphereServiceManager-Win-32-x86-en_US-1.0.0.1.0490-1.exe), double-click the executable (.exe) file, and if necessary, click Run to start the installation wizard. Follow the instructions that appear. After you accept the license agreement, the wizard verifies that the server is running the supported Java Runtime Environment (JRE) version. If the server is running an earlier version of the JRE, select Yes when asked if you want to continue with the installation. The installation wizard prompts you to download and install the later version of the JRE from the Sun website. If the server is running a later version of the JRE, the installation wizard asks if you want to continue with the installation. Select Yes to continue or No to quit the installation. If you select Yes, the USM may not work properly with a later JRE version. If you select No, and manually remove the later version of the JRE, other applications that require this version may not work properly. 5. When the installation is complete, click Done.
3.
4.
8
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
Register your Celerra to use the Celerra Upgrade Tool (CUT) 1. Register your system on EMC Online Support to enable CUT to download software. Go to the Service Center tab and select Register a Product. 2. 3. Enter the Product ID and click Continue. You are now entitled to use CUT.
Run the Unisphere Service Manager The USM will allow you to upgrade the NAS and FLARE software on your system. As a general rule, you should update the NAS version to the latest qualified release (6.0) prior to updating the FLARE version to its latest release (R30). Upgrade NAS 1. Double click the Unisphere Service Manager icon on your desktop.
2.
Click Login and log in to a Celerra. CUT supports version 5.6 system login as root only. The Name field is prefilled with the username root. The root password is required.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
9
New features and changes
3.
In the menu bar, click Software > System Software to run the Celerra Upgrade Tool. Three wizards will guide you, step-by-step, through the process of upgrading the Celerra software, which includes updating the Control Station and the blades. Select the wizards in the order presented on the screen: Download Software Pre-upgrade Health Check Perform Software Update
4.
The Celerra Upgrade Tool online help provides information about using the wizards. 5. Select the CUT files for download. For example: 6.0.36.4_apps.iso 6.0.36.4_emcnas_CD1.iso pkg_6.0.36.4_emcnas.tar
Note: After running the Health Check, be sure to resolve any errors before performing the Software Update.
10
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
Upgrade Flare 1. Double click the Unisphere Service Manager icon on your desktop. 2. Click Login and log in to your storage system. Enter the storage array admin account name and password and verify that Scope is set to Global.
3. 4.
In the menu bar, click Software > System Software and select the Prepare for Installation wizard. In the Prepare for Installation wizard, select the Download software and verify storage environment option, and follow the steps to the Select Software screen. Select the Any supported software option. This sequence of steps allows you to select only the Flare package for downloading, rather than all other associated but optional software items. When the Prepare for Installation wizard is finished, start the Install Software wizard. Follow the instructions to install the desired software packages, for example, Flare 30 release: 04.30.000.5.004. The actual download file is named CX4-Bundle-04.30.000.5.004.pbu.
EMC Celerra Network Server Version 6.0.70.4 Release Notes 11
5. 6.
New features and changes
7.
When the installation is complete, select Notify your service provider with updated storage system configuration information, and click Finish.
Version 6.0.36.4
The following new features are supported as of Celerra Network Server version 6.0.36.4. 256TB capacity/blade With the introduction of the Celerra 6.0 software release, we have increased the capacity/blade of the NSG8 and NS960 platforms to 256TB per blade. This enables users to increase the number of disk drives per blade without needing to acquire an additional blade, thereby reducing the user's TCO.
64-bit OS
The Celerra 6.0 release introduces the 64-bit version of the DART operating system. 64-bit technology enables blades with large memory configurations to effectively utilize system resources in order to provide capacity and scalability enhancements across the board.
Control Station auditing system
The system auditing feature has been enhanced with new trigger mechanisms to detect potential system access violations and to start up automatically on single and dual Control Station configurations. For more information, see the Auditing in the Celerra Control Station Technical Note.
Celerra Upgrade Tool
The Celerra Upgrade Tool (CUT) lets you upgrade your Celerra systems without assistance from EMC or an EMC service provider. CUT communicates over the Internet to EMC Online Support and over a local network to a Celerra system. CUT is part of the Unisphere Service Manager (USM), and you will now be able to use the same tool to upgrade both Celerra and CLARiiON systems. CUT uses three wizards to safely guide you, step-by-step, through the process of upgrading the Celerra software.
12
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
The Celerra Download Wizard selects the latest Celerra software for your system. Software selection is based on your hardware configuration, the software currently installed on the Celerra, and the storage system software running. The wizard can examine your system to obtain this information, or you can provide the information. The Download Celerra Software Wizard gives you one or two choices depending on your configuration, a maintenance release with the latest bug fixes or a new version with new features. The wizard then helps you download the appropriate version of the Celerra software to a central repository on a PC or laptop. The Pre-Upgrade Health Check Wizard examines the health of the system to determine if the Celerra will support a successful software upgrade. The Celerra Upgrade Wizard runs the pre-upgrade health check on the Celerra, upgrades the software, and then runs a post-upgrade health check.
Unified Storage system privileges required to run Celerra Upgrade Tool wizards With the release of NAS 6.0, Unified Storage systems can now participate in a CLARiiON Storage domain as a member system. After becoming a member of the CLARiiON domain, whether the system joined automatically or manually, global user accounts are automatically mapped to the Celerra with Operator (nasadmin) privileges by default. By design, the migrated accounts do not have the necessary privileges to run any of the Unisphere Service Manager's (USM) Celerra Upgrade Tool (CUT) wizards. USM displays only the Celerra Download and the Celerra Health Check wizards. The Celerra Upgrade wizard is displayed only if the fullnas (nasadmin) or root privileges are assigned to the migrated Users. In addition, to use all three CUT wizards, the Control Station shell allowed privilege must be enabled for each user. This right allows users to fully use the Celerra Download and the Celerra Health Check wizards. In order to run the Celerra Upgrade wizard, users must also have fullnas (nasadmin) privileges. To fully use the CUT wizards: 1. 2. Run Unisphere and enter a hostname or IP address. In the login dialog, log in to the Celerra as root.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
13
New features and changes
3. 4. 5.
From the Scope drop-down list, select Local. Select Settings > User Management > Users. Highlight the user account to which you want to assign privileges and click Properties. In properties, from the Group (Role) Membership field, select fullnas (nasadmin). From the Client Access field select Control Station shell allowed.
Once you have set these privileges for users, they can run USM and log in to a Celerra system and upgrade it with CUT.
Celerra VG2 and VG8
Celerra now supports the new low-end and high-end gateway platforms VG2 and VG8. These two platforms are based on the newest Intel Technology which provides greatly increased performance. The low-end platform, VG2, introduces UltraFlex™ technology for the first time in this level of NAS product and improves the system availability with the option of two Control Stations for high availability failover.
CEPA support for NFS audit applications
This feature provides event notifications and context to audit applications that monitor the NFS file system activity on Celerra, which run in the Windows environment. The audit applications require event notifications, from the Data Mover, to organize the access of information that is stored on the Celerra file system. To provide this functionality, the Celerra Event Publishing Agent (CEPA) API was modified to allow the applications to obtain the required Data Mover event information without blocking the CIFS or NFS client applications. To use this feature, the audit applications need to register for post-event notifications through the CEPA API. The CEPA API consists of an IDL file and an XML DTD file. These files contain information that is required by the audit application to interact with the CEPA. The audit application may coexist with the Celerra Event Enabler (CEE) framework on the same client (where Local RPC is used for communication) or on the remote client (where MS-RPC is used for communication). The CEE facilitates the use of selected third-party applications with the Celerra file systems. It provides event values containing the required context as defined by the audit applications for each class. As more vendor applications are added to each class, the events and associated context are modified to accommodate the applications.
14
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
The Celerra Event Publishing Agent technical module provides more information.
CLARiiON SATA EFD disk volumes
CLARiiON SATA EFD disk volumes are now supported, in addition to CLARiiON Fibre Channel disk volumes. Celerra support includes the ability for users to provision storage and create / extend file systems on these drives. The SATA EFD drives are treated identically to FC EFD drives which mean both drive types can be part of the storage pool, one drive type can be replaced with another, and file systems can span FC EFD and SATA EFD drives. The Managing Celerra Volumes and File Systems with AVM book has more details relating to EFD drives.
Control Station Linux upgrade
As of this release, the Celerra Control Station is upgraded to Red Hat Enterprise Linux 5 (RHEL 5). RHEL 5 leverages new security features, updates, and bug fixes in the kernel and applications provided by RedHat.
Dual port 10GbE SLIC for Data Movers
A new dual-port 10GbE UltraFlex I/O Adapter is being made available for the NS960, NSG2 and NSG8 platforms. This I/O Adapter doubles the effective throughput from the prior generation, which consisted of a single port card. In addition, the UltraFlex form factor allows for easy installation and upgrade to any chassis that supports UltraFlex adapters.
Express Install
Express Install introduces a method to reset the Celerra system back to its default factory configuration.
FAST Cache for the Provisioning wizard
As of this release, Celerra supports EMC FAST Cache as an extension to the dynamic random access memory (DRAM). This extended cache can potentially improve performance of the overall system by increasing write cache size at a comparably low cost. FAST Cache can be configured anytime by using Unisphere. The provisioning wizard in CSA and UMPW provides you with an option to
EMC Celerra Network Server Version 6.0.70.4 Release Notes 15
New features and changes
skip the provisioning of the EFD drives in order to configure them as the FAST Cache drives (maximum 6). You can continue to provision the remaining drives for file / block use.
FLARE R30
Celerra Network Server now supports the latest CLARiiON FLARE release, R30 when installed on CLARiiON CX4 storage arrays. It supports all R30 features, but does not actively support Fully Provisioned Pool LUNs, Auto-Tiering and Compression. Refer to the E-Lab Interoperability Navigator at http://Support.EMC.com for more specific FLARE version numbers.
TM
IPv6
IPv6 is fully supported and configurable by means of the CLI, and partially supported (read-only and some limited configuration) by means of Unisphere. CLI commands All CLI commands now support IPv4 and IPv6 addresses. There are some new commands provided specifically to handle IPv6. Some existing commands have updated syntax to support IPv6 addresses. Supported areas include: NFSv2/v3, FTP, TFTP, NIS, DNS, partial LDAP, Kerberos, CIFS, iSCSI, Network, NTP, NFSv4, Secure NFS, UserMapper, ReplicationV2, XLT, CDMS, FileMover, Control Station, SNMPv1 traps, SNMPv3, MirrorView, SRDF, CAVA/VirusChecker, ConnectEMC and Celerra XML API V2. The Configuring andManaging Celerra Networking technical module provides detailed information about IPv6 networking. IPv6 read-only support Unisphere provides IPv6 read-only support with a few exceptions. If IPv6 is configured by means of the CLI, that configuration is visible on the Unisphere. The information about auto-generated link-local interfaces and routes is not visible by means of Unisphere. The IPv6 addresses in drop down lists and checkboxes display IPv6 configuration information and are selectable. IPv6 addresses in hyperlinks navigate to the appropriate destination. Some screens allow the selection of IPv6 address for configuration in CIFS and Replication. The netmask field displays either an IPv4 subnet mask for IPv4 addresses or a non-editable IPv6 prefix for IPv6 addresses.
16
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
The broadcast address is blank for IPv6. The IPv4 network configurations are fully supported. When there are combined configurations of IPv4 and IPv6, changes via Unisphere are prohibited. Unisphere wizards, including the CIFS, Control Station, iSCSI, Network, Replication, and Set up Celerra wizards provide read-only support for IPv6. Lists containing IPv6 address are selectable, but not editable. CBM applications Client-side CBM applications may not work with an IPv6 configured iSCSI LUN. Celerra Monitor Celerra Monitor provides read-only support for IPv6. When IPv4 is configured, the screen displays the IP address, netmask, and broadcast fields. When IPv6 is configured, the screen displays IP address and the subnet prefix length fields. CSA The Celerra Startup Assistant (CSA) does not provide IPv6 support at this time. A new CLI command (nas_cs) provides IPv6 support. IPv4-only or IPv4/IPv6 environment? Should you configure IPv4-only environment or a dual stack IPv4/IPv6 environment? When deciding this, you need to answer the following question: In my environment, do any nodes or peers served by the Celerra use the IPv6 protocol? If you answer yes, the Celerra and associated infrastructure, such as routing, Domain Name System (DNS), and Network Information Service (NIS), should be configured for a dual stack environment. Link-local information The Celerra XML API V2 supports IPv6, however the auto-generated link-local information is not visible for the XML API V2. The application protocols are not supported via link-local interfaces. The link-local related information is supported only for IPv6 core protocols such as Duplicate-Address-Detection, Router Discovery, Neighbor Discovery, and ICMP6 and is exposed only via the server_ping6 command at this time. Only the CLI will have the view of all the IPv6 interfaces. MMC Microsoft Management Console (MMC) Snap-in provides full IPv6
EMC Celerra Network Server Version 6.0.70.4 Release Notes 17
New features and changes
support. That is, MMC accepts both IPv4 and IPv6 addresses that you input. Machines hosting MMC must be running Vista or Windows Server 2008 or later to support IPv6.
Kerberos authentication support for LDAP-based directory services
In this release Celerra adds support for Kerberos authentication to the LDAP-based directory services provided by Active Directory with Microsoft Windows Services for UNIX (SFU) or Identity Management for UNIX (IdMU). When configuring LDAP, the Data Mover’s directory client can now use Kerberos, simple (password), or SSL authentication. Anonymous authentication is not allowed. Kerberos authentication means the Data Mover, configured as a CIFS server, uses a KDC to confirm the Data Mover’s identity when accessing the Active Directory. The Configuring Celerra Naming Services technical module describes how to configure a Data Mover which is configured to use Active Directory with SFU or IdMU to use Kerberos authentication.
MAC OS 10.6
In this release, EMC qualified MAC 0S 10.5 and 10.6 in a CIFS protocol environment. File access protocol using native CIFS protocol stack in MAC-OS client has been verified in term of compatibility with Celerra CIFS implementation. The 'AppleSingle' feature which is using data stream to store resource information instead of a second file associated to the data file has also been validated (with MAC-OS 10.6, this is the default behavior on the client mount option).
MPFS Operating Systems
MPFS now supports Windows 7 clients (x86 and x86_64), Windows 2008 R2 (x86_64 only), CentOS 5.4 (x86 and x86_64), SLES 10 SP3 (x86, x86_64) and SLES 11 (x86 and x86_64 only). CentOS is supported over MPFSiSCSI only. All other operating systems are supported over Fibre Channel and iSCSI. For Windows 7 and Windows 008 R2, MPFS supports SMB1, SMB2 and SMB2.1 versions of the CIFS protocol.
18
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
NDMP
Increased NDMP backup streams The increased NDMP backup concurrent streams feature enables the Celerra Network Server to increase the maximum number of simultaneous concurrent backup sessions from four to eight. It verifies that the system memory is at least 8 GB, and then sets the value to eight. If the system memory is less than 8 GB, the value defaults to four simultaneous sessions. The Celerra Network Server Parameters Guide provides more detailed information about this feature. To support eight concurrent backup or restore streams: The backup client (DMA) must be compatible and configured to run eight concurrent streams. The Data Mover must have at least 8 GB memory.
Increasing the number of concurrent sessions to eight may result in an increase in the utilization of CPU and system memory. The Configuring NDMP Backups on Celerra technical module provides more information. NDMPcopy The NDMPcopy feature is an open source program that allows users to copy file system data within the Data Mover or between different Data Movers that are on the same or different Celerra. The NDMPcopy enables users to directly transfer data from the source to the destination file system, without traversing the NDMPcopy host machine or network (if the source and destination file systems are on the same Data Mover). It uses NDMP (Network Data Management Protocol) to transfer the file system data. With the NDMPcopy program, the NDMP host and network (if the source and destination file systems are on the same Data Mover) are not overloaded, because the data transfers take place directly from the source to the destination. The NDMPcopy program contacts the NDMP server processes on the source and destination file systems, and authenticates the connection. It then initiates data transfers between the source and destination file systems. The NDMP host receives status messages only. The NDMPcopy also provides multiprotocol (NFS/CIFS) support. EMC recommends using the NDMPcopy program for data transfers only. The NDMPcopy may affect the scheduling of backup activities in the same source, as it modifies the BackupDates file on the Data Mover. It uses the text mode for authentication and so the password is sent across the
EMC Celerra Network Server Version 6.0.70.4 Release Notes 19
New features and changes
network in a non-encrypted format. The Configuring NDMP Backups on Celerra technical module provides more information. NDMP port range The NDMP port range feature allows users to configure the port range to use a limited range of TCP ports that correspond to the ports for which the firewall allows inbound connections. EMC recommends setting the NDMP port range using dynamic or private ports (49152-65535) of IANA. The required minimum value of the port width is 32; however, EMC recommends setting the width to at least 100. Setting the port range does not imply that these ports are reserved for NDMP connection. When the port range is set, NDMP selects a random port number from the specified port range as the data connection listening port. If the port range is not set, the default port range is used. NDMP is incompatible with NAT (Network Address Translation) firewalls. It uses the existing server_param management interface and is independent of the DMA. Configuring the NDMP port range allows the Data Mover to use a limited range of TCP ports that correspond to the ports for which the firewall allows inbound connections. The Configuring NDMP Backups on Celerra technical module and the Celerra Network Server Parameters Guide provide more detailed information about NDMP port range.
pNFS
With this release, the Celerra Network Server provides a conforming implementation of NFSv4.1 in order to support pNFS. The pNFS is a part of NFS version 4.1 that allows clients to access storage devices directly and in parallel. In order to support pNFS, the Celerra Network Server fully supports all required NFSv4.1 functionality. The pNFS architecture enhances the scalability and performance of the existing NFSv4 servers. This is achieved by the separation of data and metadata, and moving the metadata server out of the data path.
Quotas
Choose and set a quota policy As of this release, administrators can change the quota limit check policy without turning off (disabling) all of the quotas on the file system. The new policy change will be applied after the Data Mover is rebooted.
20
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
The technical module Using Quotas on Celerra provides additional information about this change.
RSA Common Security Toolkit module
In this release the RSA Common Security Toolkit (CST) module replaces the CAMS Common Sign-On (CSO) module which was previously used to handle remote authentication of Celerra administrative users against an external directory server. CST changes the operation of the directory server settings test function. The Celerra Security Configuration Guide and Unisphere Settings > User Management online help provide more information about this feature.
Samba 4
Celerra supports Samba 4, except for the Samba 4 domain controller. Samba is an open source / free software suite that provides file and print services to all manner of SMB/CIFS clients, including the numerous versions of Microsoft Windows operating systems.
SAN Copy
SAN Copy™ replication is an array based replication technology. The performance and scalability of array based solutions is better for largescale migration than host based applications. SAN Copy is integrated into the Unisphere console and simplifies the configuration and management of LUN migrations through a user friendly GUI interface. The SAN Copy CLI Reference Guide and the on-line help contain additional information about SAN Copy.
SMI-S v1.4
The new SMI-S feature enables generic SMI-S compliant clients to manage a Celerra using CIM (Common Information Model) operations defined for NAS. This feature delivers two components of WBEM (Web Based Enterprise Management) infrastructure; namely a CIM Server and a Provider. Both components are embedded on the Celerra Control Station. SMI-S contains an API that storage management applications can use to provide standardized view of a Celerra. Providing a compliant API in SMI-S terms means supporting profiles and sub-profiles of the NAS domain. For Celerra, this means supporting the NAS Head Profile at a minimum, along with its mandatory subprofiles.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
21
New features and changes
SNMPv3
The Simple Network Management Protocol (SNMP) collects and processes valuable network information. It gathers data by polling the devices on the network from a management station at fixed or random intervals. In this release, a new SNMP agent, the SNMP daemon (SNMPD), which runs on the Data Mover, supports SNMPv1, SNMPv2c, and SNMPv3. SNMPv3 supports IPv4, IPv6, and enhanced security over SNMPv1 and SNMPv2c. SNMPv3 provides important security features such as: Authentication to verify that the message is from a valid source Encryption of packets to prevent snooping by an unauthorized source Message integrity to ensure that a packet is not tampered with in transit
Tolerance for 8 GB FC in Symmetrix
When a Celerra gateway with 4G Fibre Channel (FC) is attached through a switch to a Symmetrix platform incorporating an 8 GB FC card, there is no loss of or change in functionality. The communication rate automatically drops to 4 GB (Symmetrix platform data rate negotiation) and operation continues as expected.
Tools
Windows event log auto archive With Windows operating system, applications can use the event logging mechanism to log their own events. VNX currently supports three such event logs that are security, system, and applications. The physical format of these logs use a Microsoft format called 'evt' that has a limitation of 4 GB in size because there are some fields stored on 32 bits integers. Windows 2008 has introduced a new format 'evtx' that does not have this limitation. The event log auto archive feature allows you to automatically archive an event log on a particular trigger and to continue the logging on a new event log without losing any events. The archive is triggered on a time and/or on an event log size basis defined by parameters in the Windows registry. You can specify a retention policy to keep the event log archives before they can be recycled based on the duration and/or the total archive disk size.
22 EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
All parameters are stored in the Windows registry of each VDM. Therefore, each VDM will have its own configuration. The parameters can be viewed and edited with tools like regedit. The Configuring and Managing CIFS on VNX technical module provides more information about this feature.
Unisphere
Version 6.0.x introduces Unisphere. Unisphere is a unified SAN/NAS application providing uniform and intuitive management and monitoring services. It replaces Celerra Manager, Navisphere Manager, and Navisphere Express. Highlights of the new Unisphere application include: Single sign-on for a user identity common to multiple Celerra and CLARiiON systems. This functionality is facilitated by creating a global domain user who can also manage Celerra systems. This common user only needs to log in to one system in the domain in order to access and manage all systems. Management of up to 50 Celerra and CLARiiON systems within a single Unisphere instance. System summary views in a new dashboard that provides alerts and status summaries for the managed system list populated by the user. Participation in storage domains by both CLARiiON and Celerra systems. Cross-linking capability for LUN, disk group, storage pool, and storage system properties from Celerra to the relevant CLARiiON.
Note: All existing Celerra features, with the exception of message-of-the-day and system monitoring after session timeout, are supported within Unisphere.
The following table lists where each Celerra Manager view can be found in Unisphere. In some cases, Celerra Manager views map only to tasks found in Unisphere’s left-pane task list.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
23
New features and changes
Table 1 Celerra Manager to Unisphere Feature Mapping Celerra Manager Celerra Home > Status Celerra Home > Licenses Celerra Home > Statistics Celerra Home > System Info Celerra Home > Control Station Properties Data Movers Data Movers > Virtual Data Movers Data Movers > Properties > Notifications Storage > Systems Storage > Systems > Properties Storage > Systems > Properties > Disk Groups Storage > Systems > Properties > Spare Disk Devices Storage > Volumes Storage > Pools Storage > Pools > Usage Statistics Unisphere All Systems > Dashboard > Alerts System (System tasks) > Manage Licenses Monitoring (Statistics tasks) > All Performance System > Details System > Details System (System tasks) > Manage Control Stations System > Data Movers Sharing > CIFS Monitoring > Notifications System > Details System > Details Storage > RAID Groups Storage > Hot Spares Storage > Volumes Storage > Storage Pools Storage (Storage Pools tasks) > View Usage Statistics Monitoring (Statistics tasks) > Storage Pool Usage Storage > File Systems Storage > File Systems > Properties Storage > File Systems > Properties Storage > File Systems > Mounts Storage > File Systems > Properties Storage > File Systems (File Systems tasks) > Manage Quota Settings Removed Storage > File Systems Storage (File Systems tasks) > View I/O Statistics Monitoring (Statistics tasks) > File System I/O
File Systems File Systems > Properties File Systems > Properties > Checkpoint Schedules File Systems > Properties > Mounts File Systems > Properties > Checkpoint Storage File Systems > Properties > Quota Settings File Systems > Properties > Notifications File Systems > Mounts File Systems > I/O Statistics
24
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
File Systems > Usage Statistics
Storage (File Systems tasks) > View Usage Statistics Monitoring (Statistics tasks) > File System Usage1 File Systems > Properties Storage > File Systems Storage > File Systems (File Systems tasks) > Manage Quota Settings Replicas > Checkpoints Storage > Data Migration Sharing > CIFS Sharing (CIFS tasks) > View Statistics Monitoring (Statistics tasks) > CIFS Sharing > NFS Sharing (NFS tasks) > View Statistics Monitoring (Statistics tasks) > NFS Sharing > iSCSI Sharing (iSCSI tasks) > Manage Settings Replicas > Replications System > Network System (Network tasks) > Manage NIS Settings System (Network tasks) > Run Ping Test System (Network tasks) > View Statistics Monitoring (Statistics tasks) > Network Device Storage > Virtual Tape Monitoring > Event Logs Monitoring > Notifications System > Hardware Monitoring > Background Tasks All task lists Settings > User Management Settings (task list) > Manage Idle Timeout System > Network Settings > Public Key Certificates Monitoring (task list) > Launch Celerra Monitor n/a System (Network tasks)
EMC Celerra Network Server Version 6.0.70.4 Release Notes
File Systems > Deduplication Settings File System Quotas File System Quotas > Settings SnapSure Checkpoints Data Migration CIFS CIFS > Statistics NFS Exports NFS Exports > Statistics iSCSI iSCSI > Configuration Replicator Network Network > NIS Settings Network > Ping Network > Statistics VTLUs Events Notifications Inventory Task Status CLI Commands Security > Administrators Security > Celerra Manager Security > Network Services Security > Public Key Certificates Tools > Celerra Monitor Tools > Launch Navisphere Wizards – Network
25
New features and changes
Wizards – CIFS Wizards - File System Wizards – Replication Wizards – iSCSI Wizards – Celerra Support
Sharing (CIFS tasks) Storage (File Systems tasks) Replicas (Replications tasks) Sharing (iSCSI tasks) System (System tasks) System (Service tasks)
Upgrade media
Dual Compact Disk As of this release, the upgrade media is delivered on dual CDs. An outof-family upgrade, for example, V5.6 to V6.0 requires both CDs. The software upgrade prompts the user to swap CDs when necessary. DVD The upgrade media is available on one DVD for users of Control Stations that have optical drives capable of reading DVD discs.
Users and roles
Version 6.0.x introduces several changes to the Celerra user and rolebased access feature. CLI role-based user access A user account is always associated with a primary group and each group is assigned a role. A role defines the privileges (that is, the operations) the user can perform on a particular Celerra object. In version 6.0.x, Celerra extends support for assigning role-based privileges to a user accessing Celerra through the CLI. The ability to select a predefined role or define a custom role that gives a user the necessary privileges to execute a command is not available for all CLI commands. Certain commands can only be performed by the default user accounts root and nasadmin or, in some cases, by a user account associated with the root and nasadmin roles. You create and manage role-based user access in Unisphere Settings > User Management > Roles. You must be root or a user who has root or security operator privileges to create a user account and to associate it with a group and role. The Celerra Security Configuration Guide, Celerra Network Server Command Reference, and Unisphere Settings> User Management online
26
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
help provide more information about the specific command actions for which privileges can be controlled.
Unisphere storage domain users and roles
To simplify access and privileges for users managing both CLARiiON and Celerra systems in Unisphere, user accounts created to manage CLARiiON storage are mapped to Celerra user accounts, groups, and roles so that CLARiiON storage users can be given privileges to access and control Celerra objects. Groups and roles with the prefix sd_listed in Unisphere Settings> User Management views are mapped from CLARiiON storage domain groups and roles.
Known problems and limitations
CDMS
Connecting to a Solaris secondary server Users are unable to create a CDMS or a FileMover connection to a Solaris secondary server using TCP with IPv6. IPv6 with NFSv3 will not be able to create this connection. If UDP is chosen instead of TCP, this type of connection will work with a Data Mover NFS server. This problem has only been seen with Solaris NFS servers. Workaround Specify UDP as the transport protocol instead of using the default TCP. This option can be set with both the server_cdms and fs_dhsm commands. Celerra Event Enabler The following known problems and limitations pertain to Celerra Event Enabler: CAVA/CEPA servers Each Data Mover should have a CAVA pool consisting of a minimum of two CAVA servers specified in the Data Mover’s “viruschecker.conf” file, or a CEPA pool consisting of a minimum of two CEPA servers specified in the Data Mover’s “cepp.conf” file.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
27
Known problems and limitations
CIFS and NFS clients The CAVA feature of the CEE product is for CIFS and NFS clients. If FTP protocols are used to modify files, the files will not be scanned for viruses. No UNIX antivirus application support is provided. File-Level Retention It is strongly recommended that the antivirus (AV) administrator updates the virus definition files on all resident AV engines in the CAVA pools, and periodically run a full file system scan of the file system to detect infected FLR files. The Using Celerra AntiVirus Agent technical module provides additional information about virus definition updates. Microsoft Access Do not scan Microsoft Access database files in real time. Files inside a database are not scanned. Windows 64-bit operating systems In order to run CEE on Windows 64-bit operating systems, the Celerrato-CEE communications must be over Microsoft RPC (MS-RPC). The version of CEE that runs on Windows 64-bit operating systems is supported with Celerra version 5.6.45 or later and CEE version 4.5.0.4 or later. Celerra iSCSI AIX A known problem with SMIT (System Management Interface Tool) in AIX versions 5.2, 5.3, and 6.1 prevents it from creating and handling compliant IQN names for iSCSI nodes. As a workaround, manually edit the file /etc/iscsi/targets to enter or modify IQN names. Check the IBM support website for additional information. Configurations Celerra iSCSI configurations in the backend replication environment do not support TimeFinder®/FS, NearCopy, and FarCopy. Host connectivity issues Check the following guides for issues related to the Microsoft initiator, QLogic HBAs, and compatibility with the Windows operating system:
28
Host Connectivity Guide for Windows Host Connectivity Guide for Linux
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
These guides are available from EMC Online Support at http://Support.EMC.com. Microsoft iSCSI Software Initiator EMC Celerra iSCSI supports the most current version of the Microsoft iSCSI Software Initiator, which is available as a free download from Microsoft’s website (search for iSCSI initiator). This section contains known issues with the Microsoft iSCSI Software Initiator. Initiator uses only the first network portal of a target An iSCSI target on the Celerra Network Server may have multiple network portals configured for it; however, the Microsoft iSCSI initiator GUI uses only the first network portal of a target.
Note: Applies only to Microsoft iSCSI Software Initiator versions 2.0 and earlier.
Workaround Use the Microsoft initiator CLI to access other portals.
Control Station
Authentication failure Following a dual Control Station upgrade to Celerra Network Server version 6.0, users can log in successful using LDAP on the primary. When the Control Station fails over, the LDAP login no longer works. The following error is displayed: Session timeout. Relogin and try this operation again. This issue only affects LDAP users; local root/nasadmin accounts and global accounts are still available. Workaround Run /nas/site/cst_setup -reset to resync the CST settings. Once those settings are synced, you will be able to failover/failback and successfully be able to login each time using LDAP. Configuring LDAP domain settings When the Control Station is freshly installed and the Custom Active Directory page is used to configure the initial LDAP domain settings, the distinguish name is not set properly. Testing the Domain Settings generates an error similar to the following: Primary server 12.3.45.67 bind error. This issue only occurs during the initial setting. Once the
EMC Celerra Network Server Version 6.0.70.4 Release Notes
29
Known problems and limitations
initial setting is done, the issue never happens again. Workaround Use the 'Default Active Directory' option to configure the initial LDAP domain settings and then use the other options to change the settings. Copying large files to remote over NFS When a user copies very large files from the Control Station to a remote file system over NFS on a slow network, other users of that Control Station may experience slow performance when running CLI commands. For dual Control Stations, this issue can cause the Control Stations to fail over. Over time, the Control Station will correct itself, and performance will return to normal. Domain attributes After a Control Station failover on a Celerra that is a member of a storage domain, the administrator will not be able to use or see domain changes made while the Celerra was in the CS failover state. Workaround Continue using the Celerra with the domain attributes it currently has, or remove the Celerra from the domain and re-add it. Linux automount The /nasmcd/quota and /nasmcd/rootfs directories are used to nfs mount the root FS of the Data Movers using the Linux automount package. Occasionally, automount's attempts to perform this nfs mount fail, resulting in the requested Data Mover file system not being available. This problem happens more frequently when there are multiple automount requests for the same share over a short period of time. When the mount does fail, retrying the attempt after a short delay will usually be successful. One area that may be impacted by a mount failure is quota management, as it relies on access to files under /nasmcd/quota to retrieve quota information. The quota functionality and other operational aspects of the Data Movers are not affected. Security auditing When security auditing is stopped for any reason, for example, due to a lack of disk space, no alert is raised to the administrator.
30
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
Deduplication
Deduplicated data space not reclaimed A race condition occurs between removal of deduplicated files and file system unmount operations. Under certain conditions, deduplicated data space may not be reclaimed from the file system when files are removed. Messages similar to the following are displayed in the server_log when this issue occurs: 2010-05-09 09:40:57: 13163036672: DEDUPE: 3: Cannot find file system with id 15, cannot cleanup object <dart://rde..." Free Space During the deduplication process, the file system must have enough free space available that is equivalent to the size of the original file to be deduplicated plus the size of the compressed version of the file to be stored. An additional 1 percent of the file system must be free, or if autoextension is enabled, an additional 1 percent below the auto-extension threshold must be free. Restoring space-reduced files Restoring space-reduced files from a PAX-based NDMP backup into a file system causes the deduplication state to be set to suspended if the deduplication state is not currently enabled. I/O errors reported or deduplicated data space not reclaimed A race condition occurs during domain re-initialization. Under certain conditions, I/O errors can be reported or deduplicated data space may not be reclaimed from the file system when files are removed. Messages similar to the following are displayed in the server_log when this issue occurs: removeFromSecondaryStorage could not determine ssp
File-Level Retention
Prior to Celerra version 5.6.43.9, FLR protected empty locked files from being deleted. As of version 5.6.43.9, both FLR-E and FLR-C allow the deletion of empty locked files and empty append-only files. Any locked or append-only file that contains data will be protected. This behavior will be changed in a future release such that all locked and append-only files will be protected regardless of whether they contain
EMC Celerra Network Server Version 6.0.70.4 Release Notes 31
Known problems and limitations
data.
File system
Mounting file systems When mounting a file system as read-only on two (or more) Data Movers, the following error message appears on the copy file system pop up: "There are no Data Movers interconnect available on server_3, server_2". The error message is misleading and can be ignored. EMC does not support creating a copy of a file system that is to be mounted to more than one Data Mover. nas_fsck command A race condition occurs when using nas_fsck command on file systems smaller than 2G. This causes the file system to be unavailable to the user. Timestamp mismatch After a restore operation, a timestamp mismatch occurs on the softlinks with file names longer than 83 characters. The timestamp shows current time (i.e. time of restoration) instead of the original timestamp on the link.
Installation and upgrades
Celerra now joined to domain during upgrade The Celerra is now being joined to the CLARiiON domain automatically during an upgrade. You can use Unisphere to elevate the privileges of any user that has been transferred from the Domain to the Celerra. This does not occur for gateway machines, however. For gateways, users will need to join the Celerra to the domain they want it to be a part of. Celerra Startup Assistant After a fresh factory Express Install of an NS40FC system, the CSA fails on the Pre-Configuration apply changes page with the following error: CLARiiON's Domain Join Setup | Command Failed. If this occurs, ignore the error; this step should not have been tried, and failure poses no operational problems.
32
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
CallHome messages The Celerra uses connectEMC 2.6 software to deliver callhome messages to EMC. The outgoing mail functionality of this version issues duplicate HELO/EHLO commands. This is in accordance with RFC 2821, however, some mail servers do not support this (especially older mail servers). If this occurs in your environment, consider updating your mail server. Control LUN requirements enforced There is strict enforcement of the control LUN size and HLU numbers for fresh installations. An exact match between the LUN capacity and the correct HLU value is required. Mismatches will generate a "Control LUN Check Failed" error message during installation. You must correct the control LUNs before installation can complete successfully. Determining NAS version While using CUT to perform an upgrade from 5.6 to 6.0, the CUT tool presented a pop up saying it could not determine the nas_version, after or during the Linux upgrade portion of the upgrade. Retrying the task succeeds. Error message While upgrading from version 5.6 to 6.0, the following error message is displayed: Error - cannot open /usr/sbin/ext2online /dev/hda1. This error can be ignored, as it does not stop the upgrade and the task completes successfully. Gateway and FC-enabled systems During upgrades of gateway and FC-enabled systems, a message stating that the Celerra was joined to the back-end domain is presented. That message is incorrect, and can be ignored. The Celerra can be joined to the domain using CLI commands or via UEM on the domain master system. Incorrect certificate Users may experience an invalid certificate after fresh installing their system. Workaround Set the correct time on the Control Station and regenerate the certificate, or wait for the certificate to become valid and login again.
EMC Celerra Network Server Version 6.0.70.4 Release Notes 33
Known problems and limitations
nasadmin credentials If you pause and resume the upgrade, the "Clariion domain nasadmin Credentials" screen is not displayed. Workaround Use the nas storage command to change the password of nasadmin. nasadmin password When logging in globally and attempting to change the nasadmin password, an authentication error is generated and users are not allowed to change the password. The workaround is to use a nas storage command to change the password of nasadmin. Password confirmation The "Clariion domain nasadmin Credentials" screen in the Celerra Upgrade Tool (CUT) currently does not have a confirm password field. Users can mistype the password without knowing it, and will be unable to log in. PUHC "Check for running tasks" fails While upgrading from version 5.6 to 6.0 using USM, the PUHC "Check for running tasks" fails and offers a set of actions to take. These actions do not work. The first step says to type the following; "<br> * nas_license -list". Neither of the 2 steps offer strings work if typed in as specified and produce errors. Supported hardware When attempting to run USM on unsupported hardware (CNS|CFS|NS350|NS500|NS600|NS700), the Celerra Upgrade Tool will correctly exit but exit code is not passed back to the USM. Therefore, the USM tries to continue the upgrade process but will fail on the next step that interacts with the Celerra system. The error returned at that point does not give a clear indication of what the actual failure was. If you see unclear error messages while using USM, make sure that you are using a supported hardware platform. Upgrade paused During an upgrade with the CLI, and when it was paused during the Control Station reboot, the Upgrade wizard and on the welcome screen tells user that the upgrade is paused and allows user to resume the
34 EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
upgrade. Do not use the wizard to resume an upgrade that is started in the CLI. Multibyte support Using multibyte characters in fields in which they are not supported. If you use multibyte characters in data element fields that do not support multibyte characters (for example, checkpoint schedule descriptions), these fields will not display correctly. If you want these fields to be readable, you will have to recreate or rename the existing data element text using the ASCII character set. Refer to PRIMUS emc224755 for more detailed information about this limitation. Using multibyte characters with ECC Celerra version 5.6.40.3 and later may cause discovery and name display issues for ECC 6.x if certain Celerra object names use non-ASCII characters. The relevant names are: File system name Mount point name CIFS share name NFS export name Quota user name
Discovery issue If non-ASCII characters are used and if the name is more than 63 characters, discovery may not work. Garbled name display issue If these names use non-ASCII characters, the names do not display properly in ECC. However any ASCII characters used in the name do display properly. Furthermore, the name itself and the underlying data on the Celerra system are not affected. Suggested Workarounds Use only ASCII characters in the specified object names when using ECC for Celerra discovery. If multibyte characters must be used, keep all applicable object names under 63 characters. Since ASCII characters display properly, append or prepend ASCII characters to object names so you can use these characters to find the correct name.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
35
Known problems and limitations
Refer to Primus emc229599 for more details about this issue.
NDMP
Incremental NDMP backups and NDMPcopy operations Incremental NDMP backups and NDMPcopy operations use the same last backup timestamp tracking for each data set on a Data Mover. A data set in this context means the point where the backup or copy operation starts (e.g., the root of a file system or a specific sub-directory within a file system). This means that if incremental NDMP backup and NDMPcopy operations of the same data set (i.e., starting from the same point within a file system) are interleaved, then the following will occur: The incremental backup will contain only changes in that data set since the NDMPcopy operation instead of since the last backup operation. The incremental NDMPcopy of that data set will copy only those files changed since the NDMP backup instead of since the last NDMPcopy operation. To avoid this, do not interleave NDMP backup operations with incremental NDMPcopy operations of the same data set; if you do, perform a full NDMP backup of that data set after any NDMPcopy activity. Vendor support Not all NDMP vendor versions support this version of Celerra Network Server. For the latest NDMP Vendors/Versions/Patches and restrictions supported in this version of Celerra Network Server, refer to the Backup section of the E-Lab Interoperability Navigator.
Network
Do not disable IP reflects when using IPv6. System may panic when using IPv6 if IP reflect is disabled.
NFS
Cached files If a client mounts exported parent and child paths of the same file system, it is possible to access the same file from two different mount points. If client side caching is enabled, the client may end up maintaining two different versions of a cached file.
36
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
Celerra Data Migration System Access to online migration file systems while using CDMS is not supported by NFSv4 clients until migration completes. When NFSv4 is used, it triggers directory migration of subdirectories and requires global lock for the entire Data Mover. Therefore, the Data Mover will be inaccessible during the migration. NFS export failure If the export of a corrupted file system fails, the export entry of the file system which was corrupted & cleaned using FSCK is not seen in the export list. However, if user manually exports it again, the entry can properly be seen in the NFS export.
Quotas
Performance statistics Termination summary is not sorted for Set statistics if a statpath name is duplicated in the server_stats command. The following example will not have a sorted termination summary: server_stats server_x monitor store.diskVolume,store.diskVolume -sort writes. This issue does not impact the sorting of the sampled data displayed periodically during the server_stats session, only the termination summary at the end of the session is impacted. Workaround Not using duplicate statpath names when -sort is specified will eliminate this issue. For example: server_stats server_x –monitor store.diskVolume -sort writes.
Replication
nas_copy command When using Replication V2 and the nas_copy command, there is a window where if the X-Blade reboots, results may not be properly reflected. The transfer will complete but the task is not cleaned up and the appropriate status is not reported back to the user. Once completed, the task can be manually aborted.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
37
Known problems and limitations
Security
Changing LDAP settings When running the server_netstat command, or updating the LDAP settings, the following error was displayed: The lockbox stable value threshold was not met because the system fingerprint has changed. Reset the SSV values by running "/nas/sbin/cst_setup -reset" from the CLI to resolve the issue.
SnapSure
Parameter file Do not insert a new-line character at the end of a nas_param file (/nas/site/ or /nas/sys/). The empty line causes fs_ckpt to return an encodeable.extract error. Writable checkpoints operations Writable checkpoints do not support the following operations and features: Checkpoint refresh Checkpoint schedule Celerra replication CDMS File-level retention File system extension iSCSI MirrorView/SRDF TimeFinder/RDF VTLU
Data mover panic (425057) The data mover panicked. The following error was displayed: DART panic/fault message: >>PANIC: I/O not progressing LastVol touched Sn891 Kind 5 (ptr=0x65d74b08) Workaround Contact your service provider.
38
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
SRDF/S
nas_rdf –activate The nas_rdf –activate –reverse command may fail to swap device-pair personalities during the activate. This occurs when the source Celerra has a TimeFinder/FS snapshot/s created on R1BCVs which use TimeFinder Emulation mode (Symm 8 supports Emulation mode only). The error message below is specific to the personality swap and having remotely protected storage, and does not impact the successful activation of the destination Celerra system. Error message in nas_rdf –activate –reverse: An RDF 'Swap Personality' operation execution is in progress for device group '1R2_500_3'. Please wait... Cannot use the device for this function because it is a Copy Session target ERROR 13431799810: Swap operation of device group 1R2_500_3 failed due to the improper backend state or configuration. Run 'nas_message -info 13431799810' to get full description and recommended action. Workaround Canceling the relationship between the R1BCV mirrored volumes and the R1STD volumes is needed to complete the swap operation. After the SRDF restore completes, the SYMAPI will automatically request a full establish during the subsequent fs_timefinder –M refresh command to the snapshots built on the canceled R1BCV volumes.
Table 2 nas_rdf –activate –reverse procedure Step Action
1
To check symm id of source Symmetrix, run /nas/symcli/bin/symcfg list on the destination SRDF system. To check the mirrored volumes, run nas/symcli/bin/symmir –sid <source symm id> list on the destination SRDF system. You can also double check if Emulation mode is used from the output. To find R2 device group, run /nas/symcli/bin/symdglist on the destination SRDF system.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
2
3
39
Known problems and limitations
4
To find R1BCV volumes, run /nas/symcli/bin/symrdf –g <r2 group name> query –bcv on the destination SRDF system. If the same device ID is seen in the output that is the R1BCV device you need to cancel. To prepare for the cancel operation, create a file which has all R1BCV volume pairs to cancel on destination SRDF system. Example of the file: # cat file # <standard device id><bcv device id> 0068 00B8 0069 00B9
5
6
To cancel the R1BCV volumes, run /nas/symcli/bin/symmir –sid <source sid> -f <file name> cancel on the destination SRDF system. To confirm the canceled volume, run /nas/symcli/bin/symmir –sid <source symm id> list on the destination SRDF system. The canceled volumes should not be listed in the output. To compete the activate reverse operation, run /nas/sbin/nas_rdf –activate –reverse on the destination SRDF system.
7
8
nas_rdf –restore The nas_rdf –restore command may fail on SRDF system which uses TimeFinder Emulation mode (Symm 8 supports Emulation mode only) with R2BCV mirrored volumes. Error message in nas_rdf –restore: An RDF 'Failover' operation execution is n progress for device group '1R2_500_3'. Please wait... Cannot use the device for this function because it is a Copy Session target CRITICAL FAULT: SRDF restore failed for 1R2_500_3 Workaround Canceling the relationship between the R2BCV mirrored volumes and the R2STD volumes is needed to complete the restore operation.
40
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
After SRDF restore completes, full establish is required when we do mirror on again to the canceled R2BCV volumes.
Table 3 nas_rdf –activate –restore procedure Step Action
1
To check the mirrored volumes, run /nas/symcli/bin/symmir –sid list on the destination SRDF system. You can also double check if Emulation mode is used from the output. To find R2 device group, run /nas/symcli/bin/symdg list on the destination SRDF system. To find R2BCV volumes, run /nas/symcli/bin/symrdf – g <r2 group name> query –bcv on the destination SRDF system. If the same device ID is seen in the output, that is the R2BCV device you need to cancel. To cancel the R2BCV volumes, run /nas/symcli/bin/symmir –g <r2 group name> cancel -force on the destination SRDF system. To confirm the canceled volume, run /nas/symcli/bin/symmir –sid list on the destination SRDF system. The canceled volumes should not be listed in the output. To complete restore operation, run /nas/sbin/nas_rdf – restore on the destination SRDF system.
2 3
4
5
6
Nas_rdf –init command fails Problem: During the SRDF discover remote storage devices process, the nas_rdf init command fails and returns the following error: "Unable to setup remote devices". Workaround: Delete the device group for R2 devices using /nas/symcli/bin/symdg delete <DgName> -force. For example: /nas/symcli/bin/symdg delete 1R2_500_13 –force. Then run nas_rdf –init, to re-create the R2 device group.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
41
Known problems and limitations
SSD drives not supported SRDF/S does not support solid state disk (SSD) drives. If you attempt to configure an SSD drive with SRDF/S, Celerra health check operations reject the device as an unsupported storage.
Unisphere
The following notes apply to Unisphere. Adding a system to a Celerra-CLARiiON domain When users attempt to add a system to a Celerra-CLARiiON domain and uses a LOCAL scope to authenticate to the remote system, it appears to work. However, the system does not actually get added to the domain. The correct procedure is to authenticate to the system to be added with a GLOBAL scope user that is part of the root group. Adding a system to a Celerra-Symmetrix domain When attempting to log into a Celerra that has a Symmetrix backend and is part of a storage domain, you may see the following message on the console: "Target system already managed." If this occurs, log out, wait 15 minutes, and log in again or close the domain master page and restart your browser session. Adding a system to a second domain Adding a Celerra to one domain followed by adding it to a second domain from a non-master CLARiiON may result in the Celerra appearing to be in both domains at the same time, depending on which system scope you are in. Workaround Remove the Celerra from the first domain before adding it to a different one. If you have already added it to both you can still remove it from the one that you are trying to get it out of. File Type field is Blank in Export to File Dialog with JRE 1.7 The “File Type” field is empty when the “Export to File” dialog is displayed. This happens only with JRE 1.7. No file is created in the destination folder and the dialog closes without an error. Workaround The workaround is to select the file type from the “File Type” drop down.
42
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
JRE install auto link gives an error Unisphere will prompt a message to install JRE 1.6 if JRE is not installed. The JRE install link in this message does not work. Workaround The workaround is to install JRE 1.6 directly from the Oracle website.
Alerts For performance and memory usage reasons, the alerts displayed in Unisphere are now limited to 400. A new alert has been added to notify the administrator if there are more than 400 alerts:
Severity: Critical System: <hostname> Domain: Local Created: <timestamp> Message: Too many alerts on <hostname>. Full Description: This system has more than 400 alerts. Until you delete the alerts, you will only see the top 400 alerts (by date and severity) for this system. Recommended Action: Delete the alerts on this system to clear them from the list of alerts. Event Code: 0x240060005
Do not attempt to delete the "Too many alerts" alert in Unisphere. Unlike most of the other alerts, this alert will clear itself if the condition is resolved and the total number of alerts for a system is reduced to fewer than 400. If you include this new alert in a selected list to be deleted, the alerts will appear to be deleted, but will be reloaded once a new browser session is started. A failure will be posted to /nas/log/webui/apl_tm.log, but no failure will be displayed in Unisphere. In /nas/log/webui/apl_tm.log:
2010-06-24 10:20:21.260 webui-N:[[email protected]]:0:29366:S:402345-Delete alert(s). 2010-06-24 10:20:28.783 webui-N::0:29366:E:402345-Delete
EMC Celerra Network Server Version 6.0.70.4 Release Notes
43
Known problems and limitations
alert(s).(complete=true,failed=true) <- Note failed=true
Browsers EMC requires the use of one of the following browsers to run Celerra Manager: Internet Explorer 6.0 SP2, 6.2.3 through 7.x.; Internet Explorer 7.1 is recommended. Internet Explorer 8.0 is also supported. Firefox 3.0 or later.
Firefox Firefox Password Manager (PM) is enabled by default. This feature records and stores the password that you enter on a website and then automatically fills in the password entry field on your next and subsequent visit. The password field masks the password with asterisks (*). To prevent Firefox from remembering passwords: 1. 2. Click Tools > Options... and select the Security tab. Under Passwords, clear the Remember password for sites checkbox.
Browser session hang Depending upon a client's particular browser configuration (browser type, version, JRE version, native OS, etc.), launching the USM tool from the UEM may result in a browser session becoming hung. If you verify that there are no hidden dialogs awaiting confirmation, terminate the browser sessions and all associated processes. Rebooting the client system may be required to restore full browser operation. To prevent this issue in the future, launch the USM directly (i.e., not via UEM). Changes from other user interfaces Any changes made to the network server using another user interface such as the Celerra CLI are not automatically shown in Unisphere. Refresh the browser to see these changes. Changing default password Changing the CLARiiON array default admin password breaks the ability of the Celerra to communicate to the CLARiiON or make any changes to its database [diskmark, discoveries, etc. will all fail].
44
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
Checkpoint schedule page If the time zone of the Celerra Control Station and the client system running the UEM are different the checkpoint schedule page will display times in UTC (coordinated universal time). Confirmation dialog inactive While using Unisphere, a confirmation dialog appears briefly, and then moves to the background. Users need to manually re-focus the window and bring the dialog box to the foreground. Dashboard screen The data is not presented in the Celerra Capacity View screen in an appropriate amount of time. However, the other dashboard screens are loaded successfully. Workaround Close the Unisphere window and clean out the persistence files. On the next restart of Unisphere - you will not see this issue. To clean out the persistence files, follow these steps: 1. 2. 3. 4. Close the Unisphere application. Go to the Unisphere client log folder; the default should be C:\Documents and Settings\<username>\emc\Unisphere. Delete the persistence.ser file. Re-launch the Unisphere application
Dashboard summary The Dashboard Summary page did not contain view block. This is unexpected, but those blocks can be re-added using the customize button. File system delete confirmation The 'OK' button is disabled on the Unisphere File System Delete Confirmation page for a file system with File-Level Retention enabled at the Compliance level. The table on the File System Delete Confirmation page indicates that protected files still exist on the file system even though the File System Properties page and the 'nas_fs -i' CLI command both show that all file protection has expired for the file system. The user is unable to delete the
EMC Celerra Network Server Version 6.0.70.4 Release Notes 45
Known problems and limitations
file system using Unisphere. Workaround: The File-Level Retention status is now cached for performance reasons. Force the file retention data to be updated by selecting the Unisphere Refresh for the File System List page (in the upper right corner of the File System List panel) or by opening a new browser session. Help sessions While getting a session timeout in a help session, some help text is displayed, but none of the icons are displayed. Clicking OK button has no effect. Workaround: Use the close button of browser to close the window. JRE memory limit For optimal UI performance, EMC recommends that the memory limit allocated to the JRE be increased. The default size of 64 MB is often too small and may result in Out of memory errors. To set the memory size, from the Start menu in Windows, select Settings > Control Panel. Double-click the Java Plugin icon. Select the Advanced tab. In the Java Runtime Parameters, enter the following: -Xmx512m The argument specifies the maximum memory. Typically, about ½ the size of your overall workstation's memory is recommended as the maximum memory value. For instance, if your workstation has 1 GB of memory, then use -Xmx512m for the JRE runtime parameter. If your workstation has 2 GB or more memory, you can adjust the max JRE memory as needed to avoid running out of Java memory and possibly improving overall performance. Logging in after browser crash After a browser crash, the GUI reports "system already managed" when user tries to log back in. Logging in via multiple screens Logging into Celerra via two or more screens may succeed, but logging out of one of the sessions causes the other sessions to lose contact. Login screen When you log into Unisphere via a Celerra system, Unisphere will not display a login screen if you click the “yes” button on the security
EMC Celerra Network Server Version 6.0.70.4 Release Notes
46
Known problems and limitations
warning dialog, without also checking the “Always trust content” checkbox on the same dialog. MMC snap-in MMC installer program may not completely uninstall all components when requested to do so. However, when a new version is installed, the components are overridden and the application runs as expected. Modifying system properties When storage system properties (specifically - security credentials) are modified in the GUI, the following (example) warning message is displayed in addition to the expected icons: Modify CLARiiON CLARiiON CX4-960 APM00093300340. Invalid component information :BackendResume: This error is a result of the component information cron job querying the status while the credentials are in the process of being modified. The correct component information will be refreshed during the next iteration of the cron job (approx. 5 to 10 minutes). Mounts table view The file system column in the Mounts table view gets stuck at "Loading" for multiple minutes. The rest of the table is populated correctly. Workaround: Refresh the page using the green arrow icon. nbsnas/jserver filling up at 100% Unisphere jserver mounted on /nbsnas/jserver filling up at 100% prevents installation. When this occurs, the following error is displayed:
The feature identified has detected a potential capacity issue that may need the attention of the system administrator. The device name and mount point is indicated in the event posting. Recommended Action: Check that the mount point's condition is in fact approaching cautionary levels. Use normal administrative procedures to see if some free space can be returned to the mount point, such as removing unnecessary files or reorganizing areas of this mount point off onto other less full mount points on the system. If no means are available to reduce the space consumption on this mount point then you may want to investigate if there is a possibility that this mount point needs to be expanded to allow for more free space. Contact your Authorized Service Provider if you need assistance with this process or need to discuss acquiring additional storage system capacity.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
47
Known problems and limitations
Provisioning wizard If an error occurs while in the Provisioning Wizard, you are instructed to correct the error and are given a choice of either restarting the provisioning process or canceling the activity. Rolling back intermediate tasks may fail or hang. If you cancel the activity by clicking the NO button, all the previous tasks are rolled back and you will be redirected to the first screen of the Provisioning Wizard. Slice volumes not listed in wizard Occasionally, slice volumes created through the CLI or Unisphere Volume Create page are not listed as available volumes in the New File System Wizard when selecting Manual Volume Management and then attempting to create a new stripe volume from the slice volumes. Storage domain When the Celerra belongs to a storage domain that is destroyed, the storage domain users on the Celerra Control Station are not automatically moved to the "defunct" state. To correct this, manually remove all the storage domain users from the system by running the command: /nas/http/webui/bin/admin_management.pl defunctStorageUsers Then remove the Celerra from the domain. User’s tab displays unrelated messages Unisphere > Settings > User's tab displays messages which are unrelated to that screen. These messages are a result of a query that was initiated by the user on some previous table screen. Workaround Visiting another table screen and then to coming back to the table screen where the messages were seen should remove these messages.
Other restrictions
Caution Do not install Linux non-Celerra applications on the Control Station. Doing so consumes resources and might adversely affect system operation. Celerra Control Station is a dedicated controller and not an applications platform.
48
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
Do not manually edit the nas_db database without consulting EMC Customer Service. Any changes to this file could cause problems with your Celerra installation.
Command line scripts The NAS database stores specific information required for each Data Mover. The Celerra Network Server automatically performs a backup of the entire database every hour and saves it to a file named nasdb_backup.1.tar.gz. These backups start one minute after the hour. EMC recommends that you check your scripts to make sure they are not scheduled to start at the same time as the backups. During backup, the database becomes locked or inaccessible, and some commands that rely on the database might fail. Correctable hardware errors Transient correctable hardware errors are currently being handled by the firmware. Although rare, if enough of these errors occur, a watchdog panic can be triggered. In a future release, these errors will be handled by the Celerra operating system instead of the firmware which will improve efficiency and prevent the watchdog panic. server_cpu The server_cpu ALL -halt now command does not complete when using Celerra NSX Network Servers. This is because the NSX needs at least one Data Mover to be up in order to access its NAS partitions through NBS. server_http The server_http command fails if you issue it on a Data Mover while running fsck. Other commands (including FileMover API) may not work until fsck completes. server_stats On a Data Mover with more than 500 mounted file systems, the following statistics fail to display: - store.logicalVolume.metaVolume metaVolumes-std. If any of these statistics are supplied as an argument to server_stats -monitor option, the server_stats will hang. Use Ctrl-C to terminate the session. server_sysconfig If you use the option string of server_sysconfig, the entire string must be
EMC Celerra Network Server Version 6.0.70.4 Release Notes 49
Problem prevention
enclosed by double quotes. Virtual device naming When creating virtual devices, such as Fail-Safe Network (FSN) devices, Ethernet channels, and link aggregations, do not use the following names as the name of the device:
Amgr Blksrv Client Cmfap Dep dfe echo escon http i82596 icmp icmp6 ip ip6 lockd lpfn Mac meshforwarder nd nfs nfsPrimary nwh pax pcnfs pfkey pipe pmap rcp rfa scsicfg smb smpte spstream Ssl sslpipe statd StreamHead tcp tcpnfsmesh tcprpc tcprpc Tftp Toeframechk upd udpnfsmesh
Problem prevention
The EMC Customer Service center has reported a number of calls concerning the following issues. Please take a moment to familiarize yourself with these potential issues.
Administrative activities
Log in as nasadmin As with any UNIX-based operating system, refrain from performing normal administrative activities as root user. Log in using the administrative user account instead. Shutting down a Data Mover or deleting an internal interface Shutting down a Data Mover or deleting an internal interface will prevent you from issuing any server_ commands against that Data Mover. Whenever you issue any command on the Celerra Control Station CLI server, the command is sent to and executed by a Data Mover. If that Data Mover is down or unreachable, the command will fail.
50
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Problem prevention
Celerra AntiVirus Agent (CAVA)
AntiVirus (AV) threads If a large number of files need to be checked at the same time, the CAVA file checking process may be slow depending on number of threads configured, number of CAVA servers, and speed of the CAVA servers. If the CAVA servers cannot process file checks quickly enough, then all AntiVirus (AV) threads will become busy and the AV “collector queue” will increase, indicating that file check requests are queueing up. Increasing the number of AV threads (10 by default) will allow more files to be checked in parallel and may improve performance. However, the number of CAVA engines may need to be increased to satisfy a heavy load. Number of CAVA engines It is possible to configure your EMC Celerra installation with only one Celerra AntiVirus Agent (CAVA) engine. However, this non-highavailability configuration can make your installation susceptible to data unavailability in the event that the CAVA engine goes offline for any reason. The best method for preventing this type of outage is to increase the number of available CAVA engines. Additional information for configuring the Celerra AntiVirus Agent can be found in the Using Celerra Anti Virus Agent technical module. This document describes the parameter setting that will specify shutdown actions to take when no servers are available. Removing AV user from Data Mover password file If you remove an AV user from the Data Mover password file, CAVA will not work. Depending on the setting you have on CIFS, you could use up all the CIFS threads because they are used for virus checking or it will shut down CIFS altogether.
CIFS
Deleting machine account from active directory Deleting a machine account from active directory will result in loss of access to the CIFS server and its resources. Deleting the computer account of any CIFS server, including a Celerra Data Mover, from the domain of which it is a member will prevent that CIFS server from being
EMC Celerra Network Server Version 6.0.70.4 Release Notes 51
Problem prevention
able to authenticate both itself and users attempting to connect to it with the domain. This will mean that users will not be able to access data hosted by the CIFS server. Deleting server out of the domain Deleting a server out of the domain will result in shares associated with the server to be unavailable. The same applies to deleting the machine account. Naming CIFS shares Do not name your CIFS share "all" or you risk deleting all of your shares when trying to delete it. NetBIOS name conflict Instead of destroying the conflicting NetBIOS, simply rename it using the existing server_cifs <server_x> -rename -netbios <oldname> <newname> command. Changing a NetBIOS name using this command will update the share and local group and enable you to reuse it. LGDB and ShareDB information associated with the server are persistent. If the target name of a rename conflicts with LGDB or ShareDB information, re-create the old server and rename it to a name that will not conflict with your naming structure to free the old slot.
Control Station
Applications The Control Station’s primary function is to manage the Celerra. The use of applications that do not come installed on the Control Station is not recommended. The Celerra Control Station is not tested or certified with any of these applications and there could be issues with memory usage or CPU time that may affect normal operation. Dual Control Stations If you have dual Control Stations on either (CFS/CNS/NSxxx) and they concurrently try to access the /nas file systems, the nasdb may become corrupted and the Data Movers will be unable to communicate with the Control Station.
52
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Problem prevention
File System
Deleting files from Data Mover or file system Files that you delete will be permanently lost unless you have a backup copy. Files can be backed up and restored through the Celerra SnapSure, TimeFinder/FS, Replicator or by using standard tape backup methods. Deleting file systems File systems that you delete will be permanently lost unless you have a backup copy. Backups can be created via TimeFinder/FS, Replicator or other backup methods. File system full Insufficient space in a file system will prevent users from writing to that file system. In addition to monitoring file system usage, set an event notification for file system full condition and enable automatic file system extension with a conservative threshold. Refer to the Configuring EMC Celerra Events and Notifications technical module for more information. Mounting file system to /tmp Some variants of UNIX or Linux periodically delete some or all of the files under the /tmp directory. If you mount a file system under /tmp, you risk having the contents of that file system deleted by the automatic cleanup process. Splitting BCVs When a production file system (PFS) is mounted in a read/write fashion to a Data Mover and has BCVs established, the command to split (or mirror -off) the BCVs from the PFS volumes must be initiated using the Celerra fs_timefinder command. The fs_timefinder command ensures consistency of the BCV snapshot by managing file system access and Data Mover log operations when the volumes are split. These additional considerations are not available directly from Solution Enabler commands (symmir).
Name Resolution SunOne/iPlanet
When using SunOne/iPlanet, make sure that the defaultServerList
EMC Celerra Network Server Version 6.0.70.4 Release Notes 53
Problem prevention
attribute on the LDAP server is set to use IP addresses and not host names. If hostnames are specified and the Celerra is configured to use SunOne/iPlanet for host resolution, a deadlock can occur.
Networking
Changing an IP address Changing an IP address on a Data Mover can result in a naming (DNS/NIS) mismatch and render the IP address inaccessible. Changing or removing DNS settings When you make changes to the DNS server, such as IP addresses make sure you reflect those changes in the Data Mover. Failure to do so may result in accessibility issues. Removing the DNS server will prevent the CIFS server from communicating with the domain. Any CIFS server, including those configured on a Celerra Data Mover that is a member of a Windows 2000 or later domain, use DNS to locate the Windows domain controllers that it needs to communicate with to operate. Disabling DNS in such an environment will prevent the CIFS server from locating Windows domain controllers, therefore, prevent the CIFS server from authenticating users attempting to connect to it. This means that users will be unable to access the data hosted by the CIFS server. Connecting an internal switch to an outside network The internal networks in the Celerra are designed to provide private communication paths between Data Movers and the Control Station. If that network is connected to an external network, the resulting routing, addressing issues, and/or volume of network traffic may disrupt the communications between the Data Movers and the Control Station, which will in turn disrupt the operation of the Celerra. Duplicate IP Duplicating an IP address on two Data Mover interfaces will have the same result as duplicating IP addresses on any two machines in a network. The network will either be unable to determine which location it should deliver packets to or block all traffic to both machines on the assumption that one of them is attempting to gain unauthorized access to information by impersonating the other.
54
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Problem prevention
FSN devices Create Fail-Safe Network (FSN) devices on multiple network interface cards (NICs) to ensure connectivity in the event of a hardware failure. Incorrectly adding network route Incorrectly adding a network route will cause network access to be limited to the same subnet/broadcast domain. The default gateway, or route, address tells any network client, including a Celerra Data Mover or Control Station, where to send IP packets that are outside its local subnet. If the default gateway is misconfigured, the network client will not be able to contact any host that is not within its local subnet. If the Celerra relies on hosts, such as domain controllers, outside its local subnet, it may not be able to offer service.
Parameters
Changing parameters Changing parameters could significantly affect the function of the Celerra Network Server. Take the following precautions before you make changes to Celerra parameters: Do not change system parameters without first consulting with EMC Customer Service. Change server parameters one at a time, in small increments, preferably in a non production environment. Be sure to thoroughly test each change before making additional changes. Keep a detailed log of which server parameters you changed and the reason for the change. This information will be useful if a problem arises because of the change.
Managing parameters EMC recommends using the server_param command or the Celerra Manager to manage parameters. If you manually edit the param file, note the following: Parameters are case-sensitive and must be entered exactly as described in the Celerra Network Server Parameters Guide. A typographical error, the omission of a dot, incorrect capitalization, or a value out of range could result in adverse effects.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
55
Technical notes
Storage
Multiple LUN 0s Assigning multiple LUN 0s will result in Data Movers being unable to boot successfully. Powering down all SAN Switches The Celerra does not store any user data itself. All data is stored on the Symmetrix or CLARiiON disk array(s) that the Celerra is attached to. With a Celerra gateway model the Celerra is attached to the disk array(s) it uses via a Fibre Channel SAN. If that SAN is unavailable then the Celerra will not be able to access the disk array(s) it relies on and hence will not be able to offer normal service. Reassigning volumes with SDDR or making fibre changes Like any host of a Fibre Channel SAN or array, the Celerra will remember the address of the LUNs that it uses in the attached disk arrays. Any change in the identity or location of those LUNs without a corresponding reconfiguration of the Celerra may result in the Celerra being unable to access the LUNs it needs to provide normal service. Reassigning Celerra LUN with Navi Same as Reassigning volumes with SDDR or making Fibre changes. Unbinding/rebinding LUNs Same as Reassigning volumes with SDDR or making Fibre changes above. Write disabling R1 and R2 In SRDF configurations, setting R2 disks to "read only" will cause R1 disks to become "read only" as well, and cause access issues.
Technical notes
Configuration guidelines
The values listed in this section represent the highest or lowest values tested by EMC. In some cases, actual capacities may exceed the capacities tested by EMC. Theoretical means in theory, not always practical. Refer to the E-Lab Interoperability Navigator for the most up-to-date information regarding these guidelines.
56
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Table 4 CIFS Guidelines Guideline/specification CIFS TCP connection Maximum tested value 64K (default and theoretical max.), 20K (max.tested) Comment Param tcp.maxStreams sets the maximum number of TCP connections a Data Mover can have. Maximum value is 64K, or 65535. TCP connections (streams) are shared by other components and should be changed in monitored small increments. Unicode: The maximum length for a share name with Unicode enabled is 80 characters. ASCII: In ASCII mode, the maximum share length is 12 characters. You may notice a performance impact when managing a large number of shares. Limited by the number of network interfaces available on the Data Mover. From a local group perspective, the number is limited to 509. NetBIOS and compnames must be associated with at least one unique network interface. NetBIOS names are limited to 15 characters (Microsoft limit) and cannot begin with an@ (at sign) or a - (dash) character. The name also cannot include white space, tab characters, or the following symbols: / \ : ; , = * + | [ ] ? < > “. If using compnames, the NetBIOS form of the name is assigned automatically and is derived from the first 15 characters of the <comp_name>.
Share name length
80 characters (Unicode). 12 chars (ASCII)
Number of CIFS shares
10,000 per Data Mover 40,000 per Celerra 509 (max.)
Number of NetBIOS names/compnames per Data Mover
NetBIOS name length
15
EMC Celerra Network Server Version 6.0.70.4 Release Notes
57
Technical notes
Comment (ASCII chars) for NetBIOS name for server
256
Limited to 256 ASCII characters. Restricted Characters: You cannot use double quotation ("), semi-colon (;), accent (`), and comma (,) characters within the body of a comment. Attempting to use these special characters results in an error message. You can only use an exclamation point (!) if it is preceded by a single quotation mark (’). Default Comments: If you do not explicitly add a comment, the system adds a default comment of the form EMC-SNAS:T<x.x.x.x> where <x.x.x.x> is the version of the NAS software. For integration with Windows environment releases greater than Windows 2000, the CIFS server computer name length can be up to 21 characters when UTF-8 (3 bytes char) is used. The maximum number of Windows domains a Data Mover can be a member of. To increase the max default value from 32, change param cifs.lsarpc.maxDomain. The Parameters Guide contains more detailed information about this parameter. Maximum buffer size that can be negotiated with Microsoft Windows clients. To increase the default value, change param cifs.W95BufSz, cifs.NTBufSz or cifs.W2KBufSz. The Parameters Guide contains more detailed information about these parameters. This value is fixed and defines the number of requests a client is able to send to the Data Mover at the same time, (for example, a change notification request). To increase this value, change the maxMpxCount parameter. The Parameters Guide contains more detailed information about this parameter.
Compname length
63 bytes
Number of domains
10 tested 509 (theoretical max.)
Block size negotiated
64 KB
Number of simultaneous requests per CIFS session (maxMpxCount)
127
58
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Total number files/directories opened per Data Mover Number of Home Directories supported Number of Windows/UNIX users
200,000 (NS/500/700, DM514, DM 514T,NSX, NS40, NS80) 20,000 unlimited
A large number of open files could require high memory usage on the Data Mover and potentially lead to out-ofmemory issues.
Earlier versions of the Celerra Network Server relied on a basic database, nameDB, to maintain Usermapper and secmap mapping information. DBMS now replaces the basic database. This solves the inode consumption issue and provides better consistency and recoverability with the support of database transactions. It also provides better atomicity, isolation, and durability in database management. To increase the default value, change param cifs.listBlocks (default 64, max 255). The value of this parameter times 256 = max number of users. Note: TID/FID/UID shares this param and cannot be changed individually for each ID. Use caution when increasing this value as it could lead to an out-ofmemory condition. Refer to the System Parameters Guide for parameter information Number of threads for CIFS users (if there is 1 GB of memory on the Data Mover, the default is 96 threads; however, if there is over 1 GB of memory, the default number of threads is 256). This is the default, you can increase the number using the server_setup command with cifs start=xx value.
Number of users per TCP connection
16K (default), 64K (max)
Number of CIFS threads
96 (1GB memory) 256 (>1 GB memory)
EMC Celerra Network Server Version 6.0.70.4 Release Notes
59
Technical notes
Number of CIFS connections (share accessed) per TCP connection
16K (default), 64K (max)
To increase the default value, change param cifs.listBlocks (default 64, max 255). The value of this parameter times 256 = max number of connection. Note: TID/FID/UID shares this param and cannot be changed individually for each ID. Use caution when increasing this value as it could lead to an out-ofmemory condition. Refer to the System Parameters Guide for parameter information. To increase the default value, change param cifs.listBlocks (default 64, max 255). The value of this parameter times 256 = max number of files/directories opened per CIFS connection. Note: TID/FID/UID shares this param and cannot be changed individually for each ID. Use caution when increasing this value as it could lead to an out-ofmemory condition. Be sure to follow the recommendation for total number of files/directories opened per Data Mover. Refer to the System Parameters Guide for parameter information. Note: The total number of VDMs, file systems, and checkpoints across a whole cabinet cannot exceed 2048.
Number of files/directories opened per CIFS connection
16K (default), 64K (max)
Number of VDMs per Data Mover number of ntxmap User Mapping entries Table 5 FileMover Guideline/specification Max connections to secondary storage per primary (Celerra) file system Number of HTTP threads for servicing FileMover API requests per Data Mover
29
1000
Maximum tested value 1024
Comment
64
Number of threads available for recalling data from secondary storage is half the number of whichever is the lower of CIFS or NFS threads. 16 (default), can be increased using server_http command, max (tested) is 64.
60
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Table 6 File systems guidelines Guideline/specification Maximum tested value Mount point name length 255 bytes (ASCII)
Comment The "/" is used when creating the mount point and is equal to one character. If exceeded, Error 4105: Server_x:path_name: invalid path specified is returned. For nas_fs list, the name of a file system will be truncated if it is more than 19 characters. To display the full file system name, use the –info option with a file system ID (nas_fs -i id=<fsid>). With Unicode enabled in an NFS environment, the number of characters that can be stored depends on the client encoding type such as latin-1. For example: With Unicode enabled, a Japanese UTF-8 character may require three bytes.1 With Unicode enabled in a CIFS environment, the maximum number of characters is 255. For filenames shared between NFS and CIFS, CIFS allows 255 characters. NFS truncates these names when they are more than 255 bytes in UTF-8, and manages the file successfully. Note: Make sure the final path length of restored files is less than 1024 bytes. For example, if a file is backed up which originally had path name of 900 bytes, and it is restoring to a path with 400 bytes, the final path length would be 1300 bytes and would not be restored.1 This is a hard limit and is rejected on creation if over 255 limit. The limit is bytes for UNIX names, Unicode characters for CIFS. 1 This is a hard limit, code will prevent you from creating more than 65,533 directories. This max number includes VDM and checkpoint file systems.
File system name length
240 bytes (ASCII) 19 chars display for list option
Filename length
255 bytes (NFS), 255 characters (CIFS)
Pathname length
1,024 bytes
Directory name length
255 bytes.
Subdirectories (per parent directory) Number of file systems per Celerra
65,533
4096 (NSX, NS40/NS80) 2048 (All other models)
EMC Celerra Network Server Version 6.0.70.4 Release Notes
61
Technical notes
Number of file systems per Data Mover
2048
Mount operation will fail when the number of file systems reaches 2048 with an error indicating maximum number of file systems reached. This max number includes VDM and checkpoint file systems. Unified platforms — Running setup_clariion on the NS-960 platform will provision the storage to use 1 LUN per RAID group which may result in LUNS > 2 TB, depending on drive capacity and number of drives in the RAID group. On all CX4 integrated platforms (NS-120, NS480, NS-960), the 2 TB LUN limitation has been lifted--this may or may not result in LUN size larger than 2 TB depending on RAID group size and drive capacity.For all other unified platforms, setup_clariion will continue to function as in the past, breaking large RAID groups into LUNS that are < 2 TB in size. Gateway systems — For gateway systems, users may configure LUNs greater than 2 TB up to 16 TB or max size of the RAID group, which ever is less. This is supported for NS-40G, NS80G, NSX, and NS-G8 NAS gateways when attached to CX3, CX4 and Symmetrix DMX-4 backends. Multi-Path File Systems (MPFS) — MPFS supports LUNs greater than 2 TB. Windows 2000 and 32-bit Windows XP, however, cannot support large LUNs due to a Microsoft Windows OS limitation. All other MPFS Windows clients support LUN sizes greater than 2 TB if the 5.0.90.900 patch is applied. Use of these 32-bit Windows XP clients on NS-960 systems require RPQ (Request for Price Quotation) approval.
Maximum Disk Volume Size
dependent on RAID group (see comments)
62
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Total storage for a Data Mover (Fibre Channel Only)
Celerra Version 6.0 16 TB = NX4 32 TB =NS20, NS-120 64 TB =NS40, NS480, NS40G, NS-G2 128 TB = NS80, NS80G, NSX 256 TB = NS960, NS-G8 Celerra Version 5.6 8 TB = CNS (514 DM) 10 TB = NX350, NS500 16 TB = NX4, NS700 32 TB = NS20, NS120 64 TB = NS40, NS480, NS40G, NS-G2, VG2 128 TB = NS80, NS960, NS80G, NSX, NS-G8 See right column for TB information per Data Mover
These total capacity values represent Fibre Channel disk maximum with no ATA drives. Fibre Channel capacity will change if ATA is used. Note: On a per-Data-Mover basis, the total size of all file systems, and the size of all SavVols used by SnapSure must be less than the total supported capacity. Exceeding these limit can cause out of memory panic. Refer to the Celerra Capacity Limits tables for more information, including mixed disk type configurations.
File system size
16 TB (hard limit, enforced) NS700, NSX, NX4, NS120, NS480, NS960, NS-G8 10 TB NS350, NS500 8 TB 514/514T Data Mover When planning a file system, with respect to maximum desired size, be sure to factor in the time required for FSCK, ACL checks, file system build & restore. This guideline applies to both Fibre Channel and ATA drives. This hard limit is enforced and cannot be exceeded. Same as number of inodes in file system. Each 8 KB of space = 1 inode.
File size Number of directories supported per file system Number of files per Directory
16 TB (4 TB with file size policy of quota)
500,000
Exceeding this number will cause performance problems.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
63
Technical notes
Maximum number of files and directories per Celerra file system
256 Million (default)
This is the total number of files and directories that can be in a single Celerra file system. This number can be increased to 4 billion at file system creation time but should only be done after considering recovery and restore time and the total storage utilized per file. The actual maximum number in a given file system is dependent on a number of factors including the size of the file system.
1 All other industry-standard caveats, restrictions, policies, and best practices prevail. This includes, but is not limited to, FSCK times (now made faster through multi-threading), backup and restore times, number of objects per file system, snapshots, file system replication, VDM replication, iSCSI LUN replication, performance, availability, extend times, and layout policies. Proper planning and preparation should occur prior to implementing these guidelines.
Table 7 ISCSI guidelines Guideline/specification Maximum tested value LUN size Minimum: 3 MB Maximum: 2 TB less 1 MB
Comment Because an iSCSI LUN requires overhead, you cannot create a LUN size equal to the total available space; the maximum size for a LUN is slightly less (typically 1 MB less) than the amount of available space in the file system. Each target can have as many as 256 LUNs (numbered 0 through 255). Windows, however, limits each target to 255 LUNs (numbered 0 through 254). To comply with the Windows limitation, Windows - compatible mode, which is enabled by default on the Data Mover, limits the number of LUNs to 255. If you use a non-Windows iSCSI initiator, you can disable Windows--compatible mode by modifying the iscsi.WinCompat parameter. Refer to the System Parameters Guide for parameter information. Due to a volume formatting limitation on Microsoft platforms, when extending a LUN, the total LUN size after extension must be greater than 129 MB.
LUNs per target
256
LUN extension size
Minimum: 8 MB Maximum: same as LUN size
64
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Target alias
255 bytes
This name is an alias for the formal iSCSI name. The alias is an alphanumeric string of as many as 255 characters. The alias can include Unicode characters in UTF-8 encoding. Because the alias is the key field in the Celerra Network Server’s iSCSI databases, each alias must be unique on a given Data Mover.1 EMC strongly recommends you leave this field empty and let the Celerra Network Server generate a legal iSCSI name. If you type a name, be sure it is globally unique and in either IQN or EUI format. An IQN can have as many as 223 characters. The Celerra Network Server converts all uppercase characters in an IQN or EUI to lowercase characters. Only letters, numbers, "-", ".", and ":" are allowed in iSCSI IQN names. CHAP secrets must be a minimum of 12 characters and a maximum of 255 characters. Microsoft iSCSI initiator supports CHAP secrets of 12 to 16 characters. Maximum includes temporary writable snapshots.
iSCSI qualified name (IQN)
223 characters
CHAP secret
255 characters
Snapshots per LUN Replication session alias Replication application label Replication attached comment text or binary data Configured replication sessions per Data Mover Concurrent active replication sessions per Data Mover Destinations per source file system
2000 64 characters 128 characters 4096 bytes
1024
256 (theoretical max)
Number of configured iSCSI replication sessions that can be actively transferring data at one time. Exceeding this number may cause performance implications on the source file system.
4
EMC Celerra Network Server Version 6.0.70.4 Release Notes
65
Technical notes
iSCSI timeout
See comments
EMC recommends client iSCSI initiator timeout be longer than failover time. Note: MS iSCSI Initiator has a default timeout of 1 minute.
Table 8 Naming Services guidelines Guideline/specification Maximum tested value Number of DNS domains 3 - WebUI unlimited - CLI
Comment Three DNS servers per Data Mover is the limit if using WebUI. There is no limit when using the CLI (command line interface). You can configure up to 10 NIS servers in a single NIS domain on a Data Mover. Note: A Data Mover supports only one NIS domain. Each time you configure a NIS domain and specify the servers, it overwrites the previous configuration. A Data Mover can read 1004 bytes of data from a NIS record.
Number of NIS Servers per Data Mover
10
NIS record capacity Number of DNS servers per DNS domain
1004 bytes 3
Table 9 NFS guidelines Guideline/specification Maximum tested value Number of NFS exports 2,048 per Data Mover Tested Unlimited theoretical max. 64K with TCP (theoretical) Unlimited with UDP (theoretical) 16383
Comment You may notice a performance impact when managing a large number of exports using Celerra Manager.
Number of concurrent NFS clients
Limited by TCP connections.
Netgroup line size
The maximum line length that the Data Mover will accept in the local netgroup file on the Data Mover or the netgroup map in the NIS domain that the Data Mover is bound to. 2 billion max value of any GID. The maximum number of GIDs is 64K, but an individual GID can have an ID in the range of 0- 2147483648
Number of UNIX groups supported
64K
66
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Table 10 Networking guidelines Guideline/specification Maximum tested value Link aggregation/ether channel 8 ports (ether channel) 12 ports (link aggregation (LACP)
Comment Ether channel: the number of ports used must be a power of 2 (2, 4, or 8). Link aggregation: any number of ports can be used. All ports must be the same speed. Mixing different NIC types (i.e., copper and fibre) is not recommended. IEEE standard. Theoretically 509. By default the value is (in theory) 0xFFFF, but it is also limited by the number of TCP streams that can be opened. To increase the default value, change param tcp.maxStreams (set to 0x00000800 by default). If you increase it to 64K before you start TCP, you will not be able to increase the number of FTP connections. Refer to the System Parameters Guide for parameter information.
Number of VLANs supported Number of interfaces per Data Mover Number of FTP connections
4094 45 tested Theoretical value 64K
Table 11 Quotas guidelines Guideline/specification Maximum tested value Number of Tree Quotas Max size of Tree Quotas Max number of unique groups Quota path length 8191 256 TB 64K 1024
Comment Per file system Per file system
EMC Celerra Network Server Version 6.0.70.4 Release Notes
67
Technical notes
Table 12 Replicator V2 guidelines Guideline/specification Maximum tested value Number of Replication sessions per Celerra Max number of Replication sessions per Data Mover Max number of local and remote file system and VDM replication sessions per Data Mover Max number of loopback file system and VDM replication sessions per Data Mover 1365 (NSX) (other platforms): 682 1024
Comment
This enforced limit includes all configured file system, VDM, and iSCSI replication sessions as well as copy sessions
682
341
The max number of iSCSI LUN replication sessions per Data Mover is the difference between the max number of replication sessions (1024) minus the number of configured file systems and VDM replication sessions per Data Mover. Table 13 SnapSure guidelines Guideline/specification Maximum tested value Number of checkpoints per file system 96 read-only 16 writable
Comment Up to 96 checkpoints per file system are supported, as well as 16 writable checkpoints.
Table 14 Virus Checking guidelines Guideline/specification Maximum tested value Virus checker file pending 90% of available VNODES in system
Comment The percentage of total cached open files available in the system that can be pending on Virus Checking. An event is sent to the Control Station when the maximum is reached. To change the default value, modify param viruschk.vnodeMax(default=2000) and/or param viruschk.vnodeHWM(default=0x5a(90% ). Refer to the System Parameters Guide for parameter information.
On systems with Unicode enabled, a character may require between 1 and 3 bytes, depending on encoding type or character used. For example,
68
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
a Japanese character typically uses 3 bytes in UTF-8. ASCII characters require 1 byte.
Capacity guidelines Celerra capacity limits - NS series
Following is a list of the versions 6.0 Celerra Capacity limits for the NS series. The E-Lab Interoperability Navigator is updated monthly and contains the most up-to-date capacity limits.
Table 15 Celerra capacity limits – NS series
NX4
NS20
NS120
NS40
NS480
NS80 (XB-60 or XB-65)
4
NS960
Usable IP Storage Capacity limit per blade (all disk 5 types and uses) Usable MPFS Storage Capacity limit per blade (all disk types and 5 uses) Max FS Size Max # FS per 7 DM/Blade Max # FS per 7 cabinet Max configured replication sessions per DM/Blade for 3 Replicator v2 Max # of checkpoints per 2 PFS Max # of NDMP sessions per DM/Blade
16TB
4
32TB
4,6
32TB
6
64TB
4
64TB
128TB
256TB
N/A
128TB
1
128TB
1
128TB
1
128TB
1
128TB
1
256TB
1
16TB 2048 4096
16TB 2048 4096
16TB 2048 4096
16TB 2048 4096
16TB 2048 4096
16TB 2048 4096
16TB 2048 4096
1024
1024
1024
1024
1024
1024
1024
96
96
96
96
96
96
96
4
4
4
4
4
4
4
EMC Celerra Network Server Version 6.0.70.4 Release Notes
69
Technical notes
Table 16 Celerra capacity limits – NS series (continued)
NS40 G
NS-G2
NS80 (XB-60 or XB-65)
4
NSX (XB-60 or XB-65)
4
NS-G8
V-G2
VG8
Usable IP Storage Capacity limit per blade (all disk 5 types and uses) Usable MPFS Storage Capacity limit per blade (all disk types and 5 uses) Max FS Size
64TB
4
64TB
4
32TB
128TB
128TB
64TB
256T B
128TB
1
128TB
1
128TB
1
128TB
1
128TB
128TB
256T B
16TB
16TB
16TB
16TB
16TB
16TB
16T B 2048 4096
Max # FS per 7 DM/Blade Max # FS per 7 cabinet Max configured replication sessions per DM/Blade for 3 Replicator v2 Max # of checkpoints per 2 PFS Max # of NDMP sessions per DM/Blade
2048 4096
2048 4096
2048 4096
2048 4096
2048 4096
2048 4096
1024
1024
1024
1024
1024
1024
1024
96
96
96
96
96
96
96
4
4
4
4
4
4
4
Notes: 1. 2. 3. No snaps or replication –If planning to use snaps or replication, the Data Mover capacity limit will be the same as non-MPFS. PFS is Production File System. A maximum of 256 concurrently transferring sessions are supported per DM.
70
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
4. 5.
When using Replicator –To achieve storage capacities listed, Replicator v2 is required. This is the usable IP storage capacity per blade. For overall platform capacity, consider the type and size of disk drive and the usable Fibre Channel host capacity requirements, RAID group options and the total number of disks supported. NS20, NS120 – All ATA configuration can go up to 48 TB per blade. This count includes production file systems, user defined file system checkpoints, and two checkpoints for each replication session.
6. 7.
Celerra virus checking solution
The following table lists the requirements for using CAVA with thirdparty antivirus engines.
AV Vendor Celerra Version CAVA Version
CA
eTrust r8.1 eTrust r8 TrendMicro ServerProtect for EMC NAS 5.3.1 ServerProtect for EMC NAS 5.5.8 Sophos AntiVirus 3.x All currently-supported DART versions1 1.8.9; 2.2.4; 3.4.0; 3.5.6; 3.5.7.1, 3.5.8, 3.6.2 All currently-supported DART versions1 1.8.9; 2.2.4; 3.4.0; 3.5.6; 3.5.7.1, 3.5.8, 3.6.2 All currently-supported DART versions1 1.8.9; 2.2.4; 3.4.0; 3.5.6; 3.5.7.1, 3.5.8, 3.6.2
AntiVirus 5.x
AntiVirus 6.x AntiVirus 7.x AntiVirus 9.x Kaspersky Kaspersky Anti-Virus for Windows Servers Enterprise Edition McAfee VirusScan 8.0i All currently-supported 1.8.9; 2.2.4; 3.4.0; 3.5.6; 3.5.7.1, 3.5.8, 3.6.2 5.6.46.4 and later Celerra Event Enabler (CEE) version 4.5.1
EMC Celerra Network Server Version 6.0.70.4 Release Notes
71
Technical notes
VirusScan 8.5i Patch 2 and above Symantec Symantec Endpoint Protection 11.04 SAV for NAS 5.1.x SAV for NAS 4.3.x
DART versions1
5.6.48.x and later All currently-supported DART versions1
4.5.2.2 and later 3.6.2 3.5.7.1, 3.5.8, 3.6.2
1Currently supported NAS versions include the entire 5.6.x and 6.0.x TM families. Refer to the EMC E-Lab Interoperability Navigator for the latest supported version information.
Celerra scripting guidelines
Following are guidelines to keep in mind when creating Celerra command scripts.
Issue Scheduling Action NAS Database Backups: The Celerra Network Server backs up the NAS database that stores specific configuration information required for each Data Mover every hour, at one minute after the hour. During part of the backup, the database is locked, and some commands that rely on the database might not have access. EMC recommends that command scripts avoid starting at one minute after the hour. Note that scripts with complex commands that run for an extended period may overlap the backup period. The duration of the backup may vary. Use the following Linux command to check the state of the backup process prior to executing scripts: ps -ef | grep nasdb_backup. If a lock condition occurs, wait a few minutes and retry. Some commands must lock the database in order to execute. If multiple user-entered commands or scripts are active at the same time, some of these commands may lock the database and prevent other commands from executing. To avoid this, you should sequence commands whenever possible. Some processes within a script can take time to complete. Use proper timing and adequate sleep statements to prevent timingrelated issues. Piping script outputs through grep is a helpful tool to check the status of the script. Use periodic checks to grep for file or database locked messages, timeouts, resource unavailable warnings, and other failure or success messages, and use this information to check status, pause the script, or halt it. All commands return a UNIX-style return code (for example: 0 for success or 1 for failure) or a text-based status code (for
Command sequencing
Sleep statements
Pipe and grep
Return code check
72
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Issue
Action example, done) that can be used to help determine if the command completed or if there was an error or a conflict with the NAS database backup, or other commands being run. If a lock condition occurs, wait a few minutes and retry. If you create and run scripts, be sure to incorporate return code checks and verify for proper return codes from individual operations.
Scripting examples: The RECOVERABLE variable contains the following errors to retry on: Unable to acquire lock Resource temporarily unavailable Unable to connect to host Socket: All ports in use Database resource vanished Connection timed out NAS_DB locked object is stale
RECOVERABLE="unable to acquire lock|Resource temporarily unavailable|unable to connect to host|socket: All ports in use|database resource vanished|Connection timed out|NAS_DB locked object is stale".
Following is an example of what the RECOVERABLE variable looks like:
The res variable contains the command output.
#!/bin/sh ######################################################## # File: nas_cmdrcvr.sh # Created by: NAS Engineering # Date: Thursday, May 25, 2006 # Version: 1.0 # Notes: # 1) script will retry commands for specified period of time #2) script will log messages to file only if there's available disk space ######################################################## NAS_DB=/nas export NAS_DB PATH=$PATH:$NAS_DB/bin:$NAS_DB/sbin:/usr/sbin:.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
73
Technical notes
export PATH RETRIES=60 SLEEPTIME=30 LOGDIR="/home/nasadmin" LOGFILE="$0.log" LOGLAST="$0.last" DISKFULL=98 RECOVERABLE="Resource temporarily unavailable|\ unable to acquire lock|\ unable to connect to host|\ socket: All ports in use|\ database resource vanished|\ Connection timed out|\ NAS_DB locked object is stale" # # function to log messages to a file # nas_log() { DISKCHK=`df -k $LOGDIR|awk 'NR>1{print $5}'|sed 's/\%//'` # if there's enough free disk space, append to log if [ $DISKCHK -lt $DISKFULL ]; then TDSTAMP=`date +%Y-%m-%d' '%T` echo $TDSTAMP: $LOGMSG >> $LOGDIR/$LOGFILE fi # regardless of available space, always write last error echo $TDSTAMP: $LOGMSG > $LOGDIR/$LOGLAST } # # function to execute (and potentially retry) commands # nas_cmd() { # initialize variable(s) retry_count=0
EMC Celerra Network Server Version 6.0.70.4 Release Notes
74
Technical notes
# loop until either successful of retry count exceeded while [ $retry_count -le $RETRIES ]; do # execute command and gather response RES=`$CMD 2>&1` # check if response means command is recoverable if [ `echo "$RES"|egrep -c "$RECOVERABLE"` -ne 0 ]; then # check retry count if [ $retry_count -ne $RETRIES ]; then # retry count has not been exceeded LOGMSG="Command ($CMD) failed with ($RES)...retrying in $SLEEPTIME s" nas_log sleep $SLEEPTIME else # retry count has been exceeded LOGMSG="Command ($CMD) failed with ($RES)...exiting (retry count of $RETRIES exceeded)" nas_log exit 1 fi else # command was either successful or failed for an unknown reason LOGMSG="Command ($CMD) successful with ($RES)" nas_log retry_count=$RETRIES exit 0 fi #increment counter for retries retry_count=`expr $retry_count + 1` done } # # main
EMC Celerra Network Server Version 6.0.70.4 Release Notes
75
Technical notes
# CMD="nas_volume -d mtv1" nas_cmd
Using the NAS database and query facility EMC has partially changed the layout or format of Celerra internal databases. This change can impact the use of awk or grep utilities when used in scripts that assume specific positions of fields in databases. To enable searching of the NAS database, Celerra has developed a new query subsystem that appears as a hidden option on some of the nas_commands. This query subsystem enables you to specify the information you are interested in, allows you to format the output, and is independent of the database format. Caution Do not use grep and awk to scan the database files. Database positions may change and substrings may return false matches for database objects. Following is an example of a query to view unused disks: nas_disk -query:inuse==n -format:'%s\n' -Fields:Id To filter root disks, refer to the List all non-root disks that are not in use in Table 10. Examples Use the following commands to view the tags (fields) that you can query: nas_disk –query:tags nas_fs –query:tags nas_volume –query:tags nas_slice -query:tags Table 17 contains a list of examples to help you get started. Note that these commands can be run on the Control Station CLI, so the hardcoded values can be replaced with shell script variables.
Table 17 Task and query examples Task Query the ID of a named file system Example
nas_fs -query:Name==RLL_fs10 format:'%s\n'-Fields:Id
76
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Query the ID of a named file system without the new line Query the name of a file system ID that corresponds to a particular ID List of all server ID List of all server names List of all the checkpoint file systems List type of file system with ID 20
nas_fs -query:Name==RLL_fs10 format:'%s'-Fields:Id nas_fs -query:id==20 -format:'%s\n' -Fields:Name nas_server -query:* -format:'%s\n' Fields:Id nas_server -query:* -format:'%s\n' Fields:Name nas_fs -query:type==ckpt fields:name -format:"%s\n" nas_fs -query:id==20 -format:'%s\n' -Fields:Type nas_fs -query:inuse==y format:'%s\n' -Fields:Name or nas_fs -query:inuse==y format:'%s\n' -Fields:Id nas_fs -query:id==28 -format:'%s' Fields:BackupOf nas_server -query:id==2 format:'%s\n' -fields:name`" nas_fs -query:Name==my_fs -format:%d -fields:VolumeID nas_volume -query:Name==my_meta3 format:%d -fields:Blocks nas_volume -query:Name==JAH_meta3 format:%d -fields:BlockSize nas_fs -query:name==fs123 format:'%s\n'-fields:ServersNumeric nas_disk query:inuse==n:IsRoot==False -format:"%s\n" -fields:name nas_volume query:inuse==n:IsRoot==False:name=dc -format:"%s\n" -fields:name nas_disk -query:inuse==n:SymmID==$ symm_id:IsRoot==False -format:"%s\n" -fields:name
List the file systems that are in use
Identify file system of which file system ID 28 is a backup List the name of the server with ID 2
View which volume file system is built on
View the block count of meta volume
View the block size of meta volume
Find which server IDs use fs123
List all non-root disks that are not in use
List unused volumes that contain “dc” in the volume name List all available disks on a particular storage device (symm_id is a script/env variable)
EMC Celerra Network Server Version 6.0.70.4 Release Notes
77
Environment and system requirements
Query operators Table 18 lists the operators to use when building your queries:
Table 18 Operator definitions
Operator = == ==+ =* =^ =^= =^=^+ =^*
Definitions
Having the pattern (sub-string match) Exact string match Integer Minimum (not less than) Integer Maximum (not more than) Any NOT having the pattern NOT an exact match NOT Integer Minimum (is less than) NOT Integer Maximum (is more than) NOT Any (None)
Environment and system requirements
Symmetrix microcode and CLARiiON FLARE levels
The E-Lab Interoperability Navigator contains the latest microcode and FLARE information. The E-Lab Interoperability Navigator is updated monthly and can be found on EMC Online Support at http://Support.EMC.com.
Centera FileArchiver not supported
The Centera® FileArchiver policy engine is not supported when using Celerra Network Server version 6.0.36.4 or later.
Java Runtime Environment (JRE)
Minimum JRE 1.6. After February 2013, Oracle will no longer post updates of Java SE 6 to its public download sites. Existing Java SE 6 downloads already posted as of February 2013 will remain accessible in the Java Archive on Oracle Technology Network. This Unisphere Client will run with
78
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Documentation
JRE 1.7 but it has not yet been fully qualified with JRE 1.7. Future releases of Unisphere will be qualified with JRE 1.7 and the future Unisphere Client can be used as a workaround if there are issues specific to JRE 1.7.
Documentation
Celerra Network Server Version 6.0 documentation
EMC provides the ability to create step-by-step planning, installation, and maintenance instructions tailored to your environment. To create VNX customized documentation, go to: https://mydocs.emc.com/VNX.
Documentation updates
CDMS The Celerra Data Migration Service documentation leads the user through a multi-step process to execute a migration. The latest version of this document highlights the success and failure criteria for various steps in this process. Celerra CDMS Version 2.0 for NFS and CIFS provides more information about CDMS. The following have also been added as warnings and cautions: A verify command should not be run if there are any migration threads running or active on other CDMS connections in the same file system. A verify command should not be run if there is another verify command still in progress on a different file system on the same Data Mover.
Multibyte support for Celerra data elements In the Celerra management interfaces, multibyte support allows to input, store, and display multibyte characters. Users can input and view Celerra object fields in their native language. The multibyte feature is supported since Celerra NAS 5.6 and later. The following data elements are enabled with multibyte support and also support the Unicode 3.0 standard: CIFS Share name
EMC Celerra Network Server Version 6.0.70.4 Release Notes
79
Documentation
CIFS share comment CIFS export path Tree quota pass Tree quota comment Quota user name CIFS server netbios name CIFS server computer name CIFS server alias CIFS server comment CIFS domain name CIFS workgroup Organizational Unit (OU) name DFS root name CDMS source share name CDMS local path CDMS source export path Mount point name File system name Checkpoint name NFS export name NFS export path DFS root name CDMS source share name CDMS local path CDMS source export path Mount point name File system name Checkpoint name NFS export name NFS export path
80
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Installation and upgrades
Installation and upgrades
Celerra installation process
This section lists some facts you should know about the Celerra Network Server and the installation and configuration process before proceeding further. How to prepare for Celerra installation To prepare for Celerra installation and configuration, complete the following tasks: 1. Perform the pre-site installation tasks as required by EMC to complete the installation and connectivity of the Celerra Network Server. Tasks include preparing electrical facilities, wiring the network, adding a telephone line, checking environmental conditions and related requirements, as stated in the Professional Services Statement of Work (PSSOW).
Note: For a copy of the PS-SOW document, contact your Professional Services Account Team. For examples of Celerra network topologies, refer to the Configuring and Managing Celerra Networking technical module.
2.
Configure your environment for UNIX clients, Microsoft Windows clients, or a combination of Windows and UNIX clients, as applicable. Complete network assessment to ensure proper performance and protocol configurations. (Optional EMC services can be ordered to assist with this process.)
3.
Celerra Network Server installation and configuration EMC Customer Service installs the Celerra Network Server hardware—a process that includes the physical setup, cabling, power application and diagnostic testing of the Celerra Network Server. Next, EMC Professional Services or an EMC partner might configure and integrate the Celerra Network Server. For example, for a Celerra Network Server with a Symmetrix system, this process typically includes the following tasks: Coordinating the activities to enable installation, cabling, and connectivity to system cabinets Loading and configuring the Celerra Network Server software
EMC Celerra Network Server Version 6.0.70.4 Release Notes 81
Installation and upgrades
Creating, mounting, and exporting file systems Configuring network protocols Integrating with a supported, existing user authentication environment Configuring standby Data Mover(s), if applicable Configuring and demonstrating system management and administration tools
EMC Professional Services can perform additional Celerra Network Server configuration tasks, such as host provisioning (with Celerra MPFS), backup protection, and network connectivity, depending on the individual EMC customer agreement defined in the PSSOW. These services are part of the EMC Network-Attached Storage Design and Implementation service offering, which is designed to provide continuous technical expertise as you manage your organization’s growing network storage needs. For more information about this service offering, contact your EMC Customer Support representative. After EMC installs, configures, and tests the Celerra Network Server, it is fully operational and ready to support networked storage operations.
Celerra upgrades
You can upgrade the Celerra NAS software two ways: Celerra Upgrade Tool Command Line Interface
Celerra Upgrade Tool You can upgrade the software for many Celerra integrated systems by using the Celerra Upgrade Tool (CUT), which is part of the Unisphere Service Manager (USM). CUT lets you upgrade Celerra systems without assistance from EMC or Customer Support by communicating over the Internet to EMC Online Support, and over a local network to a Celerra system. Three wizards guide you, step-by-step, through the process of upgrading the Celerra software, including updating the Control Station and the blades, but not the storage system: Go to EMC Online Support at http://Support.EMC.com for information about USM and CUT and for information about downloading a copy of USM to a Windows system. The Upgrading Unisphere for File online help on USM provides information and instructions about using CUT and its wizards. The Celerra Upgrade Tool can upgrade Celerra Network Servers running software versions:
82 EMC Celerra Network Server Version 6.0.70.4 Release Notes
Installation and upgrades
• •
5.6 to 6.0 6.0 to 6.0.x or later
The Celerra Upgrade Tool cannot upgrade systems with dual Control Stations, nor can it be used to upgrade Gateway systems. If your Celerra is not eligible for upgrade by means of CUT, you will need to use the CLI to upgrade your system. Running multiple upgrade wizards The USM Celerra Upgrade Tool wizard does not explicitly prevent CLARiiON upgrade wizards from running while it is in use. The Celerra Upgrade Tool presents a prominent warning on its 'Welcome' page in the GUI. Note that this scenario is the same as simultaneously running multiple instances of USM and/or CLI against the systems, which cannot currently be prevented. Also, the CLARiiON wizards correctly prohibit the Celerra Upgrade Tool from launching while they are in use. In general, upgrading the Celerra NAS and the CLARiiON FLARE software at the same time can cause serious system problems and result in system downtime. Other CUT restrictions: Selecting and rebooting blades at different times can cause errors. Select and reboot all blades at same time during the upgrade in the Reboot Blades dialog. The Celerra Upgrade Tool cannot upgrade systems with dual Control Stations.
NAS and FLARE upgrades EMC requires NAS upgrades to be performed before FLARE is upgraded on any attached CLARiiON. The reason for this is to maintain compatibility between the storage API (NAVI) on the Control Station and the FLARE on the CLARiiON. The API is built in as part of the NAS management software and is used to gather data about the backend such as statistics, health, and configuration. If the FLARE is upgraded first, the API may encounter issues while communicating with the new FLARE. Upgrading the NAS first allows the API to have limited backward compatibility with the FLARE code. Out-of-family upgrades The following issues have been reported during out-of-family upgrades.
EMC Celerra Network Server Version 6.0.70.4 Release Notes 83
Installation and upgrades
Error message during upgrade While upgrading from version 5.6 to 6.0, the following error message is displayed: Error - cannot open /usr/sbin/ext2online /dev/hda1. This error can be ignored, as it does not stop the upgrade and the task completes successfully. Pre-Upgrade Health Check (PUHC) fails When attempting to upgrade the software via CLI from version 5.5 to 6.0, PUHC fails abruptly with a run-time failure message which does not explain the cause of the failure. Note: A 5.5 to 6.0 upgrade is not supported. Upgrade Replication task While performing an out-of-family upgrade (5.6.x to 6.0.x) on a dual Control Station system, the Upgrade Replication task exits with the message: =======================Tasks======================= 23:06 [ 3/80 ] Upgrade replication 0 seconds Please shutdown control station in slot 1 and retry When this occurs, users need to power off the secondary Control Station, just running 'shutdown -h' is not enough. The secondary Control Station must be in 'power-off' mode during the upgrade. Reboot required at upgrade CNS software The CNS software is comprised of two installation components, a Linux package and a CNS software package. The software currently running on the Celerra and the CNS upgrade software versions determine the number of Control Station and Data Mover reboots required and whether Data Mover reboots can be scheduled to a later time after the upgrade is complete. The DART software runs on the Data Movers and is part of the CNS software package. The new version of DART is loaded and run on the Data Movers only after a Data Mover boot. Control Station rebooting Most software upgrades require a Control Station reboot. When upgrading CNS software, if either of the first two digits of the new version number change (for example, 5.6.x to 6.0.x), the Control Station requires rebooting.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
84
Installation and upgrades
At the beginning of upgrade, the system information section lists whether or not the Control Station reboot is mandatory. Note that this initial output is long and may scroll off the screen. You will be prompted whether to continue the upgrade after all the upgrade information is displayed. If you enter no, the upgrade stops, and you are returned to the command prompt. ===========System Information=========== Install Manager Version: 6.0.34-4 Install Manager Command: /celerra/upgrade/pkg_6.0.34.4/install_mgr Starts at: Thu Jun 3 10:36:56 2010 From version: 5.6.49-1 To: 6.0.34-4 Dual Control Stations: yes Slot: slot_0 Blade slot(s): 2, 3, 4, 5 Cabinet family: NS Cabinet type: NS-HAMMERHEAD Control Station reboot: multiple (mandatory) Blade reboot: 1 time(mandatory) Total number of attempts: 0 ========================================= . .< other information is posted here> . Do you wish to continue with the upgrade now [yes or no]? > yes Data Mover rebooting The Data Movers require a reboot in order to run the new version of DART. When upgrading CNS software, if either of the first two digits of the new version number change (i.e., 5.6.x to 6.0.x), the Data Mover requires rebooting.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
85
Installation and upgrades
The system information screen is the same as the information in the Control Station rebooting screen. Limited scheduling of the Data Mover reboot exists. An estimated time for when the Data Mover reboot will happen is shown at the beginning of upgrade: ==============Estimated Time for Services ===============
Current Time: Estimated Time when NAS service will be stopped: Estimated Time when Data Movers will be reset: Estimated Time when NAS service will be restarted: Estimated Time when upgrade will be complete: 10:36 10:38 11:28 11:39 11:50
================================================== The Upgrade Data Movers task prompts before it reboots the Standby Data Movers. It then provides a menu to reboot the Primary Data Movers similar to the following: ==================================================
11:58 [ 46/70 ] Upgrade Data Movers 7 minutes Start at: Thu Jun 3 11:58:38 2010 All servers must now be rebooted. Ready to reboot the Standby Data Mover (3). Press the ENTER key to begin... The following server(s) need to be rebooted to run the new release: 2) server_2 Enter slot number(s) or ALL to select server(s) for reboot:
================================================== When upgrading DART, if both of the first two digits of the new version of software remain the same (for example, 6.0..x to 6.0.y), you may delay the rebooting of the Data Movers. The system information screen is the same as the information in the Control Station rebooting example except for the following: Data Mover reboot: 1 time (advisory) The Upgrade Data Movers task asks if you wish to reboot the Data Movers. If you enter yes, you will be prompted to press ENTER to reboot the standby Data Movers, then prompted to reboot the primary Data
86
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Installation and upgrades
Movers. If you enter no, the Data Movers continue running the current (pre-upgrade) version of CNS software until they are rebooted. =================================================
12:29 [ 44/67 ] Upgrade Data Movers 7 minutes All servers need to be rebooted to run the new release. Do you wish to reboot them now? [yes or no]: Ready to reboot the Standby Data Mover (3). Press the ENTER key to begin... The following server(s) need to be rebooted to run the new release: 2 ) server_2 Enter slot number(s) or ALL to select server(s) for reboot: Waiting for ping response from server_2... :up (0 secs)
================================================= Schedules during upgrades As part of the upgrade procedure, schedules that run between midnight and 1:00 A.M. may lose those run times. This affects only schedules that run every X hours of the day, (X being from 1 through 12), and not schedules that run at specific hours of the day. This is a known problem and can be corrected by modifying the affected schedules after the upgrade. Upgrade requirements Existing read/write-able file system must have 128K minimum free space prior to upgrading to CNS version 6.0.
CLARiiON upgrades
Occasionally, CLARiiON storage connected to a Celerra Network Server requires a software upgrade. When upgrading CLARiiON software, do not run heavy loads on the Celerra, as operations may be extremely slow or interrupted. Notes CLARiiON array software cannot be upgraded unless the Celerra Control Station software is upgraded first. Ignoring this rule may make the CLARiiON unmanageable by the Celerra. Incremental CLARiiON arrays cannot be added to systems running
EMC Celerra Network Server Version 6.0.70.4 Release Notes 87
Fixed problems
older Celerra software unless they are running a compatible release of CLARiiON software. A warning will appear when you upgrade the Celerra software if the CLARiiON software needs to be updated. If you receive this warning, finish upgrading the Celerra software and then upgrade the CLARiiON software as soon as possible to avoid compatibility issues.
Consult the E-LabTM Interoperability Navigator at http://Support.EMC.com for all software interdependencies.
Fixed problems
Fixed problems categories
The Fixed problems section is grouped by the following categories:
Category CEE Celerra Management Applications CIFS Control Station CORE Deduplication FileMover FLR Hardware MPFS Install, Config, and Upgrades Internationalization NDMP/Backup Networking NFS Replication Security Meaning Celerra Event Enabler, CAVA Unisphere, Celerra Startup Assistant, MMC snap-ins, and so on. Common Internet File System (Quotas, Usermapper, ans so on) Celerra Control Station Celerra Data Movers Deduplication Celerra FileMover File-Level Retention Hardware Multi-Path File System Celerra installs, callhomes, configuration, upgrades Internationalization (I18N) Network Data Management Protocol or Backup Networking issues Network File System Celerra Replicator, SnapSure, Checkpoint LDAP
88
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.70.4
SRDF Storage Tools UFS
Symmetrix Remote Data Facility Storage Storage EMCOPY, EMCACL, LGDUP, SHAREDUP UNIX File System
Version 6.0.70.4
Backup Severity Symptom description
Data Mover panic Page Fault Interrupt caused by sending NDMP an empty software set file filter or directory filter string.
Fix summary
The code has been modified to fix the filter logic for the empty file filter string or the directory filter string.
Tracking number
1
496542 / 48722540
Celerra Management Applications Severity Symptom description Fix summary
The code has been modified to resolve the issue by optimizing the Tree Quota query processing on the XML API server. The code has been modified to avoid collecting the wasted space statistics. The former behavior is still available by running acl database printstats slow.
Tracking number
2
High CPU and memory usage by the XML API server process when running the VNX Monitoring and Reporting Tool. All CS operations were impacted.
455143 / 43142954
1
On file systems with a very large number of acls, running the acl database printstats command might take a very long time.
467244 / 45564250
EMC Celerra Network Server Version 6.0.70.4 Release Notes
89
Version 6.0.70.4
CIFS Severity Symptom description
Could not immediately connect to share on the system when many users were connecting and disconnecting simultaneously. Sometimes could not connect to share when the system was overloaded by share connections.
Fix summary
Tracking number
2
The code has been modified to improve the connection logic. The length of the encrypted ticket component is now checked before allocating memory, and an error will be returned if the length is zero. Also, the share mount request fails, and the error is logged in the server log by the calling code. The single sparse FT log file has been replaced by two regular files which will be written to one by one. When the first is full, it will be renamed with a bak extension and the second one will be created. The code has been changed to not trigger the scan if a user closes a file opened as read-only, after another user has previously opened and modified it. The code has been modified to move a joined compname tranparently accross VDMs of the same Data Mover.
495620 / 48114692
1
Data Mover panic caused by receiving an invalid Kerberos service ticket.
505036 / 49707766
1
When all Common Event Publishing Agent servers were offline, the CIFS access could be lost briefly if the faulttolerance log file became full.
512489 / 50330376
2
Poor CIFS performance leading to DU could be caused by unnecessary virus scans after a file was opened as readonly.
519560 / 50672554
1
Issues were encountered after moving a compname to another VDM due to the Active Directory replication problem, especially for sites with multiple Domain Controllers.
522085 / 51136100
90
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.70.4
CIFS Virus Check Severity Symptom description
When the AV Engine was not available, the following message was logged: Warning: RecvStream failed I/O error.
Fix summary
The code has been modifed to suppress alert messages that occur while killing a MSRPC client connection. The code has been modified to disable CEPP events for users who have the CEPP bypass right, and also to enable the CEPP event bypass right for users with this right who are not on a CAVA/CEPP server or connected on CHECK$. The code has been modified to recognize this state and handle it correctly.
Tracking number
499970 / 48272012
2
2
CEPP events were generated even if the user had the CEPP bypass right. CAVA/CEPA users with "bypass" rights can be suppressed.
504473
2
When a file system was unmounted with files waiting to be scanned by a virus checker, the files were not scanned after the remount of the file system on another path.
506486 / 49683544
Control Station Severity Symptom description Fix summary
The code has been modified to mount NDMP checkpoints of a VDM root file system to the VDM server. The code has been modifed to send the fsck request to the parent Data Mover.
Tracking number
2
The checkpoint creation on the VDM root file system could fail during the VBB backup.
476488 / 46629630
2
The Data Mover could fail to start fsck on a writeable checkpoint mounted on an unloaded VDM.
502351 / 49425182
EMC Celerra Network Server Version 6.0.70.4 Release Notes
91
Version 6.0.70.4
2
A CIFS server created from GUI could fail to join the domain due to a leading dash in the administrator's password. NDMP backup failed for the production file system mounted with the mount point that contained two root vdm prefixes, for example, /root_vdm_1/root_vdm_1/productin_fs. The SNMP trap did not get sent for a file system notification.
The code has been modified to correctly handle such a password. The code has been modifed to avoid this problem. The code has been modified to avoid this problem.
503070 / 47564604
2
525208 / 51439440
2
487350 / 47698828
DART Core Severity Symptom description Fix summary
The code has been modified to better handle the case when clock interrupts stop unexpectedly and to allow more time for the system to recover.
Tracking number
1
An idle Data Mover might panic with SYSTEM WATCHDOG or System clock was stalled when there was no explanation for why this happened.
500581 / 49117864
EE Escalations Severity Symptom description
If the Control Station couldn’t resolve the IP address of ftp.emc.com, the following error could appear in the system log: transfer_support_materials[241 18]: The transfer script has failed. It cannot transfer data to ftp.emc.com or the remote directory. Please contact your authorized service provider to make sure the remote server has sufficient free disk space (100MB). The code has been modified to use the new IP address for ftp.emc.com as the default IP address for transferring support materials.
Fix summary
Tracking number
2
520928 / 48169232
92
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.70.4
FLR Severity Symptom description
Data Mover panicked due to the use of a NULL pointer when a file system that is mounted read-only attempted to log something to the flrLogHandler.
Fix summary
The code has been modified to check for read-only file systems before they attempt to use the flrLogHandler.
Tracking number
505349 / 49743518
1
IP Replication v2 Severity Symptom description
The memory leak in the Replication V2 error path was caused by the improper manual cleanup of a session. This could result in an out of memory panic.
Fix summary
The code has been modified to remove the memory leak that occurred during some error recovery operations.
Tracking number
494281 / 48485034
1
NFSv4 Severity Symptom description
The NFSV4 access might take a long time if multiple clients work on the same files. The delegation recalling process led to poor performance, especially when some clients did not reply to delegation recalls immediately. Page Fault Interrupt. Virt ADDRESS: 0x62f870 Err code: 0 Target addr: 0x10
Fix summary
The code has been modified to enhance the locking mechanism, allowing multiple requests to run in parallel. The code has been modified to synchronize the release of a lock owner with other lock owner operations. The code has been modified to avoid a deadlock related to the NFSv4 delegation recall and conflict checking.
Tracking number
2
490956 / 48099274
1
492876 / 48323880
1
The NFS access was lost during backup.
496917 / 48792858
EMC Celerra Network Server Version 6.0.70.4 Release Notes
93
Version 6.0.70.4
1
When using NFSv4, Data Mover panicked with a dump header similar to: DART panic/fault message: *** GP exception. Virt ADDRESS: 0x00006af142. Err code: 0 *** The NFSv4 service was not responding; "Pool:NFSD_Exec BLOCKED" appeared in the datamover log. The NFSv4 response was slow with the LDAP resolver.
The code was modified so that the NFSv4_setAttr could write, exceeding the mblk limit, in this rare condition. The code has been modified to resolve the dead-lock issue. The code has been modified to no longer query group members.
502522 / 49446140
1
503276 525643 / 50177872
2
Quotas
Severity Symptom description When the user renamed a directory on which online quota check was running, a race condition could occur which caused threads to be blocked and file system to be inaccessible. The system would panic if users tried to enable quota on an MGFS filesystem. Fix summary Tracking number
1
The code has been modified to eliminate this race condition. The code has been modified to disallow the quota feature on MGFS to avoid the panic.
509083 / 50156424
1
519520 / 50889786
RDE - PE
Severity Symptom description The state of deduplication showed 0% saved data even though the files had been deduplicated. This occured when a deduplication scan was not completed due to a file system or a checkpoint resource error. The hash table mechanism used by the Policy Engine to track hard links could cause memory fragmentation issues after the table was deleted.This could lead to an out of memory panic with a header like: PANIC in file: ../bufpool.cxx at line: 1309 : couldn't get a free page or PANIC in file: Fix summary The code has been modified to allow the FRDE PE to continue scanning the file system after encountering a resource error. The code has been modified to create a private hash table using a memory mechanism that avoids the risk of fragmentation. Tracking number
2
372977 / 35129780
1
499671 / 49145068
94
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.70.4
../addrspac.cxx at line: 503 : Out of memory The code has been modified to handle dedupplication of empty files in a manner that does not lead to poor performance.
1
The Data Mover became inaccessible after the EM copy was done. Need to reboot the Data Mover to get access back.
512668 / 50460864
Transport Protocols
Severity Symptom description Under rare circumstances, Data Mover might panic with the following stack: gdb) bt #0 PANIC (string=0x1918be0 ">>PANIC in file: ../malloc.cxx at line: 662 : \n Memory corrupted") at ../panic.cxx:277 #1 0x0000000000243e60 in free (base=<value optimized out>) at ../malloc.cxx:662 #2 0x00000000002440e2 in operator delete (p=0x500) at ../_delete.cxx:23 #3 0x0000000000485def in CLIENT::operator delete (p=0x500) at ../client.cxx:522 Fix summary Tracking number
1
The code has been modified to remove the real-time race condition by adding synchronization in deleting paths.
473726 / 46291616
UFS
Severity Symptom description The Data Mover used the netcap command to capture the network traffic of a network device and store the network packets to an internal file. When the file was full, a staging file would be used. If the staging file was not successfully opened, the Data Mover would panic. Fix summary Tracking number
1
The code has been modified to check the file handle of staging buffer before it is used.
511398 / 50349810
EMC Celerra Network Server Version 6.0.70.4 Release Notes
95
Version 6.0.70.4
1
When the Data Mover was enumerating file system checkpoints from a Windows client, it panicked with the following DART panic/fault message: Page Fault Interrupt. Virt ADDRESS: 0x000191688f Err code: 0 Target addr: 0x0000000008 **. This happened when at least two checkpoints for the file system had different time stamp for Month field while having the same stamp of Seconds, Minutes, Hours and Days.
The code has been modified to correct an invalid boundary check. This resolves the issue of the incorrect return of a NULL pointer for a valid date range.
523621 / 51266466
UFS Log
Severity Symptom description Fix summary The code has been modified to increase the size of the array, to sense and handle the overflow, and to track current users of the array elements to enhance diagnosability. Tracking number
1
Overflow of an array in a filesystem inmemory data structure led to a deadlock and hang condition.
522045 / 50805196
Backend Features
Severity Symptom description "nas_rdf -restore" failed with storage profile mismatch error if the file system was created from a storage pool. Fix summary The issue is resolved by refreshing storage profile name reference for rdf restore. Tracking number
1
531824
System Management
Severity Symptom description Hitachi EFD was showed as STD disk in the gateway system. Fix summary This issue is fixed by adding the code to correctly recognize the drive type from Hitachi. Tracking number 49746724/ 508000
2
96
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.70.4
VBB backup of VDM root file system failed. 2
This issue has been fixed by mounting NDMP checkpoints of VDM root file system to VDM server instead of server_2.
46629630/ 476488
1
XML API server processing cost high CPU and memory usage when running VNX Monitoring and Reporting Tool. All CS operations were impacted .
The issue has been resolved by optimizing the Tree Quota query processing on XML API server.
43142954/ 455143 51087704/ 521751 50912918/ 522374
SNMP trap did not get sent for a filesystem notification. 2
The issue has been resolved by improving the code responsible for configuration of SNMP traps. The issue has been resolved by sending fsck requests to the parent Data Mover of the unloaded VDM.
47698828/ 487350
It failed to start fsck on a writeable checkpoint mounted on an unloaded VDM. 2
49425182/ 502351
3
CIFS server created from GUI failed to join the domain due to a leading dash in administrator's password.
The issue has been resolved by adding a transient internal password prefix. The issue has been resolved by removing the code that strips off the root VDM prefixes from the mountpoint.
47564604/ 503070
1
NDMP backup failed for production filesystem mounted with the mount point containing two root VDM prefixes, for example: /root_vdm_1/root_vdm_1/producti n_fs.
51439440/ 525208
EMC Celerra Network Server Version 6.0.70.4 Release Notes
97
Version 6.0.65.2
NDMP backup failed due to automatic ckpt creation failure when the file system was created with a physical DMbut mounted with a VDM path through GUI. 2
This issue has been fixed by adding a check that will prevent the GUI from creating a file system with a physical DM but mounted with a VDM path.
48710464/ 496443
Version 6.0.65.2
Backup Severity Symptom Description
The Data Mover panicked with a Page Fault Interrupt when VBB backup ran on a file system larger than 2TB.
Fix Summary
The code has been modified to prevent panic due to integer overflow.
Tracking Number
458269 / 44465210
1
Celerra Management Applications Severity Symptom Description Fix Summary Tracking Number
396490 / 38065374
1
Users could not log on to Unisphere if the passwords contained URL special characters like "&, ; ,+".
The issue has been fixed by encoding the password and user name to URL encoding. This issue was fixed by correcting the formula for calculation of the graph data.
2
"% CPU Idle" graph on "Resource Utilization History" page in Celerra Monitor showed value of 0%.
481652 / 46960472
CIFS Severity 2 Symptom Description
Files renamed by SMB2 clients could not be submitted to the virus checker.
Fix Summary
Code has been added to submit files renamed by SMB2
Tracking Number
472623 / 45655462
98
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
Severity
Symptom Description
Fix Summary
clients to the virus checker.
Tracking Number
2
The user was not able to delete a DFS link that existed beneath another DFS link.
The code disallows creation of a DFS link beneath another DFS link. This is consistent with Windows server behavior. The ACE now displays the owner name resolved through NIS and its uid number. It is no longer necessary to go to the owner tab. The code has been fixed to restore ability to enumerate and report user' s quota. The code has been updated to not scan a file if it has just been created during the open operation. The code has been fixed to correctly handle unreachable path even when it is longer than 220 characters. The code has been improved to rearrange some tests in the lock manager so that checking lock conflicts consumes less CPU. The code has been fixed to allow open a file for read attribute even when another process has the file open.
475820 / 46376654
1
When a user changed the UNIX rights of a file located on a file system mounted with MIXED access policy, the ACE displayed on Windows showed only "Owner Rights". To check who corresponded to this ACE, the user should go to the owner tab. The access checking was not affected. Quotas enumeration stopped responding indefinitely on Windows7 and W2K8 clients (clients using SMB2 protocol). A file could not be created on File Level Retention (FLR) if scan on read was enabled.
478365 / 46563750
2
481974 / 46640214
2
482464 / 47200498
1
server_cifssuport -cred -path <path> command panicked on Data Mover when the path specified was longer than 220 characters and was unreachable.
486226 / 47626966
1
Data Mover CPU consumption reached 100% when there were a lot of file range locks.
487750 / 47801326
2
Internet Explorer returned an error when downloading files on Windows 7 with 64bit CAVA.
490177 / 47973450
EMC Celerra Network Server Version 6.0.70.4 Release Notes
99
Version 6.0.65.2
Severity
Symptom Description
OS X 10.7 could not mount CIFS FS because the reply of opening CIFS might report invalid maximum access rights of the logged user if rights were set from NFS on a file system mounted with mixed access policy. The Data Mover panicked when virus checker was enabled with param waitTimeout set.
Fix Summary
The open reply now correctly reports maximum access rights of the logged user. The code has been fixed to avoid the panic condition.
Tracking Number
2
491591 / 45540912
1
DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x000110f214 Err code: 0 Target addr: 0x0000000800 ** Data Mover panicked with “out of memory” when CEPP was started with FT level=1 and no more CEPP servers were available. The code has been fixed to avoid the panic condition.
496627 / 48742734
1
The following message was displayed: DART panic/fault message: >>PANIC in file: ../bufpool.cxx at line: 425 : couldn't get a free page
499427 / 49063234
CIFS Virus Check Severity Symptom Description Fix Summary
The code has been fixed so that when a client disconnects, all CIFS threads working for the client are unblocked. The code has been fixed to generate the correct event.
Tracking Number
2
The disconnection of a client did not unblock CIFS threads.
476743 / 46658574
2
CEPP generated a “file open read” event incorrectly instead of “open dir” event.
500020
100
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
Control Station
Severity Symptom Description Fix Summary Tracking Number 467458 / 45558122
1
Some of the Unisphere GUI pages like "New File System Page" and "File System Properties Page" took over 20 minutes to load on a gateway system connected to Symmetrix array.
The performance issues have been fixed by revising the code. The Unisphere GUI pages take approximately 2 seconds to load with the fix. CCMD message definition has been modified to clarify the term "matching storage pools". This issue has been fixed by displaying "NoLimit" instead of "EXPIRED" when file system quota grace period is defined as "unlimited". This issue has been fixed by removing an old restriction which limited the new HLU number to be assigned to a newly discovered LUN. This issue has been fixed by limiting filesystem inconsistent density check to CLARiiON-based filesystems only.
2
Creation of a Virtual Data Mover on a mirrored storage pool failed with an unclear error message "No space left in matching storage pools". nas_quotas report displayed "EXPIRED" in "Timeleft" column even when filesystem quota grace period was defined as "unlimited".
375163 / 35536804
2
466085
2
nas_diskmark failed to mark a new LUN and displayed a warning similar to the following:- "Warning: 17716815787: server_2 c16t0l6 skipping reserved LUN id, APM00111803762 stor_dev=0x00ED (237 decimal), maps to a reserved host LUN id 0x6." If nas_diskmark executed on a gateway system tried to attach to a Symmetrix array, it failed with an error similar to the following:- "Error 5008: server_5 :13421849340: File system root_avm_fs_group_1 is made up of disk volumes with inconsistent density."
472963 / 46066538
2
474902 / 46033776
EMC Celerra Network Server Version 6.0.70.4 Release Notes
101
Version 6.0.65.2
2
For a production file system mounted on VDM, if a new checkpoint was created with a name that has one or more spaces and with automount=yes, the checkpoint was created but remained unmounted. The following error was returned: Error 5005: Invalid argument failed to complete command. Disk Provisioning Wizard failed on NX4 system with the following error: "Not able to process templates. n Command failed, error code: 1, output: errorMessage:string="Timeout (180 seconds) waiting for state SS_TEMPLATES_GENERATED". " CLI command "nas_pool -list" failed with the following error: "Error 2237: Execution failed: Segmentation fault: Operating system signal. [VIRTUAL_DISK_TABLE.is_profile_s upported] "
The issue has been fixed by replacing every space with "_" character in mount path created from the checkpoint name.
476767 / 4665800
2
This issue has been fixed by revising the code responsible for generation of hot spare templates.
483778 / 46920576
2
The issue has been fixed by initializing the storage_profile_
name to empty string when the name is not present in database record. This issue has been fixed quoting file system name in "rootnas_fs" command to allow spaces.
487182 / 47678224
3
File system auto extension failed when there was space in file system name.
478780 / 46910078
102
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
DART Core Severity Symptom Description
Log event interface showed "Parsing error. Invalid XML” messages. The server_log showed following valid error messages, but with garbage characters at the end of each message. 2012-03-28 09:13:37: 150596354069: TIMESYNC: 3: ReadStream failed in time synchronization poll, error: _ŠÁ 2
Fix Summary
The code has been corrected to display the accurate error messages. The log event no longer displays "Parsing error. Invalid XML” messages.
Tracking Number
479218 / 46947888
.
2012-03-28 09:13:37: 150596354081: TIMESYNC: 3: Connection to the timeserver lost 2012-03-28 09:13:38: 73286942722: TIMESYNC: 1: 2: The NTP servers are not accessible or responding to the time service from [Wed Mar 28 09:13:37 2012]. The Data Mover panicked with the following message: DART panic/fault message: The code has been fixed to avoid the panic condition. 483525 / 47383678
1
>>PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG
Deduplication Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
103
Version 6.0.65.2
1
Due to a race condition with DART's internal dedupe process, deduplication could be attempted on an inode that was deleted and recycled into another type of file. This could cause an internal file to be mistakenly deduped. The result could be loss of access due to blocked threads.
The race condition has been fixed. It is no longer possible for the DART deduplication process to accidentally compress the wrong inode.
487774 / 47818786
File System Severity Symptom Description
When a file system was nearly full or reached a quota limit, a large copy from an SMB2 client could fail. If you retry the copy, it may complete successfully. Fsck panicked with header "PANIC in file: ../DeltaIO.cxx at line: 725 : DeltaIO_Read_CB::failed" due to bad block in the opaque data mapping pointers. This caused an extended DU situation. When the amount of metadata changed, it caused the ufslog to become full. Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../staging.cxx at line: 439 : failed to write UFS log The result was that file systems should be checked before they could be mounted.
Fix Summary
The Code has been fixed to correct the method to reserve and consume blocks during a copy from SMB2. The code has been enhanced to check for and skip bad blocks.
Tracking Number
1
477292 / 46343472
1
483028 / 46874118
1
The code has been fixed to prevent the panic. When the Staging buffers are full, the current buffer will only be updated if the next buffer is unused.
490046 / 48019080
MPFS
Severity Symptom Description A discrepancy may build between the du and df commands for an MPFS file system. Fix Summary Tracking Number 463971 / 45169514
1
The issue has been corrected to free space
104
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
on MPFS mounted file system when files are deleted.
Network
Severity Symptom Description Fix Summary Tracking Number
DART panicked when the user issued a very large SNMP bulkcache request with the following message:
DART panic/fault message: 1 >>PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG The code has been fixed to avoid the panic condition. 478969 / 46919840
The following panic might occur while running Nesus security scanner: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x000020f310 Err code: 0 Target addr: 0x0000000028 ** Panicking Processor: 2 DART stack trace: 2860: 0xc2aef0 _ZN7DskDump9checkdumpEPKc+0x7a 28a0: 0xc2b0fe dump_return+0x4 28b0: 0x14d0b2 _ZN12Memory_Frame13setProtectio nEm19Memory_Prot 28d0: 0x14d73d fault_dump+0x89 28f0: 0x14165d handle_fault+0x1d 69910: 0xd89489 _ZN13cPipeSimulate11checkPolicy EP13smb_threadC 69950: 0xde498b _ZN21cMsRPC_LsarpcSimulate15lsa rCloseHandleEP1 699b0: 0xdf202e _ZN21msrpc_legacyOperationI21cM sRPC_LsarpcSimu
1
The code has been fixed to handle invalid MSRPC requests gracefully, returning an error.
484635 / 47497156
EMC Celerra Network Server Version 6.0.70.4 Release Notes
105
Version 6.0.65.2
NFS
Severity Symptom Description It was possible to receive an illegal RPC reply with a NULL body and a status of "YP_TRUE" when using NIS with an NIS Server AIX 5.2 and/or AIX 5.3. The system expected the RPC message to contain at least 4 bytes and was not prepared to handle a NULL body RPC request, so it processed the request normally. This would result in storing data more than one byte into one byte space. That overwrite led to the following panic: DART X86_64 panic message: >>PANIC in file: ../malloc.cxx at line: 662 : Memory corrupted Fix Summary Tracking Number
1
The code has been fixed to check for a NULL RPC body to properly process the NULL message and reject the bad NIS response.
469561 / 45818200
NFSv4
Severity Symptom Description The Data Mover panicked when a file system was unmounted while there was NFSV4 activity on it, with the following message: Fix Summary Tracking Number
1
Page Fault Interrupt. Virt ADDRESS: 0x00006a4ba3 Err code: 0 Target addr: 0x0000000010 ** The premature revocation of LOCK states by the server caused BAD_STATEID errors. The server might revoke the LOCK state as soon as there was no file lock left attached to it;
A spinlock has been introduced in the code to avoid reusing NFSv4 states which are being released.
476566
2
This did not comply with the RFC.
The delay between the last LOCKU operation and the revocation of the LOCK state could vary between 0 and 5 minutes (nfsv4.cleaningInterval).
The Data Mover no longer releases the idle LOCK states associated with files which are open and idle.
484690 / 47472438
106
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
Proactive Health Checks
Severity Symptom Description If a CIFS server was present with WINS servers (used wins= option with server_cifs while adding CIFS servers), the server_checkup failed with the following error: Warning 17451974760: server_2 : VDM1: The WINS server '<IP address>' does not contain the correct IP address for the name '<CIFS server name>'. This one should be associated with <IP address>. The CIFS clients of this NetBIOS server might not be able to access the Data Mover. The code has been fixed to enhance the IP address comparison function used during server_checkup. Fix Summary Tracking Number
2
479534 / 46324040
Quotas
Severity
Symptom Description
While you had named streams in files, after running online quota check, quota block and inode usages would be less than the previous number. Querying user quotas on a tree produced no result, no error message from Linux client over NFS.
Fix Summary
The code has been fixed to calculate named stream and container files in quota check correctly. The code has been fixed to check the quota tree on/off status and return correct quotas.
Tracking Number
2
466450 / 43457680
3
473741 / 45791204
EMC Celerra Network Server Version 6.0.70.4 Release Notes
107
Version 6.0.65.2
3
Any query for tree quota usage following the modify tree quota operation returned 0. The same query would return values 24 hours later.
The code has been fixed; the query to return tree quota usage information following the modify tree quota operation now returns accurate results. A new option has been added to the param quota useQuotasInFsStat to enable reporting only file system and tree quota accounting to df queries from Unix clients. The existing values will behave as before. 0: (default) only the file system capacities will be reported to Unix clients 1: the file system and all quotas (user, group and tree) are taken into account when capacities are reported to Unix clients 2: file system and the tree quota are taken into account when capacities are reported to Unix clients The fix ensures the tree quota accounting is accurate.
474357 / 46279178
There was no option to report capacities and usage based on file system and tree quotas that did not take user and/or group quotas into account.
1
488023
1
When a large file was deleted, this was handled in the background as a deferred delete. If this process was interrupted by the file system being unmounted, it could result in tree quota accounting being incorrect.
493473
108
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
Replication
Severity
Symptom Description
Data mover panicked with the following message:
Fix Summary
Tracking Number
1
DART panic/fault message: *** Invalid Opcode exception. Virt ADDRESS: 0x0000006bf6 ***
The code has been fixed to avoid the panic condition.
489235 / 47953422
Snapsure
Severity Symptom Description Data Mover bootup or failover/failback could be slow due to checkpoint recovery when there were mergeDone checkpoints.These were caused by deleting or refreshing checkpoints which were not the oldest for the file system. The chunks for the deleted/refreshed checkpoints were merged into existing checkpoints and these chunks caused extra processing to be done during recovery after a reboot or failover/failback. Fix Summary Tracking Number
2
The checkpoint recovery code has been improved to skip reading the mergeDone chunk headers as it is not necessary.
408286
1
When latest ckpt was very large (several TB), it would take a long time (longer than 5 minutes) for reboot or failover. Depending on the content in slot B, this would cause bad cg, fs corruption or btree corruption. The data mover panicked with the following message: DART panic/fault message: finishTake: bad magic in CG
The checkpoint recovery code has been improved, leading to improved reboot or failover time.
436839
1
The code has been fixed to avoid the panic.
472453 / 46122574
EMC Celerra Network Server Version 6.0.70.4 Release Notes
109
Version 6.0.65.2
1
Writing a PFS where a checkpoint was created caused a panic with the following message: "io not progressing". The recovery process was suspended after the panic. PFS on the replication destination side might be corrupted as a result of this.
The code has been fixed to correct the Snapsure recovery process.
483546 / 47403530
1
When replication started in reverse with HWM 0, it always returned an File_NoSpace error.
The code has been fixed so that only when the usage is greater than or equal to 75%, the replication reverse would fail, otherwise it succeeds. The code has been fixed to avoid panics by: • Avoiding the insertion of invalid block numbers • Adding checks to bypass invalid block numbers that may have been written previously.
491140 / 48120034
1
Repeated Data Mover reboots could be triggered during replication transfer or checkpoint merge on >2 TB file system. Any of the following messages could be displayed: Panic in file: ../DeltaIO.cxx at line: 802 : DeltaIO_Read_CB::failed Assertion failure: 'cv>unpackBtree32(curadd) >= startAddress/VOLBLK_SIZE' on line 3286 in file '../PagedBlockMap.cxx' >>PANIC in file: ../assert.c at line: 27 : Assertion failure: 'val != INVALIDNUM' on line 428 in file '../include/PagedBlockMap.hx x' >>PANIC in file: ../assert.c at line: 27 : assert failed
497062 / 48857254
1
Rolling data mover panic occurred when a little amount of data wrote to PFS after a new CKPT creation and the corresponding save volume was corrupted. The following message was displayed:
The code has been fixed to avoid the panic condition by correcting the Snapsure recovery process.
497107 / 48872748
110
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
DART panic/fault message:
>>PANIC in file: ../PagedBlockMap.cxx at line: 2423 : Duplicate Mapping
Storage
Severity Symptom Description Data mover panicked with "Volume has zero paths" . Fix Summary The code has been changed to not prune paths to lun on which scsi inquiry fails. This avoids the panic. Tracking Number 458436 / 44563854
1
The following message was displayed:PANIC in file:
../include/volume.hxx at line: 414 : Volume has zero paths.
1
A rare panic was encountered while processing a Tachyon fibre channel driver's Outbound Completion Message (OCM) with a set Invalid Entry (INV) error bit, although neither the OCM's SEST Programming Error (SPE) bit nor the VAL bit in the associated SEST were set. This condition violated the rules of the OCM's INV bit as specified in table 9-2 of the Tachyon FC Protocol Controller User Manual. The driver explicitly panicked the Data Mover (panic code = 0x01902009 == FC_EXM_INVALID_API_ERROR_STAT US with an additional error code = 0x00003002 == FC_CTLR_PROG_ERR) whenever an OCM's invalid entry bit was set without even looking at the OCM's SPE bit or the SEST entry's VAL bit.
The fix prevents this explicit panic within the Tachyon fibre channel driver (panic code = 0x01902009 == FC_EXM_INVALID_A PI_ERROR_STATUS ) in the specific case (error code = 0x00003002 == FC_CTLR_PROG_E RR) where it violates the rules for the OCM's INV bit as specified in the Tachyon FC Protocol Controller User Manual.
459091 / 44655534
EMC Celerra Network Server Version 6.0.70.4 Release Notes
111
Version 6.0.65.2
1
DART panicked with a page fault. The following message was displayed: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000a69602 Err code: 0 Target addr: 0x0000000010 **
The offending code leading to page fault panic has been corrected.
470802 / 45934948
2
Deletion of the disk failed with error:"Error 3501: Storage API code=24: SYMAPI_C_DEVICE_NOT_IN_DG. Could not select requested device(s) from the specified device group"
The issue has been fixed by changing the way the code determines the name of the device group the disk belongs to.
494402 / 48470488
SRDF
Severity Symptom Description Fix Summary Tracking Number 473453 / 45851944
1
RDF Set mode sync operation failed when Sync license was not enabled.
The fix will ignore the license error message to continue the restore process. The fix should stop such a nd-clnt hung issue.
1
nas_rdf -activate could be hung in nbs update operation. ps -ef shows "nd-clnt 6 7" like "/usr/local/bin/nd-clnt 6 7 server_2,server_3,server_4,server_5".
475565
System Management
Severity 1 Symptom Description When a security user was renamed, the home directory did not get renamed to reflect new name. Fix Summary This issue has been fixed by renaming the home directory to match the new user name. If a directory with the new name exists, operation fails and an error is displayed. Tracking Number 90188
112
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
2
No event notifications e-mails were received when an event notification was configured with Action 'mail' and Destination 'user' .
This issue has been fixed by correcting the software error that prevented email notification to be sent to the correct destination. This issue has been fixed by correcting the user authentication code that handles disabled accounts. This issue has been fixed by proactively deleting outdated temporary files used to determine statuses of hardware components. The issue has been resolved by adjusting configuration parameter of Celerra Common Cache. This issue has been fixed by preventing Jserver from trying to access Symmetrix to check the threshold of the virtually provisioned pool. This issue has been fixed by moving the ECOM core files to /nas/var/dump/ partition, compressing the files and limiting the number of retained core files. The issue has been fixed by stripping off leading zeroes from Symmetrix serial number.
399760 / 37640588
2
After disabling a local user account, login into Unisphere through an LDAP authenticated account failed.
459428
2
Incorrect alerts about hardware components appeared in Unisphere GUI after the actual faults were corrected.
466065 / 45410530
2
XML API failed to return all provisioned volumes.
480447/ 45516028
2
Control Station periodically ran out of memory and rebooted.
486239 / 47586416
2
UIM could not discover the VNX and displayed the following error message: "HTTP error 405 method not allowed".
487238 / 47682672
2
Jserver returned disk volumes with Symmetrix serial number that had leading zeroes.
488668 / 44892982
EMC Celerra Network Server Version 6.0.70.4 Release Notes
113
Version 6.0.65.2
2
The following error was seen in sys_log and the Unisphere event log: transfer_support_materials[264 56]: The remote FTP server/directory cannot be reached. Verify your configuration in /nas/site/automaticcollection. cfg. Also, nas_checkup reported the following warning: Control Station: Check auto transfer status Warning HC_CS_18800050419: Cannot establish connection to FTP server sftphost sftp.emc.com Action : Command "/nas/tools/automaticcollectio n -verifyftp" failed to connect to FTP connection. Check the configuration in /nas/site/automaticcollection. cfg file. For more information, search the Knowledgebase on EMC Online Support as follows: 1. Log in to https://Support.emc.com and go to Support > Knowledgebase Search> Support Solutions Search. 2. Use ID emc221733 to search.
The issue was caused by a recent change in FTP server security, which caused the verification to fail. The code has been updated to work with the new FTP server security policy. See Primus article emc296211 for the temporary workaround for fixing the problem with the script until the system is upgraded to this VNX OE for File version.
491605 / 48169232
2
Checkpoint creation via XML API failed with error:- "The specified storage system '0' is not valid."
This issue has been fixed by modifying the XML API to correctly handle the configuration where both Control Station and the storage array have the same serial number.
493653 / 48326724
VDM
Severity Symptom Description Fix Summary Tracking Number
114
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
2
The size of the security.evt log file increased beyond 512 KB when located on the root file system or on the root VDM when the GPO setting was set to a higher value, thereby filling up the root file system.
The code has been modified to limit the size of the security.evt file to 512 KB when located on the root file system or on the root file system of a VDM, regardless of the GPO settings.
473430 / 46236640
Version 6.0.60.2
Backup
Severity 1 Symptom Description VBB restore operation failed in restoring directories with UTF8 characters. The Data Mover panicked when the SavVol filled up during a greater than 60-day long backup and the oldest checkpoint was inactivated. Fix Summary The code was fixed to handle directories with UTF8 characters. The code was fixed to protect against inactivated checkpoints panicking the Data Mover. The number of active versus inactive checkpoints was verified to be correct. The code was fixed to avoid the panic. Tracking Number 459567/ 43756396
1
460751/ 44831934
1
The Data Mover panicked with the following message when DMA sent garbage data in env list during ndmp restore. PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG
465634 / 45403186
Celerra Management Applications Following are the fixed problems related to the Celerra Management Applications, that is, Celerra Monitor, Celerra Manager, WebUI, or Celerra MMC Snap-ins:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
115
Version 6.0.60.2
2
Unable to join the CS to the storage domain. The CS appears to join properly and no errors are thrown but it never appears in the System List. This happened because systems on the same subnet with hostnames undifferentiated in the first 12 characters ended up with identical, non-unique System IDs.
The code was updated to generate unique System IDs, resulting in CS successfully adding to the storage domain.
435338 / 43225714
CIFS
Severity 1 Symptom Description Windows clients issuing a DOS 'dir' command which included a complex mask resulted in all CIFS access to the file server being blocked. A forced panic of the Data Mover was needed to restore client access. When the CIFS server had many open files (likely more than 100000, depends on file names size), the open files enumeration over MSRPC (MMC, srvmgr) could not complete. Fix Summary The code was changed to re-format DOS masks for optimal execution before they were submitted to the file server. The code was fixed to prevent the cleanup task from running on inprogress active requests. This allowed the open files enumeration to complete. The code was fixed to return the correct size of qtree. Tracking Number 460364/ 44773644
2
462010/ 44848002
2
When connecting by CIFS to a qtree using the share of the root of the FS and the qtree path, the size of the qtree may be incorrect (the size of the FS is returned instead of the size of the qtree). Users could experience access issues to server's resource when domain was configured to use Resource Groups through Active Directory Federation Services (ADFS), as described in the Microsoft document: http://technet.microsoft.com/
464557 / 45104472
2
The code was modified to handle credential resource groups correctly.
464981/ 45324904
116
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
enus/library/cc737832(WS.10).a spx
2 When a non server's administrator user attempted to connect C$ system share (access not granted), the access checking was not performed immediately, but access denied error was later reported on first request, when using the share. The server_stats and ”printstats cifs” did not report correctly for SMB2 TCPconnections and SMB1/SMB2 threads. The code was fixed to check for access immediately now. 466678 / 45510170
2
The SMB2 TCPconnections and SMB1 and SMB2 threads counters reported by server_stats were corrected. The code was enhanced to allow a CIFS user to create a file and get the Execute Unix mode bits set according the ACL of the file. The code was enhanced to improve performance.
468729/ No Clarify ID
2
The cifs acl.umask parameter was documented but not useable.
469523 / 45695180
2
Bad performance when copying several files from a Windows 7 client to a Data Mover share. This is because only one file was copied at a time.
473384 / 46207560
CIFS Virus Check
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
117
Version 6.0.60.2
Severity 2
Symptom Description When a file was being opened for writing by one client, and another client was reading attributes (or extended attributes) of this file, an unnecessary virus checker scan was triggered for the file causing write performance degradation.
Fix Summary With the code change, the virus checker scan could not be triggered when the second client was reading attributes only on a file that was opened for writing from the first client. The code was fixed to avoid the panic condition.
Tracking Number 463462/ 43901404
1
When an NFS user renamed a file on a file system with CEPP enabled for NFS, the Data Mover panicked with the following message: Page Fault Interrupt. Virt ADDRESS: 0x00007ec8f3 Err code: 0 Target addr: 0x0000000188
463584/ 45067744
CORE
Severity 2 Symptom Description The Data Mover panicked with the following message due to a memory leak when F-RDE was scanning a file system with a large number of hard linked files: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000159274 Err code: 2 Target addr: 0x0000000000 ** Under load, the Data Mover crashed with "Out of memory". The following message was displayed: DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory This could happen randomly Fix Summary The code was fixed to avoid the panic condition. Tracking Number 459484/ 44220834
1
The code was fixed so that the memory used by the DBMS internal mechanism was pre-allocated at the initialization of the system to avoid the lack of memory situation.
460260/ 44774392
118
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
Severity
Symptom Description when the memory was fragmented.
Fix Summary
Tracking Number
Deduplication
Severity 1 Symptom Description Heavy CIFS access to compressed file on an F-RDE enabled filesystem can cause excessive CPU load. Fix Summary The code was fixed to avoid the situation causing excessive CPU load. Tracking Number 469525 / 45397734
DHSM
Severity 1 Symptom Description The Data Mover panicked with the following panic message when the user created a DHSM http connection and tried to modify this connection using an incorrect fs_dhsm command (-user option was used but no user name was provided). DART panic/fault message: *** GP exception. Virt ADDRESS: 0x0000aa783f. Err code: 0 *** The user cannot read archived files when the stub file is restored from backup. Fix Summary The DHSM command parser code was made more robust to avoid the panic. Tracking Number 422536/ 41631464
1
The code was enhanced in the DART kernel at the file system level to support the call that VBB restore operations was using.
440647 / 43018304
File System
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
119
Version 6.0.60.2
1
Data Mover server_2 panicked due to "Unnamed stream is wrong type". The release of DART internal resources for files containing (or once containing) alternate data streams is a multi-step release. DART incorrectly allowed an update to an inode that had already been freed. That inode had already been reused by another file and its contents were overwritten.
The code was fixed to avoid this condition during the release of files containing (or once having contained) alternate data streams.
440553 / 43857930
1
Using NFSv3 and lockd, could not unmount a file system on the Data Mover, resulting in a timeout.
Fixed a race condition which can occur when the lockd client uses several interfaces to set file locks on the Data Mover. The memory allocation needed to enable a tree quota was split up so one large contiguous chunk is no longer needed. This will prevent the Data Mover panic when it has been up long enough for its memory to become fragmented.
458487 / 44588032
1
1
When creating a new tree quota the Data Mover panicked with: PANIC in file: ../addrspac.cxx at line: 529 : Out of memory DART stack trace: e6e30: 0x1417d4 _ZN19Memory_AddressSpac e10allocPagesEl12Memory e6e70: 0x242509 malloc+0x10b e6ed0: 0x243dcb _Znwm+0x1b e6ef0: 0x243df6 _Znam+0x4 e6f00: 0x812804 _ZN14UFS_FileSystem20tr eeQuotaOn_InternalEPcS0 e71c0: 0x815a26 _ZN14UFS_FileSystem16qu otaOn_InternalEPcS0_+0x e7300: 0x786df6 _ZN14VFS_FileSystem7quo taOnE9quotaTypetPcS1_+0 e7350: 0x779f92 CPU usage intermittently spikes to 5% idle when there is high read activity on large read-only file systems.
459816 / 44708688
The code was fixed so that the process to free inode block objects was
464911 / 45307774
120
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
modified, so it consumes less CPU resources. 1 A corrupted NLM request caused Data Mover panic with the following message: Page Fault Interrupt. Virt ADDRESS: 0x000087d7fa Err code: 0 Target addr: 0 Datamover stack: #0 Lockd_Initial::start(th is=0x360cdc40) lockdthr.cxx:1401 #1 Sthread_startThread_int ernal(ip=0x2d4464a60) sthread.cxx:698 When the ownership is changed for a file or directory containing Alternate Data Streams (ADS), the inodes for the ADS are not changed. This will cause user or group quotas to be incorrect. The Data Mover hung in a rare situation while there were 3 tasks running at the same time: quota check or quota on task, IO task, and snapsure task. Threads become blocked eventually leading to a loss of access. This can happen when a file or directory containing a compressed named stream is deleted. In DART 5.6.47 and above, DART 6.0, and VNX OE for File 7.0, compression of named streams is not performed so this issue is only possible on a file system where dedupe was enabled on 5.6.46 or earlier. Fixed the use of uninitialized pointer. 466579/ 45508014
2
Corrected the logic error to make the ownership of ADS consistent with the file/directory.
467158 / No Clarify ID
1
The code was fixed to avoid the hang.
468422 / 45715556
1
The code was fixed to change the locking that is required when deleting a named stream to avoid the potential of a deadlock.
469851 / 45822506
EMC Celerra Network Server Version 6.0.70.4 Release Notes
121
Version 6.0.60.2
1
Data Mover panics with: PANIC in file: ../ufslog.cxx at line: 2728 : filesystem log transaction record overflow This happens due to a rare case when a file system operation results in a ufslog transaction larger than the maximum allowed.
The code was changed so that the number of blocks which can be included in a ufslog transaction have been reduced to ensure that the total transaction size will not exceed the maximum allowed.
476163 / 46597972
FLR
Severity Symptom Description Data Mover panicked with the following message: DART panic/fault message: >>PANIC in file: ../rawirp.cxx at line: 607 : Write Verify failed for FLR data Fix Summary The code was fixed to avoid the panic. Tracking Number 461016/ 44839716
1
FSCK
Severity 1 Symptom Description If a PBR check was required during FSCK or if quota recalculation was required when mounting the file system, the inode scan routine did not print any progress message making the user believe that the operation was hung. Fix Summary The code was fixed so that the inode scan could print the progress messages. Tracking Number 463902/ 45121204
Network
Severity
1
Symptom Description If IPv6 was configured in the network environment, but IPv6 interface was not configured on the Data Mover, the Data Mover could periodically panic
Fix Summary The panic was triggered by invalid sanity checking code, which did not expect a destination
Tracking Number
440268 / 43805094
122
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
Severity
Symptom Description with a panic dump header similar to: *** Page Fault Interrupt. Virt ADDRESS: 0x0000489c28 Err code: 2 Target addr: 0x0000000010 ** Data Mover panicked with the following: PANIC in file: ../bufpool.cxx at line: 425 : couldn't get a free page If a customer joins a CIFS Server to a Windows domain that is using a non-Windows DNS Server, the subsequent secure (Kerberos) DNS update from the Data Mover will fail because the DNS Server is not registered in Kerberos for the domain. If the Kerberos credentials for the CIFS Server are cached in a certain order, memory leaks will occur due to a credential being retrieved from the cache and selected as a 'match', then not used.
Fix Summary address to be IPv6, and source address to be IPv4. This sanity checking code has been fixed.
Tracking Number
1
Additional calls to free the credential contents were added to prevent the memory leaks.
460236 / 44786942
1
The Data Mover panicked due to experiencing IP fragmentation and any of the fragments are zero length. The following message was displayed: DART panic/fault message: Page Fault Interrupt. Virt ADDRESS: 0x00003d49a8 Err code: 2 Target addr: 0x0000000000
The code was enhanced to avoid UDP and optimize infrastructure to avoid IP fragmentation.
470446 / 45912724
EMC Celerra Network Server Version 6.0.70.4 Release Notes
123
Version 6.0.60.2
NFS
Severit y 1 Symptom Description The Data Mover panicked with the following message: Fix Summary The code was fixed to avoid the panic. Tracking Number 470457 / 45931264
Page Fault Interrupt. Virt ADDRESS: 6638df Err code: 2 Target addr: 4 There were server logs like "NFS: 4: xid XXXX in progress - reset after drop more than 8 times " NFSv4
Severity Symptom Description The Data Mover panicked with the following message:
Fix Summary The code was fixed to avoid the panic.
Tracking Number 460992/ 44859406
1
DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x00006a7fc0 Err code: 0 Target addr: 0x0000000000 **
2 The AIX client hung when using NFSv4. The Server could no longer grant a read delegation if the open access mode was read + write. If a write delegation could not be granted, no delegation would be granted at all. The code was changed, so that the conditions which could trigger this infinite loop were fixed. 464549/ 44422786
1
All NFS threads were blocked causing NFS access to be lost. This could have happened due to an edge condition with NFSv4 when verifying the attributes for an NFSv4 operation leads to an infinite loop. The NFSV4 server returned the error NFS4_RESOURCE when
465354/ 45380486
1
The code was fixed to release an open
465606/ 45378140
124
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
Severity
Symptom Description there were too many open owners. This could also happen if there were few opened files. The reason was that open owners were released only after a lease period to allow close replay detection.
Fix Summary owner when it had no more files opened. This avoided the issue.
Tracking Number
Platforms
Severity 3 Symptom Description HTTP 413 “Request Entity Too Large” page might return unsanitized client header. The connectemc daemon hung and stopped sending the callhome files. Fix Summary Changed webserver configuration file to use a static error message. A solution was implemented to detect and restart the connectemc main process if it was hung, and added a regular restart mechanism of the connectemc main process as the last recovery resort. Changed the web server configuration file to disallow weak ciphers and SSLV2 by default New code was added to read local user if no domain user found, and display the local user in AUDIT_Tool output. Tracking Number 355867 / 33486008
2
453530 / 44047694
2
By default, the web server on the Control Station allowed weak ciphers and SSLV2 to be used. syslog messages report blank for Username fields in case of local user when AUDIT_Tool is run.
456362 / No Clarify ID
2
470581 / 43023590
Quotas
Severity 2 Symptom Description ”User Name” column on quota information page in Unisphere showed the User ID and not the Domain\Username. Fix Summary The code was fixed to display the user information correctly. Tracking Number 464377/ No Clarify ID
EMC Celerra Network Server Version 6.0.70.4 Release Notes
125
Version 6.0.60.2
SnapSure
Severity Symptom Description The Data Mover panicked on ckpt fs with fs corruption; or on snapsure layer with Duplicate Mapping. The following message was displayed: Fix Summary Code was fixed to avoid the panic situation. Tracking Number 439471 / 43674774
1
DART panic/fault message: >>PANIC in file: ../DeltaVol.cxx at line: 3844 : doSavVolIOVec
Storage
Severity 1 Symptom Description "server_mt -f [device] status" fails when connected to DataDomain VTL devices. The NDMP Backup failed with the following medium write error message on direct attached LTO5 drive with append only feature : Fix Summary Code was fixed so that server_mt returns the correct status. The code was changed to add support for 16 byte SCSI commands for tape/robot devices. Tracking Number 392665 / 36965502
2
431941/ 41648218
---------- server log--CAM: 3: cdb: 0a 00 00 04 00 00 00 00 00 00 00 00 | SCSI ERR ON #3200 camstat 0x84 bstat InvalidOperation | scsi stat.skey.ascq 0x02.07.5a02 Check condition | skey [ ] Write protect |____________________
EMC Celerra Network Server Version 6.0.70.4 Release Notes
126
Version 6.0.60.2
Severity
Symptom Description
Fix Summary
Tracking Number
ascq Operator selected write protect System Management
Severity 2 Symptom Description Abort of "server_ping -send" command with Ctrl+C hangs serial connection to Control Station. Fix Summary Code was modified to protect output to console operation from being interrupted by external signal, and the interrupt signal is delivered after the operation completes. The code was corrected to ensure that server events get retrieved according to the user specified severity level. Tracking Number 406307 / 39377292
2
File Events filtering by severity does not work correctly on Unisphere Event Logs page: when user requests "Info" events only, both "Info" and "Warning" server events get displayed. And when the user selects "Warning" events only, no server events get displayed at all. Upgrade failed with error message "Error
439164 / 42989446
1
IM_14505279794 : Failed to setup the following local storage devices ..." due to ongoing RAID group defragmentation on storage array.
3 Server_export command failed with error message ”Invalid argument” when a host FQDN contained "sec" clause.
The code was changed to remove obsolete 'exdisks' check performed while running 'nas_storage -check -all' command.
458484/ 44421782
The code that processed export options was modified to ensure that "sec" clause in FQDN did not get process as "sec=" export option.
460366/ 44678140
EMC Celerra Network Server Version 6.0.70.4 Release Notes
127
Version 6.0.60.2
Severity
Symptom Description Errors like the following appeared in the server log of a standby server:
Fix Summary The code was modified to disallow sending VTLU queries from Control Station to standby server.
Tracking Number 465968/ 45378244
2
2012-01-09 02:35:26: 13160087552: SVTL: 3: Command failed: poolsInfo 2012-01-09 02:35:26: 26044989440: SVTL: 6: SvtlServer() --> svtl_rename(/svtl, /.etc/svtl) failed: Stale handle 2012-01-09 02:35:26: 13160087552: SVTL: 3: mkDir(/.etc/svtl/lock s/libraries) failed: IO_Error 2012-01-09 02:35:26: 13160087552: SVTL: 3: SvtlServer() --> pTlusMutex>Init(/.etc/svtl/lock s/libraries) SVTL_STATUS_ERROR 2012-01-09 02:35:26: 13160087552: SVTL: 3: SvtlServer::Alloc(/.e tc/svtl): SVTL_STATUS_ERROR
2 Control Station exhibits high CPU load.
The code was fixed to reduce Control Station CPU load by limiting periodic filesystems queries initiated from Jserver
470280 / 45558122
UFS
Severity 1 Symptom Description Failed to mount the file system with the following message: ”Mount not allowed: too many mounts on this server”. Fix Summary The code was modified to handle this situation. Tracking Number 165703/ 33316276
128
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
Severity 1
Symptom Description CPU usage spikes of 100 percent lasting for several seconds occurred when clients accessed tens of thousands of directories. The Data Mover panicked with panic header ”Assertion failure: 'aclHdr.getHdr()->revision == ACLRev” if ACL database was full (1 million different ACLs) and there were concurrent threads trying to allocate a new ACL.
Fix Summary The code was changed to reclaim memory used in directory processing more efficiently. The code was fixed so that the solution eliminated the condition where two threads attempt to allocate the last ACL entry.
Tracking Number 457456/ 44256920
1
467142/ 45564250
Version 6.0.55.3
Celerra Management Applications
Following are the fixed problems related to the Celerra Management Applications, that is, Celerra Monitor, Celerra Manager, WebUI, or Celerra MMC Snap-ins:
Severity 2 Symptom Description Celerra Monitor failed to start with error: "Component <serial_number> is not accessible to this user" Fix Summary Data types in Jserver have been modified to fit Symmetrix director configuration attributes without truncation The code has been modified to correctly handle filesystem names that included space characters. Jserver code was modified to stop producing problem reports when stale data was received from Common Cache. Jserver now records the record problem in the log file. Tracking Number 420801/ 41209492
2
Unisphere and Celerra Monitor did not display filesystem capacity and inodes utilization information for a filesystem name that included one or more space characters. A Data Mover failing over generated a very large number of problem reports which filled up the Jserver partition. This occurred due to a space issue under /nas/jserver partition.
432871/ 42832926
2
439763/ 11111111
EMC Celerra Network Server Version 6.0.70.4 Release Notes
129
Version 6.0.55.3
2
Numerous warning messages similar to the following in sys_log files: CS_PLATFORM:ADMIN:WARNING:18:::::d skMon[1245]: FS /dev/mapper/emc_vg_lun_5emc_lv_nas_var_emcsupport mounted on /nbsnas/var/emcsupport filling up (at 100%, max = 90%). Events were missing or the entire view was empty when viewing events from the Event Log Refreshing the view restored some or all of the events. More than one refresh caused events to be dropped or the view to empty. On very large (greater than 10 TB) MGFS filesystem after a large amount of data is migrated, the system started appearing unresponsive.
The code has been fixed to prevent the emcsupport partition from filling up. The issue that caused a background process to run infinitely has been corrected. The code has been fixed to display events correctly.
456533
1
462851
1
The code has been modified to initialize cg bitmap for MGFS filesystems to avoid unresponsive behavior.
455240 / 44242956
CIFS
The following entries describe fixed Common Internet File System problems:
Severity 2 Symptom Description In Windows Vista®, a new SID, OwnerRights, can be used in ACEs to control the rights on securable objects. The File OE did not support the new Vista OwnerRights SID. Data Mover panicked when Usermapper service received unexpected packet. Typically when security scanner sent invalid RPC to the usermapper service. An open file could not be forced closed on VNXe based storage systems. Fix Summary The code has been changed to support Vista SID. Tracking Number 368438 / 34831146
1
The code has been hardened to validate and reject incoming requests that are not formatted in the correct manner. The code has been modified to allow open files to be forced closed upon request. The code has been modified to add a parameter (cifs.admin.shareC_NotC reated)to hide C$. The default for the parameter leaves C$ visible.
404497/ 39107628
2
433118
1
All the storage system file systems were visible and accessible through the C$ share.
434264
130
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
Severity
Symptom Description
Fix Summary Disabling C$ visibility will not allow users to browse the different file systems from Microsoft MMC when a share is created.
Tracking Number
2
With Windows 7 and SMB2 client access, if a Word document had its archive bit un-set in DOS attributes, a modification in the Word document did not reset the archive bit. The Data Mover panicked when an application made a very large number of accesses to the same file from many different Windows Vista and Windows 7 SMB2 clients that used SMB2 file lock request operations. The message displayed is “Run out of free blocks in Buffer_CachedPool!" Users experienced a WERR_INVALID_PARAMETER error when attempting to configure shares from SQL clustered 2008 R2 server. This is because Celerra was rejecting the NetShareSetInfo request if the shi502_netname field was not populated. The command server_checkup generated an error in its report when executed on the FTPD test with a standard default configuration of FTPD. When the default persona of the FTP server is empty (no certificate and no private keys), then it generates the error “no available certificate for persona 'default'”. This error message is not correct. In quota reports, a discrepancy was observed in the reports shown in CLI and GUI quota screen for a approximately 234 users. This was in context with domain migration.
The code has been modified. Archive bit now works as expected with Word in Windows 7 and SMB2. The code has been modified to correct the memory leak in the SMB2 file locking mechanism.
434306/ 43096796
1
436698/ 43406710
2
If a netname is missing in a NetrShareSetInfo request, the code uses the share name from qualifier to prevent this error. The code has been modified. When the command is executed the validity of certificates in the persona is checked only if SSL is allowed, required, or enabled (that is,not disabled). The code has been fixed to avoid missing entries in the quotas report.
436846/ 42935986
2
440016/ 43715054
2
440448/ 43731696
EMC Celerra Network Server Version 6.0.70.4 Release Notes
131
Version 6.0.55.3
Severity 1
Symptom Description The Data Mover panicked when the Codenomicon/Defensics security scanner sent invalid SMB2 packets. The following message was displayed “>>PANIC in file: include/stream.hxx at line: 1130: datab:freeRef: db_ref <= 0” "The permissions are incorrectly ordered“ error message was displayed when the CIFS users opened the security tab for a file or folder. This was due to improper sorting of deny and access permissions by NFSv4 ACL. Windows 8 clients were unable to map EMC server shares. The net use command failed with system error 2148073478.
Fix Summary
Tracking Number
The code was fixed to protect against corrupted SMB2 commands. The code now rejects invalid commands and returns an error to the client. The code was changed so that NFSv4 ACL correctly sorts deny and access permissions.
453320/ 44042010
2
458223/ 44455658
2
The code has been fixed to support Windows 8 compatibility. This ensures that the Data Mover selectively checks an SMB2 request and signs the response even if the server was not configured to force signing. The code has been fixed. When access to share is checked, we return Read-Only if the file system is a snap and not a write-able snap. The code was modified to accept ACEs with SID 1-5-88 without returning an error. A new NT registry key HKEY_LOCAL_MACHIN E/Software/EMC/Widelin k/Global (Set per VDM) has been added as a choice if Widelinks is globally supported or per share using a new attribute. By default, HKEY_LOCAL_MACHIN E/Software/EMC/Widelin k/Global will be set to 1 which will result in keeping the existing
459810
2
A Windows user trying to create a new file (right click in Window Explorer) did not get a new option if the share they were working on was mounted on a network drive;file system is a writeable snap and client OS is Win7/W2K8 Copying files from MacOS X Lion to a Celerra CIFS share failed when the UNIX type SID had the form S-1-15-88. When using the Widelinks feature all shares for the CIFS server will have the DFS root attribute set, even those without any Widelinks under the share. This was found to cause an issue with Windows7 clients using offline caching mode because the DFS attribute affects which referral it selects.
461052
2
437759 / 43336414
2
438207 / 43020618
132
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
Severity
Symptom Description
Fix Summary behavior. When this key is set to 0 the Widelink attribute will need to be specially set on all shares containing Widelinks. The new Widelink attribute will be supported on share type (like for ABE and Offline cache). This attribute will have to be set using the type option on the server_export command: server_export -name myShare -option type=Widelink /myPath
Tracking Number
CORE
The following entries describe fixed problems related to Celerra Data Movers:
Severity 1 Symptom Description The Data Mover panicked when a file system mount point included special character "=". The root cause of the Dart panic is NULL fs volume ID (fsVolId). It returns from parsing Mount Command string with more than one pre-defined separator "=". The Data Mover panicked when a path larger than 256 characters is entered in a command. Fix Summary The code has been modified to handle special character “=”. 431762 456060/ 42742496 Tracking Number
1
The code was modified to handle paths greater than 256 characters in length.
436681/ 43287434
EMC Celerra Network Server Version 6.0.70.4 Release Notes
133
Version 6.0.55.3
Severity 2
Symptom Description Creating a checkpoint on the Control Station while the production file system was mounted on a Unix/Linux client did not allow that client to immediately cd into the .ckpt directory from the root of the file system, and/or all the checkpoints were not visible from the .ckpt directory at the root of the file system. Once the file system was unmounted and mounted, you could then cd to .ckpt, and all the checkpoints were visible from the root of the file system.
Fix Summary The code was fixed so that the checkpoints were assigned correct values when they were created, and all checkpoints and the .ckpt directory were visible to the client from the root of the PFS immediately after creation.
Tracking Number 455858/ No Clarify ID
Deduplication
The following entries describe fixed issues related to Deduplication:
Severity 1 Symptom Description The CIFS and NFS server became unresponsive and forced a reboot when using deduplication. The Data Mover panicked when Deduplication was scanning a file system containing a large number of hard linked files. The DART panic/fault message was “Page Fault Interrupt. Virt ADDRESS: 0x0000159274 Err code: 2 Target addr: 0x0000000000” Following a datamover panic and failover, users were unable to open deduplicated files. If a problem was found with the internal deduplication policy engine the state of deduplication would be set to Off. Under normal circumstances when deduplication is set off all files are recalled, however in this situation the state was set off but no recall was performed. This resulted in users being unable to access any deduplicated files. Fix Summary The deadlock was fixed by changing statistics logging. The logging caused a lock to be acquired in the wrong order. The memory leak was fixed. The code now releases allocated memory after the scan. Tracking Number 440939/ 43864948
1
455471/ 44220834
1
Deduplication will be suspended which means no new files will be deduplicated while still allowing access to any previously deduplicated files.
457093 / 43892970
134
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
File Systems
The following entries describe fixed File System problems:
Severity 1 Symptom Description The Data Mover panicked while creating a Read/Write check point filesystem. The message shown was: “PANIC:>PANIC in file:../addrspac.cxxat line: 529:” A Data Mover could deadlock if FS checkpoint or any other tasks that needed FS pause while iSCSI LUN/snap deletes were going on in the background. Fix Summary The code was modified to correct a memory leak in file system initialization. Tracking Number
436316
1
Locks were acquired in the proper order to avoid the deadlock situation.
459827/ 44694588
iSCSI
The following entries describe fixed issues related to iSCSI LUNs:
Severity 1 Symptom Description iSCSI LUNs that resided on the same filesystem were erroneously reported to the iSCSI initiator with the same WWN when the file system was unmounted on boot-up. If the file system issue was corrected and the file system was remounted, the iSCSI LUNs were still reported to the initiator with the same WWN until the Data Mover was rebooted. Fix Summary The code has been fixed so that the iSCSI service will not register a WWN for iSCSI LUNs whose file system is un-mounted and will not report those LUNs to the iSCSI initiator. Once the file system issue is corrected and the filesystem is remounted, the iSCSI service must be restarted by running the command "server_iscsi server_x -service -stop" and then the command "server_iscsi server_x -service -start" -or- by rebooting the Data Mover.After the iSCSI service is restarted, it will compute a proper WWN for the affected iSCSI LUNs and report them to the iSCSI initiator. Tracking Number
453767/ 44110140
Installation, Configuration, and Upgrades
The following entries describe fixed issues related to installations, configurations and upgrades:
EMC Celerra Network Server Version 6.0.70.4 Release Notes 135
Version 6.0.55.3
Severity 1
Symptom Description Earlier releases used ConnectEMC 2.6 software to deliver callhome messages to EMC. The outgoing mail functionality of this version issued duplicate HELO/EHLO commands. This was in accordance with RFC 2821, however, some mail servers do not support this (especially older mail servers). Biweekly health checks ran for a very long time and consumed a lot of CPU cycles if "server_checkup" reported more than 40 errors or warnings for one Data Mover.
Fix Summary
Tracking Number
This release uses ConnectEMC 3.0.1, which addresses the callhome issue.
377477/ 378198
1
The health check code has been modified to avoid this situation.
461161 / 44639742
NFS
The following issues describe fixed issues related to the Network File Server:
Severity 1 Symptom Description The Data Mover panicked when the NFSv4 client was revoked and requests to confirm its callback path had not been cancelled. The message displayed was “path is unavailable(frozen) or invalid” Fix Summary The code was modified to eliminate the race condition by making a check to verify the callback path has been cancelled before the NFSv4 revocation takes effect. The signature verification processing code was fixed to return the correct response. The code has been fixed to return the correct error status (OLD_STATE_ID). The code has been fixed. This flag will be correctly translated in NFSV4. Tracking Number
434416
1
The signature verification on a secure NFS request to delete the session security context was incorrectly processed, resulting in an auth_error response to the client. In the case of a Mac OS X client, the error caused the client to go into a rapid loop of creating and deleting additional sessions, which severely affected the Data Mover performance. When using NFSv4, an unexpected error status was returned by the server (BAD_STATE_ID), making the file unaccessible. After running chmod for files with NFSv4 ACL setting, inherited ACL settings got lost. INHERITED flag was dropped by NFSV4 ACL manipulations, because it was not returned to the client.
453465/ 43760246
1
456022/ 44309440 453335 / 43875390
1
136
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
1
Loss of NFS access after changing special mode bits (sticky, suid etc ...) from NFSv4. This issue is only possible when Windows security auditing is enabled and the file system is mounted using Mixed access policy. The server log will show messages like: KERNEL: 3: 3: ThreadsServicesSupervisor: Service:NFSD Pool:NFSD_Exec BLOCKED for 361 seconds: Server operations may be impacted.
The code has been modified to eliminate the potential deadlock when changing special mode bits (sticky, suid, and so on) from NFSv4.
455710 / 44297386
Replication
The following issues describe fixed issues in this release related to Celerra Replicator, SnapSure, and Checkpoint:
Severity 1 Symptom Description When the Data Mover panicked during a replication, the create BLO reported success to the user without writing the session signature to the sav volume. This later caused a problem for stop/start BLO, when a session was lost. The create remote replication command failed when a session on the source had the same FSID as a session on the destination. When a checkpoint was restored, it created a temporary checkpoint and snapped restored data. If there was not enough space, the restore failed leaving the filesystem inconsistent and replication inactive. The replication stop operation would attempt to restore a checkpoint to bring the destination to a consistent state. The restore operation failed as the destination savvol was full and could not be extended. Fix Summary The code has been modified to attempt to re-write the session signature after a panic interrupts a replication. If the re-write fails, the BLO will be rolled back and a CCMD message will be issued to the user. The code has been modified to reject the creation of a new session between a source and destination that is not unique. The issue has been fixed. When replication aborts and tries to restore the PFS, the special restore , which will not create temporary checkpoint and not allocate any other chunk, is launched. Tracking Number
426987
2
431722/ 42744966
1
398615 / 38346418
EMC Celerra Network Server Version 6.0.70.4 Release Notes
137
Version 6.0.55.3
1
A queued ReplicationV2 session checkpoint operation (create/delete/refresh/restore) on the destination experienced long delays when the SavVol needed to be extended in order for the transfer to complete. DART panicked with the following panic header: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 820db2 Err code: 0 Target addr: 0 ** DART panicked with the following panic header: DART panic/fault message
The code has been changed to allow extends to continue, And, to reject any queued checkpoint operation requests.
416453 / 40756596
1
The code has been fixed to avoid the DART panic. 421194 / 41485422
1
The code has been fixed to avoid the DART panic.
439878 / 43748484
UFS
The following entries describe fixed UNIX File System issues:
Severity 1 Symptom Description The Data Mover panicked due to an out of memory condition under a high work load and with many active file systems. The condition was caused by Cylinder Group (CG) buffers being held in-use and remaining indefinitely memory resident. The following message was displayed “>>PANIC in file: ../addrspac.cxx at line: 289 : Out of memory” The Data Mover panicked with the following message: DART panic/fault message: >>PANIC in file: ../ufs.hxx at line: 1377 : UFS hashRemove: not on hash chain Data Mover panicked with DART panic/fault message: Assertion failure: 'this->back == this' on line 551 in file ”../vfsnode.cxx” >>PANIC in file: ../assert.c at line: 27 : assert failed Fix Summary Tracking Number
The CG caching algorithm code was changed to more efficiently handle the aging of CG buffers. The code has been modified to avoid the panic
394853/ 37939478
1
421397 / 41500892
1
Fixed the race condition on the mount point when file system was being unmounted and another thread is still referring the mount point.
434499 / 43149914
138
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
1
Watchdog panicked while enumerating file-system stats using server_stats command.
The code has been modified to avoid the watchdog panic. The code has been fixed to release this lock in the error path. The code was changed to correct invalid free block recoveries at the time of Data Mover recovery, avoiding further Data Mover panics.
453141 / 43990760
1
When the system is short of memory, a rwlock may not be released in the error path if the I/O is in the uncached IO path. This causes NFS threads to hang. The Data Mover panicked and reported that a file system requires an File System Consistency Check (FSCK). The panic may report a corrupted directory entry or an invalid indirect block.
461028 / 44875620
1
466190 / 45446328
Control Station
The following entries describe fixed problems related to the Celerra Control Station:
Severity 1 Symptom Description Slow overall system performance was observed with nas_pool -query being issued from JServer. The poor performance could be seen in several different ways including RDF continuously failing over, slow GUI response times, certain task starvations, etc “dbchk –p” command failed when executing "symmir list", and gave out error “Error: File system with id /nas/symcli/bin/symmir list is found in .server_config mountdisplay's output but does not exits in /nas/server/{1}/mount file.” The error doesn't describe the actual problem. Fix Summary The code now provides a number of performance optimizations to improve performance. Tracking Number 422395 / 41595948
2
The error message was replaced with ”The exit status is neither “0” nor “1” while running command symmir list.”
454984 / 44202128
EMC Celerra Network Server Version 6.0.70.4 Release Notes
139
Version 6.0.55.3
Network
The following entries describe fixed Network-related problems:
Severity 1 Symptom Description BCM 5715 1GigE NIC ports intermittently showed link bounce. Fix Summary Changed the default behavior of the BCM NIC recovery mechanism to not reset the 5715. Tracking Number 432345 / 42831902
System Management
The following entries describe fixed issues related to system management:
Severity 2 Symptom Description When control station is in BRST timezone, if schedule fire-off time falls into the transition time of BRST to BRT, scheduler operations get hung. Fix Summary The code has been fixed. Scheduler no longer hangs. Tracking Number 436472 / 43375724
Quotas
The following entries describe fixed issues related to quotas:
Severity 1 Symptom Description When a CIFS client issues quota on while a online quota check task is running, the Data Mover panicked with: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x000083f4bc Err code: 2 Target addr: 0x0000000000 ** While logging a server log message indicating that an online check was already in progress, the code was dereferencing a null pointer Fix Summary The condition is now handled correctly. Tracking Number 453514 / 44008566
140
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Version 6.0.51.6
Celerra Management Applications
The following are fixed problems related to the Celerra Management Applications, that is, Celerra Monitor, Celerra Manager, WebUI, or Celerra MMC Snap-ins:
Severity 2 Symptom Description Following a Data Mover failover, VDM CIFS servers were not listed in Unisphere. They were visible through the CLI. Persistent failures of APL tasks over time, due to faulty logic validating UPS configuration (expecting 2 UPS configuration when only one exists) resulted in accumulation of large amount of tasks related files in /nas/tasks directory. Eventually 100% of file descriptors were exhausted. NBSnas becomes non writable, and nas services will fail to start. In Unisphere, when managing the Usermapper configuration and uploading "User Mapping File" or "Group Mapping File", or both with the "replace" option caused one of the files contents to be lost. As the result Windows users may lose access to their files. Automatic DHSM connection creation and connection verification for Rainfinity FMA using XML API failing due to case sensitivity of share queries. Users could not query the location if they didn't specify the share names in the actual case. Fix Summary CIFS servers created on a VDM are now visible in Unisphere after a Data Mover failover/failback. The logic has been corrected and these APL task failures will no longer occur. Tracking Number 406031 / 39311966
3
411571 / 40109354
2
Modified Mapping Files processing algorithm to ensure that using the "replace" option does not cause loss of the contents.
412850
2
The share name comparison have been changed to case insensitive.
422160
EMC Celerra Network Server Version 6.0.70.4 Release Notes
141
Version 6.0.51.6
Severity 1
Symptom Description When there was a large number of objects in user and group quota pages, user experienced delays in screen painting. The screen refreshes itself several times, and each refresh causes significant delay which appears like a hang to the user. User and group names were not displayed on the user and group quota page. Instead user and group IDs were displayed. Jserver reported incorrect port WWN for SP ports when a setup has more than 8 FC FE ports per backend. Creation of a new notification in Unisphere fails when e-mail address for notification contains '&' character. The error messages says: "Celerra Monitoring Agent could not be contacted". When using NTP QFS software with Celerra, Celerra Manager login page failed to load. That problem disappeared after a Control Station reboot, however returns back after several days. This was due to QFS software being disconnected abruptly and leaving HTTP socket open. Unisphere showed only one event starting 31st October 2011. The alert_log file was checked, it was normal and contained multiple entries. The Schedule Name column was not visible in the view even if the SnapSure license was enabled in Unisphere.
Fix Summary Fix involves some design changes in the way tables are built and updated.
Tracking Number 396840 / 37832254
2
Changed the APL to support the correct schema reporting of the SIDs value for user and group quotas. The fix is to hash ports info based upon the entire port WWN instead of just the "port number". The code now properly handles special characters, like '&', '<', '>', etc. when included in the e-mail address for notification. Improved code that handles abruptly disconnected XML API client. Introduced a limit on the number of outstanding XML API requests.
397447
2
420482 / 40714894
2
432113 / 42573498
2
431655 / 42590566
2
The APL filtering error that was causing this issue has been corrected. Changed the string concatenation operations to ensure that when "snapsure" license is enabled, user
454109 / 44118074
1
454360
142
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description
Fix Summary can see the Schedules column.
Tracking Number
CIFS
The following entries describe fixed Common Internet File System problems:
Severity 1 Symptom Description If a share path was being used by another share command while the path was being relocated, the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 359: Memory corrupted.. The stack trace showed:
0xe94ff740: 0x141278 waitForReboot+0x90 0xe94ff760: 0x141501 fault_dump+0x67 0xe94ff780: 0x1413e9 PANIC+0x29 0xe94ff790: 0x200731 free+0x6f 0xe94ff7c0: 0x200e34 _ZdlPv+0x10 0xe94ff7d0: 0x8aecce _ZN12volatilePathIcED0 Ev+0x2e 0xe94ff7e0: 0xc89ff2 _ZN14abstractStringD2E v+0x20 0xe94ff7f0: 0xc8a13d _ZN10pathStringD0Ev+0x 15
Fix Summary Code was modified to ensure that the share path is not being used or modified when the path is relocated.
Tracking Number 431284 /42686198
EMC Celerra Network Server Version 6.0.70.4 Release Notes
143
Version 6.0.51.6
Severity 1
Symptom Description When the virus checker service was restarted with new masks and exclusion rules, these changes were not applied to files that were open. This resulted in the virus checker scanning files with an excluded extension. When there were several open resources on a Domain Controller in trusted domain which no longer responded, each SMB thread entered a 20-second timeout, potentially creating a CIFS outage. During CIFS access outage, the server_log was logging "DOMAIN_CONTROLLER_NOT_F OUND" events. In addition to this, the Data Mover was sending MSRPC Close requests when the TCP connection was already dropped. When a Data Mover requested user’s group memberships, a Domain Controller, for some unknown reason, had not sent the primary group of a user. This resulted in access being denied to files that rely on group membership as the group membership was not stored in internal credential. File system outage occurred when enumerating CIFS quotas with slow mapping mechanisms. User lost access to his shares while the quota entries window was populating CIFS quotas window. When certain kinds of malformed CIFS/SMB or NFS packets were sent to a Data Mover’s CIFS/NFS server or that of a Celerra, the Data Mover panicked, resulting in a denial of service for legitimate users. One example of this was seen when Codenomicon's Defensics v3.12 (www.codenomicon.com)
Fix Summary Code now checks whether the virus checker configuration has changed before performing a scan of a file.
Tracking Number 435234 /43216602
1
The code has been fixed and Data Mover will no longer cause CIFS outage under such conditions. The Data Mover will no longer send unnecessary close requests when connection is already dropped.
372385 /35027576
1
The fix adds missing primary group to internal credential to prevent this condition.
407896 / 39421448
2
Code was modified to reduce the file system locking time when enumerating users' quotas from CIFS. Improved handling of invalid data being sent in fields within an NFS or an SMB2 request.
412335 /39230656
1
414090
144
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description vulnerability scanner was run against a Celerra Data Mover.
Fix Summary
Tracking Number
1
With widelinks enabled and many CIFS accesses to a share from one client, the Celerra CIFS server incorrectly returned STATUS_OBJECT_PATH_NOT_F OUND for a request for a valid path. This was due to a race condition where multiple threads were checking a variable to determine whether it was a widelinks connection.
The check for whether it is a widelinks connection is now done only once during tree connect or mount of a file system and not on each protocol request which was leading to the race condition.
414865 / 40421472
EMC Celerra Network Server Version 6.0.70.4 Release Notes
145
Version 6.0.51.6
Severity 1
Symptom Description The Registry.pol file is used to maintain certain GPO settings on a Windows domain controller. User had a (possibly corrupted) Registry.pol file where fields (in a single entry) were all zeroes, so the pointer to the data field was set to NULL. When the NULL data pointer was subsequently dereferenced, the Data Mover panicked with the following DART panic/fault message: ***
Page Fault Interrupt. Virt ADDRESS: 0x0000f76e6a Err code: 0 Target addr: 0x0000000001 **. The stack trace
Fix Summary The data length field value is now checked (to be greater than 0) before dereferencing the data pointer.
Tracking Number 422696 / 41650866
showed:
0760: 0xc1ad80 _ZN7DskDump9checkdumpEPKc+0 x7a 07a0: 0xc1af8e dump_return+0x4 07b0: 0x14c7d2 _ZN12Memory_Frame13setProte ctionEm19Memo ry_Prot 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x14162d handle_fault+0x1d 60d1b0: 0xf77f2d _ZN6SysVol19getRegistrySett ingsERK6Strin gRP8L 60d320: 0xf7beb0 _ZN6SysVol14getGpoSettingsE RK6StringRP12 gpo_s 60d770: 0xf52346 _ZN9DomainGpo18updateVdmGpo ServerERK6Str ingi+=
2
One of the Group Policy settings used by the Data Mover is the Maximum machine account password age. When this setting is defined or changed in the domain,
The code was changed so that the default setting (0) is applied to the applicable CIFS Servers, when the
418168
146
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description it is updated and used by each CIFS server. If the setting was set to not defined, the applicable CIFS servers were not updated with the new setting value until CIFS was restarted.
Fix Summary setting is changed to no longer defined in the domain.
Tracking Number
1
When an administrator tried to list open files the Data Mover panicked with the following DART panic/fault message: ***
Page Fault Interrupt. Virt ADDRESS: bc9a8e Err code: 0 Target addr: 24 **. The stack trace
Fixed the race condition that occurred when an enumeration of open files is being performed at the same time as an open file request is failing.
419382 / 41240528
showed:
0x2c3b91c: 0x14124c waitForReboot+0x90 0x2c3b93c: 0x1414d5 fault_dump+0x67 0x2c3b95c: 0x13a398 handle_fatal_fault+0xa 0x2c3b980: 0xb585cb _ZN8fid_info16getFileEnum3I nfoEP12_FILE_ INFO_3+ 0xee897970: 0xb58970 _ZN14smb_connection11srvsvc AuditERiP16cS rvsvcF 0xee897a10: 0xb5837d _ZN14smb_connection11srvsvc AuditERiP20cS rvsvcA 0xee897a30: 0xb581d2 _ZN13smb_streamCtx16srvsvcA uditParseEP20 cSrvsv 0xee897a70: 0xb57f8f _ZN10smb_server16srvsvcAudi tParseEP20cSr vsvcAu
2
The Data Mover allowed the user to zero out (truncate) a file even though he did not have any rights to do so.
If there is no write permission from the ACL, the open now fails and the file is not truncated.
422369 / 41403684
EMC Celerra Network Server Version 6.0.70.4 Release Notes
147
Version 6.0.51.6
Severity 1
Symptom Description If a file system became full, when user cleared the windows log file, the Data Mover panicked with the following DART panic/fault message: >>PANIC in file:
../sched.cxx at line: 154 SYSTEM WATCHDOG.
Fix Summary This issue occurred because the memory state of the log object was not correctly restored. This has been fixed, and the Data Mover will no longer panic under this condition.
Tracking Number 425818 / 42018528
The stack trace showed:
f3a0: 0x14dd29 system_watchdog+0x61 f3b0: 0x1407b3 t_nmi+0xc3 f468: 0xa653db _ZN11BasicVolume18completio nCallbackEP10 ccb_hea f4a8: 0xae10b1 _ZN14ScsiBus_Device12sim_fi nishIOEP14Scs i_IOReq f4b8: 0xae1348 _ZN14ScsiBus_Device10sim_fi nishEP14Scsi_ IOReque f4d8: 0xaede04 _Z14FCP_completeIOPv+0xf2 f528: 0xb4db26 cpd_osw_finish_srb+0xb0 f588: 0xb46b8d fc_exm_exchange_complete+0x fa
1
While using FileMover with many files (in this case, 873), an Arkivio server returned an invalid 0xFFFF value for File ID, which led to a Data Mover panic with the following DART panic/fault message:
>>PANIC in file: /disks/Archive_Area_nsg7/6_ 0_41_303/Dart/s erver/src/cifs/client/inclu de/cifsclient_n ode.hxx at line: 281 : getting size of an unopened file. The stack trace
This issue was fixed by changing the default value (for unitialized FID) from ushort 0xFFFF to ulong32 0xCC11FFFF.
426242 / 42082680
showed:
f21400: 0x141439e _ZN17DHSM_CIFS_RemoteP15rem
148
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description
oteBlockReadEjR 11 f21590: 0x13f374a _ZN10DHSM_Vnode15remoteBloc kReadEjR11File_ St f21610: 0x13c7c67 _ZN10MGFS_Vnode14migrateToB lockEjRiPP13Buf fe f21890: 0x13ed849 _ZN10DHSM_Vnode14migrateToB lockEjRiPP13Buf fe f218e0: 0x141169b _ZN10DHSM_Vnode10blocksRead EjjRiPP13Buffer _C f219d0: 0x13f1c94 _ZN10DHSM_Vnode9blockReadEj R11File_Status+ 0x f21a00: 0x7bc897 _ZN9File_Fork17readBlockInt ernalEjRP13Buff er_ f21a40: 0x7c70dc _ZN13File_OpenFile16readDat aInternalEyjR11 Fil
Fix Summary
Tracking Number
1
A page fault PANIC might occur when changing quota state to different values (ON/OFF) from both Windows and from Control Station side at the same time for the same file system. When an SMB2 client opened a file under a Quota Tree, then disconnected just before the file system unmount processing started, a file system unmount did not complete, which led to a loss of access.
A lock on quota objects was added to prevent this condition.
426890 / 42139428
1
the unmount processing was modified to properly clean up SMB2 durable handles on a Quota Tree.
427655 / 42232932
EMC Celerra Network Server Version 6.0.70.4 Release Notes
149
Version 6.0.51.6
Severity 2
Symptom Description CIFS CEPP events were not generated or incorrectly generated. This caused some third party applications to not work properly. When user applied a strong security policy to his W2K8R2 CIFS domain, FileMover, CAVA or CEPA on this Microsoft machine may fail with ACCESS_DENIED errors due to communication errors with the domain controller. When CIFS client application used file range lock operations intensively with conflicts between different clients, a race condition led to a Data Mover panic with the following DART panic/fault message: *** Page Fault
Interrupt. Virt ADDRESS: 0 Err code: 0 Target addr: 0 **.
Fix Summary Code was modified to better manage CIFS CEPP events. Code now uses the correct CIFS-based SPN (service principal name) to access the domain controller.
Tracking Number 386670
2
425859 / 41925924
1
This race condition between range lock and close operations has been eliminated.
429671 / 42298318
The stack trace showed:
0x2c3bbd0: 0x14124c waitForReboot+0x90 0x2c3bbf0: 0x1414d5 fault_dump+0x67 0x2c3bc10: 0x13a398 handle_fatal_fault+0xa 0x2c3bc50: 0x768586 _ZN14File_LocksData6umountE y+0xc0 0x27de1c30: 0x7681ef _Z12lockd_umounty+0x123 0x27de1c70: 0x72dafe _ZN15unmountCallBack7execut eEy+0x20 0x27de1c90: 0x72e8a3 _ZN14VFS_FileSystem10unmoun tNowEi+0x4d 0x27de1cb0: 0x72ebb5 _ZN14VFS_FileSystem9doUnmou ntEv+0x109
1
Concurrent command to start usermapper, particularly when the database was disabled, caused a Data Mover panic with the following
Modified code to protect Usermapper database initialization from concurrent start.
430768 / 42669876
150
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description DART panic/fault message: *** GP exception. Virt ADDRESS: 0x00012c5c2a. Err code: 0 *** . The stack trace showed:
6760: 0xbf96c0 _ZN7DskDump9checkdumpEPKc+0 x7a 67a0: 0xbf98ce dump_return+0x4 67b0: 0x14c7d2 _ZN12Memory_Frame13setProte ctionEm19Memory _Prot 67d0: 0x14ce5d fault_dump+0x89 67f0: 0x13378d handle_fault+0x1d a57b0: 0x12323c3 __db_c_close+0x35 a57f0: 0x123271d __db_c_close+0x38f a5840: 0x1233169 __db_c_get+0x7a3
Fix Summary
Tracking Number
2
Whenever a new user or group quota was created, it didn't show up in the appropriate Unisphere quota list until an internal refresh. Unable to access CIFS share when trying to rename a file that was waiting to be checked by virus checker and all VC servers were offline. New users could not be mapped via LDAP.
Code was fixed to ensure new user or group quotas are displayed when created. Code now checks if the client has disconnected before renaming a file using CIFS. If a current LDAP configuration is deleted and a new configuration is created, CIFS code now automatically switches to the new LDAP setup.
378714
1
414744 / 40668160
2
414760 / 40449194
EMC Celerra Network Server Version 6.0.70.4 Release Notes
151
Version 6.0.51.6
Severity 1
Symptom Description During a share enumeration, if the share name contained 118N characters that generated a NULL hash key, renaming the share could cause a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x00011ef9c4 Err code: 0 Target addr: 0x0000000000 **. The stack trace showed: 0760: 0xc1ac70 _ZN7DskDump9checkdumpEPKc+0x7a 07a0: 0xc1ae7e dump_return+0x4 07b0: 0x14c7d2 _ZN12Memory_Frame13setProtectionEm1 9Memory_Prot 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x14162d handle_fault+0x1d cbb7d0: 0xc3e590 _ZN11smb_sdShare8getDBKeyEPKt+0x1a cbb820: 0xc37564 _ZN11smb_sdShare8getDBKeyEv+0x54 cbb840: 0xc377f3 _ZN9shareList9getLockedERP9smb_share PKcP9xdrS Samba clients were unable to use Kerberos authentication. The following messages were found in server_log: SMB: 3:
SSXAK=RPC_NT_INVALID_BINDIN G Client=a.b.c.d origin=550 stat=0x40000,0
Fix Summary
Tracking Number
Shares having a NULL hash key are now handled correctly, and the Data Mover will no longer panic under this condition.
418870 / 41139864
2
Code has been fixed to enable support for CKSUMTYPE_HMAC_ MD5 checksum type.
427779 / 42247186
They indicate that Kerberos checksum type cannot be checked due to usage of CKSUMTYPE_HMAC_MD5 checksum type instead of regular Windows client's CKSUMTYPE_KG_CB=0x8003. As a result, Kerberos authentication cannot complete, but client falls back to NTLM authentication seamlessly.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
152
Version 6.0.51.6
Severity 1
Symptom Description Memory was not correctly released after a DIC sending error. This memory leak led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../bufpool.cxx at line: 1309 : couldn't get a free page. The stack trace showed: 08e90: 0x1ff58e _ZN18Buffer_Cached_Pool17makeCached BuffersEm+0 08ee0: 0x14497f _ZN19Memory_AddressSpace16makeCac hedBufferEv+0 08f10: 0x1ff631 _ZN18Buffer_Cached_Pool12findAfreeBuf Emiii+0x6 08f50: 0x1ff6b9 _ZN18Buffer_Cached_Pool15getAnyCach edBufEmiii+ 08fa0: 0xa8f082 _ZN17UnCachedIO_Object9readBlockEyj1 6BlockIO_P 09060: 0x8ba443 _ZN14UFS_FileSystem9readBlockEiiRP13 Buffer_Cac 090c0: 0x8b4550 _ZN9UFS_Vnode5breadEjR11File_Statusi +0x352 09130: 0x92a676 _ZN9UFS_Vnode9blkatoff2EyRPvRP13Bu ffer_Cachedi
Fix Summary Memory allocated for the issue is now released once it has been used.
Tracking Number
432293 / 42848676
EMC Celerra Network Server Version 6.0.70.4 Release Notes
153
Version 6.0.51.6
Control Station
The following entries describe fixed problems related to the Celerra Control Station:
Severity 2 Symptom Description Administrators with a space in their ID could not process Join/Unjoin requests. Celerra replaced spaces with an asterisk which rendered the admin ID invalid Following a NAS code upgrade, when trying to query the status of SNMP on the Control Station, the system was returning the following error:
nasadmin]# /sbin/service snmpd status snmpd dead but subsys locked /sbin/service snmpd stop Stopping snmpd: [FAILED]
Fix Summary Corrected code to ensure that spaces embedded inside 'admin' are handled correctly. Code has been fixed to save the old config files and properly update the new config files during upgrade.
Tracking Number 406140
2
409494 / 39688446
1
After a power outage the CIFS servers came back as NT4 type instead of Windows 2000 type.
Code has been fixed to check for an existing server entry with the same name as the new server entry. If found, the existing entry and the new entry get merged to create a new entry. If the new and existing entries differ in server type, or domain, or netbios (W2K only), code now replaces the new entry with the existing one. The Control Station no longer ignores these event dispatch
381939 / 36515374
1
The Control Station ignored the event dispatch policies of the CLARiiON, causing a
415637 / 39816782
154
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description very high amount of call homes which led to the flooding of the call home utilities and servers.
Fix Summary policies.
Tracking Number
CORE
The following entries describe fixed problems related to Celerra Data Movers:
Severity 2 Symptom Description When issuing the warm reboot, if the Data Mover panicked for any reason, Data Mover failover occurred. Fix Summary Code has been fixed to ignore any unexpected events during warm reboot PreShutdown operation and not fail over. Tracking Number 403054
Deduplication
The following entries describe fixed issues related to Deduplication.
Severity 1 Symptom Description The deletion of deduplicated files while open transactions were in progress led to a deadlock situation where the UFS transaction log was unable to make progress. DART panic/fault message: >>PANIC in file: ../ufsvfsop.cxx at
line: 278 : sync: no progress reducing dirty list. The stack trace showed: 0x2752bf40: 0x141278 waitForReboot+0x90 0x2752bf60: 0x141501 fault_dump+0x67 0x2752bf80: 0x1413e9 PANIC+0x29 0x2752bf90: 0x7dcb25 _ZN21UFS_SyncThreadInitial5startEv +0x17f 0x2752bfe0: 0x145c26
Fix Summary Changed the way file system deletes compressed data produced by deduplication to prevent this panic.
Tracking Number 422705 / 41604572
Sthread_startThread_internal+0xe
EMC Celerra Network Server Version 6.0.70.4 Release Notes
155
Version 6.0.51.6
Severity 1
Symptom Description When using alternate data streams on a Deduplication-enabled file system, removing a deduped unnamed stream file caused the Data Mover to panic with the following DART panic/fault message: isAValidIndBlkBuf: bad bn. The stack trace showed: 0x21e47b40: 0x141280 waitForReboot+0x90 0x21e47b60: 0x141509 fault_dump+0x67 0x21e47b80: 0x1413f1 PANIC+0x29 0x21e47b90: 0x784dd7 _ZN14UFS_FileSystem28markCorruptedFs ForU nmount 0x21e47d50: 0x79b64a _ZN14UFS_FileSystem17isAValidIndBlkBuf EP 13Buff 0x21e47dd0: 0x7915c3 _ZN9UFS_Vnode17readIndirectBlockElRP 13Bu ffer_C 0x21e47e00: 0x791d25 _ZN9UFS_Vnode4bmapElRlPiS1_+0x237 0x21e47e60: 0x10366f0 _ZN10DHSM_Vnode14itruncInternalERP20 UFS_ Update User reported poor performance and high CPU utilization during a deduplication scan on file systems which contained thousands of hard linked files.
Fix Summary Code has been fixed to ensure blocks are flushed to disk properly before files are removed or truncated.
Tracking Number 432149 / 42837196
2
Replaced linked list with hash table to significantly improve scan times on file systems containing thousands of hard linked files. This new code will allow the migration inode to be reingested even if the UID/
424654 / 41006630
1
User was unable to read deduplicated files. The following error was displayed in the server_log: corrupt mignode for offline inode x. This can happen if the stub file and migration inode uid / gid is out of sync.
421654 / 41472302
156
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description
Fix Summary GID is out of sync so that stub file read operations are possible. It will also try to resync the uid / gid on file system update operations such as write, setLength, or recall.
Tracking Number
1
Compressing files caused their ctime to be updated which triggered unnecessary activity for incremental backup even though neither the data nor the file meta-data had been changed from the users perspective.
Code was modified to ensure deduplication and reduplication do not impact the ctime for either the internal Dart PE scan or for CIFS client compression / decompression.
424721
FileMover
The following entries describe fixed FileMover related problems:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
157
Version 6.0.51.6
Severity 1
Symptom Description While doing read operations using FileMover, threads hung and prevented CIFS access when the secondary storage became unavailable. Group quota usage is incorrect after recalling file from secondary file server.
Fix Summary During read operations, code was modified to require a shared lock only under this condition which will prevent hung threads.
Tracking Number 417842 / 41062586
1
This issue occurred because of a typo in DHSM feature that typed a GID as UID. This quota usage issue has been corrected.
440598
File Systems
The following entries describe fixed File System problems:
Severity 1 Symptom Description Write performance degraded substantially on a file system with checkpoints compared to a file system without checkpoints. Performance degradation was caused by inefficiencies in checkpoint handling code path. Fix Summary The code has been optimized to improve write performance on file systems with checkpoints. Code has been fixed to properly use the lock and prevent checkpoint from being falsely marked corrupted under this condition. This issue occurred Tracking Number 429333
1
Following an upgrade to version 6.0.42.0, when reading a deduplicated file for the first time and creating/refreshing a checkpoint, the checkpoint was marked as corrupted and unmounted. This was due to improper lock handling and creating a checkpoint with a dirty flag while there was a superblock update on the production file system.
429662 / 42529536
1
158
Migration between CFA and FMA was failing
382368 /
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description for a file system with Conversion file is different than an existing one error observed in the server log.
Fix Summary because the file system was replicated from an older system while a conversion was in progress. This caused the new FS to be marked as "conversion active", preventing the new conversion from failing. Code has been fixed to properly handle this condition and finish the conversion on the new destination automatically. The code has been fixed so that server_df is now able to display the correct number of free inodes regardless of the nbpi setting.
Tracking Number 36303236
2
The server_df server_x -inode command reported a negative a number of free inodes if the file system was created with lower than the default setting for nbpi (number of bytes per inode). This was because server_df was not able to report more than 2G free inodes.
376435 / 35689866
EMC Celerra Network Server Version 6.0.70.4 Release Notes
159
Version 6.0.51.6
Severity 1
Symptom Description When accessing a large directory (in this case 1.25 GB) and storing hash of the directory into memory, code did not properly release the memory which led to a Data Mover panic with the following DART panic/fault message: >PANIC in file: ../bufpool.cxx at line: 421 : couldn't get a free page. The stack trace showed: 0x79d0f8c0: 0x140ce4 waitForReboot+0x90 0x79d0f8e0: 0x140f69 fault_dump+0x67 0x79d0f900: 0x140e51 PANIC+0x29 0x79d0f910: 0x1c6af9 _ZN18Buffer_Cached_Pool17makeCached BuffersEi+0 0x79d0f940: 0x13c9c7 _ZN19Memory_AddressSpace16makeCach edBufferEv+0 0x79d0f960: 0x1c669c _ZN18Buffer_Cached_Pool15getAnyCache dBufEjii+0 0x79d0f990: 0x77e8eb _ZN14UFS_FileSystem8getBlockEliRPv+0x 49 When user tried to rename a file without write access into target directory, the rename failed with "Access denied". When this occurred, some CEPP events were generated twice for some NFS V3 operations. NFS lock was not preserved upon a Data Mover failover.
Fix Summary Code has been fixed to properly reclaim memory under this condition and prevent the panic.
Tracking Number 383034 / 36705796
2
Code was modified and will only send the CEPP PostEvents once.
491155
2
Code now preserves the statd registry when file systems are unmounted by administrator. If the client is 4.0 we return only 4.0 attributes even if 4.1 attributes are requested.
407035 / 39509656
1
Some version of AIX clients request NFSv4 4.1 attributes with a 4.0 mount. This caused AIX 6.1 host to crash when trying to mount a file system using NFSv4.
409247 / 39816612
160
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity 1
Symptom Description After running FSCK on a deduplicated file system, some files were inaccessible.
Fix Summary Improved FSCK handling of deduplicated files containing Alternate Data Streams (ADS). Code has been fixed to properly handle NFSv4 owner sequence and prevent this condition. The code has been optimized to improve write performance on file systems with checkpoints.
Tracking Number 410661 / 39762892
2
When using NFSv4, if a client sends a bad StateId this could result in NFSv4 errors, forcing the user to umount and remount NFSv4 file systems to get access again.
419732 / 40998286
1
Write performance degraded substantially on a file system with checkpoints compared to a file system without checkpoints. Performance degradation was caused by inefficiencies in checkpoint handling code path.
429333
EMC Celerra Network Server Version 6.0.70.4 Release Notes
161
Version 6.0.51.6
Severity 1
Symptom Description When using NDMP, a file was truncated when backing up the R/W file system on which the deduplicated file resided. The last coming data size exceeded the bytes in the last block and caused a Data Mover to panic with the following DART panic/fault message: Memcpy size 0xfffff28e itoo large. The stack trace showed: 909a68: 0x25147e memcpyPanic+0x42 909ac8: 0x282578 memcpy+0x54 909b40: 0x140009a _ZN19File_OpenDedupeFile11writeStreamE PhjRj+ 909bc0: 0xbf32d9 _ZN15PaxDedupHandler11WriteStreamEyP cjRj+0xaf 909c10: 0xbf469d _ZN14pax_cmd_thread13dedup_restoreEP 5ARCHDP15 909d50: 0xbafcc0 _ZN14pax_cmd_thread7extractEP5Event+0 x31a0 909f40: 0xbb1004 _ZN15restore_Initial8appStartEv+0x24a 909fb0: 0xb46723 _ZN15Uthread_Initial5startEv+0xf When using NMFS and concurrent CIFS and NDMP access, a race condition occurred if two sharedLocks on the same thread had an exclusiveLock request. Loss of access when attempting to generate a tree quota report while the tree quota directory was being deleted.
Fix Summary Code now compares the size of data it is trying to write with the remaining space of the migVnode (before EOF), and writes the smaller. If there's not enough remaining space, code returns EOF.
Tracking Number 413977 / 40557024
1
The fix eliminates the second sharedLock attempt on the thread. Changed the locking associated with looking up tree quota paths to eliminate a potential deadlock. The Data Mover now handles domain SID in Restricted Group correctly.
415992 / 40821538
1
416425 / 40875472
1
When Restricted Groups contained a domain SID, the Data Mover panicked when applying GPO that contained Restricted Groups. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS:
417912 / 41067196
162
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description 2020c9 Err code: 0 Target addr: 0 **. The stack trace showed: 0x2c4d92c: 0x14124c waitForReboot+0x90 0x2c4d94c: 0x1414d5 fault_dump+0x67 0x2c4d96c: 0x13a398 handle_fatal_fault+0xa 0x2c4d990: 0xc7c6fb _ZN17abstractConverter12strTranslateERP KcS1_+0x 0xe61fd9d0: 0xcbb8cc _ZN16localgroupServer11name2MemberEP 13smb_thre 0xe61fdaf0: 0xcbbd3c _ZN16localgroupServer13names2Members EP13smb_th 0xe61fdb20: 0xcbc99b _ZN16localgroupServer12addRGMembersE P13smb_thr 0xe61fdbd0: 0xcbc428 _ZN16localgroupServer19addRestrictedGro upsEv+0
Fix Summary
Tracking Number
2
Renaming a file from an NFSv4 client took an unnecessary lock on the parent directory as well as locking the file itself. This lock on the parent directory was not released when the rename was finished, and had to wait for 10 seconds before it timed out. When user/group quota was enabled on a file system, the default file system limits were not inherited by the user/group quota records.
The code now correctly only locks the file being renamed, and not its parent directory. User/group quota limits are now set correctly.
418716 / 40017698
2
420491 / 41347324
EMC Celerra Network Server Version 6.0.70.4 Release Notes
163
Version 6.0.51.6
Severity 3
Symptom Description Following an upgrade from Version 5.6 to 6.0 in a Linux environment, NFSv4 ACLs were not inherited from the parent and files created could not be used because the permissions were not correct.
Fix Summary When mode bits are set with NFSV4 in MIXED access policy, inheritance is preserved by a special ACL setting. Access right control on this processing was inconsistent with setAttr right checking. This has been fixed. When this occurs, the following message is now logged in the server_log: nfs4_checkSpecial A ce : SID <SID> cannot be translated to NFSV4 : ACL can't be modified. The memory leak issue has been fixed.
Tracking Number 424387 / 40438280
2
If an ACL for a file or directory contained an entry for CIFS-only well-known groups, NFSv4 users were unable to modify the ACL of the file. This is blocked intentionally, to prevent NFSv4 clients from replacing the CIFS-only ACE with an invalid entry. However, there was no indication why the user couldn’t change the ACL.
425723 / 41929606
1
When user changed ownership for millions of files, a memory leak condition accumulated to the point where the Data Mover panicked with the following message: >>PANIC in file: ../addrspac.cxx at line: 529 : Out of memory. The stack trace showed: 0x56c7fb80: 0x146dfe _ZN19Memory_AddressSpace10allocPages El12Memory 0x56c7fbc0: 0x8e54ff _ZN9UFS_Vnode15blockChainWriteEP10Bl ockChainiR 0x56c7fce0: 0x8e6183 _ZN9UFS_Vnode15blockChainWriteEiP10Bl ockChain+ 0x56c7fd50: 0x150dd10 _ZN10DHSM_Vnode15blockChainWriteEiP 10BlockCha
428118 / 42269136
164
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description 0x56c7fda0: 0x86b34c _ZN9VFS_Vnode18writeChainOfBlocksEP1 0BlockChai 0x56c7fde0: 0x7ed2c0 _ZN9File_Fork5flushERyii+0x1ac 0x56c7fe20: 0x7ed7ee _ZN9File_Fork16commit_cachedataEjji+0x1 58 0x56c7fe60: 0x7fc343 _ZN13File_OpenFile5closeEiii+0x55
Fix Summary
Tracking Number
1
User reported a higher than expected CPU usage during deduplication operations.
Eliminated unneeded debugging options which were found to be contributing to the CPU load during deduplication operations.
453242 / 43990760
EMC Celerra Network Server Version 6.0.70.4 Release Notes
165
Version 6.0.51.6
Severity 1
Symptom Description When using NFS V4, revoking an open state from an unconfirmed owner could double release the state, because in some cases the pointer to the (unreferenced) state was not reset to NULL. Since states are recycled, the release in excess caused a missing reference on a valid state. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 674dbb Err code: 0 Target addr: 0 **. The stack trace showed: 0x2c46464: 0x14124c waitForReboot+0x90 0x2c46484: 0x1414d5 fault_dump+0x67 0x2c464a4: 0x13a398 handle_fatal_fault+0xa 0x2c46580: 0x66e244 _ZN11NFSv4_Owner4openEmmP8nfs4_ctx P8stateid4RP 1 0xdffb9590: 0x6704e7 _ZN16NFSv4_StateTable13OpenClaimNull EmmP11NFSv 0xdffb9a10: 0x662b4b _Z9nfs4_openRP4msgbS1_RPhR8nfs4_ctx +0x8a9 0xdffb9d90: 0x65c870 _Z23nfs4_convertNfsv3Statusm+0x626 0xdffb9ee0: 0x63ebf1 _ZN17nfs_serverInitial5startEv+0xad7 When running an online quota check for a file system containing a quota record of anonymous ID, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x00008c7dc6 Err code: 2 Target addr: 0x000000005c **. The stack trace showed: 0760: 0xc1ac70 _ZN7DskDump9checkdumpEPKc+0x7a 07a0: 0xc1ae7e dump_return+0x4 07b0: 0x14c7d2
Fix Summary This issue was fixed, and the state pointer in the owner is reset to NULL before releasing the state.
Tracking Number 407115 / 39374052
1
Code was fixed to ensure the quota record of anonymous UID no longer causes this panic during online quota checks.
421173 / 41453586
166
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description _ZN12Memory_Frame13setProtectionEm19 Memory_Pro t 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x14162d handle_fault+0x1d f7a70: 0x8d72c6 _ZN14UFS_FileSystem14readQuotaDescsE 9quotaType f7ee0: 0x8f0f9c _ZN14UFS_FileSystem9oqcEnableEP20onli neQuotaCh f7f30: 0x8f1221 _ZN20onlineQuotaCheckTask10initializeEv+ 0x149
Fix Summary
Tracking Number
2
When using NFSv4, if a file system was unmounted, some Linux clients could not return their delegations. Delegation are correctly recalled. But the client try an OPEN_DELEGATE_CUR to convert its delegation to an open and then flush its cache. This fails because the file-system is frozen.
During the umount process, code now allows OPEN CLAIM_DELEGAT E_CUR from clients holding delegations to allow flushing the client cache Code was fixed to ensure delegation states are global to a client and no longer conflict with other states Code has been fixed to allow the nas_version command to work properly under this condition.
428074
2
When using NFSv4, client were unable to write to a file opened with a share reservation. This issue was due to I/O having write delegations conflict with share reservation
428236
1
When the NFS file system mounted on the Control Station became inaccessible, issuing the nas_version hung.
424679
EMC Celerra Network Server Version 6.0.70.4 Release Notes
167
Version 6.0.51.6
Installation/Configuration/Upgrades
The following entries describe fixed issues related to installations, configurations and upgrades:
Severity 1 Symptom Description During an upgrade, the nas_version command hung if __db.00* lock files under /var/sadm/pkg/emcnas were corrupted. Fix Summary Code now removes the lock files that are no longer in use, preventing the command from hanging due to prior corruption. With the fix, the upgrade will make sure that pre-reboot task is complete before continuing to the next step in upgrade. Code now correctly ascertains slot number when non-standard DM names are in use. A mechanism to control whether or not inventory data should be included into email notifications has been introduced. Tracking Number 404601 / 3988124
1
During NAS upgrade, NAS service was required to be stopped on Control Station. In some cases when NAS service failed to stop, and Control Station reboot was initiated, the upgrade failed after reboot on the next step because pre-reboot task had not completed successfully. The upgrade Blade parameters task failed during out-of-family upgrades from 5.6 to 6.0 if the system had nondefault host names configured for the Data Mover.
417369 / 40921968
2
414362
2
User wanted to remove a portion from their emailuser alerts. There was a "System Configuration Snapshot" at the bottom of the e-mail that the user did not want to include. It included a "/nas/sbin/model" command and a "nas_inventory -list". The primary goal was to remove the "nas_inventory" command output from these e-mails. When booting a Control Station with ExpressInstall CD/DVD, if user selected Install option, the process did not verify existing backend data, and overwrote it. This was inconsistent with non-ExpressInstall CD/DVD.
423814 / 41763194
1
Fixed the code so ExpressInstall CD/DVD work the same way as nonExpressInstall
421316 / 41453962
168
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description
Fix Summary CD/DVD. Fixed command to properly handle this error condition and no longer report these errors.
Tracking Number
2
When using CUT to upgrade to version 6.0.50, nas_version command displayed the following error: $ nas_version /nas/bin/nas_version: line 21: lsof: command not found rm: cannot remove `/var/sadm/pkg/emcnas/__db.001': Permission denied rm: cannot remove `/var/sadm/pkg/emcnas/__db.002': Permission denied rm: cannot remove `/var/sadm/pkg/emcnas/__db.003': Permission denied Celerra was not picking up failed drive events
421334
1
The file that indicates events to the Celerra operating environment software was updated so that drive failed/removed messages are handled correctly.
432919
NDMP
The following entries describe fixed Network Data Management Protocol/Backup problems:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
169
Version 6.0.51.6
Severity 1
Symptom Description When running NDMP, if the getinq command failed, it accepted the current value of a different buffer. If the different buffer had random data, it resulted in a malloc failure, which led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 515 : malloc() incorrect size - not a positive value. The stack trace showed: 0x4edce14d0: 0x25d13e malloc+0x5e 0x4edce1520: 0x1723c25 _Z18GetScsiDevices NameRP16scsi_devic e_namei+ 0x4edce17e0: 0x1724283 _ZN24NdmpdConfig GetTapeInfoMH14me ssageHandle 0x4edce1880: 0x172c669 _ZN14NdmpConnec tion15processMessag esEi+0x509 0x4edce1910: 0x172c84f _ZN14NdmpConnecti on15processRequest sEv+0x9 0x4edce1920: 0x16f836d _ZN21ConnectionFil eHandler11handleEve ntEv+0x 0x4edce1940: 0x172d38f
Fix Summary The NDMP code now checks the status of the getinq function and exits properly, if the getinq command fails.
Tracking Number 430903 / 42684894
170
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description _ZN19NdmpEventDis patcher8dispatchEij+0 x729 0x4edce1a80: 0x16fb29a _ZN21NdmpConnec tionHandler17connect ionHandle
Fix Summary
Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
171
Version 6.0.51.6
Severity 1
Symptom Description While performing a VBB restore, during file creation, the lookupPath() failed. If the func call did not find the path, it returned NULL. The Data Mover panicked with the following DART panic/fault message: *** DART panic/fault message: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 124a3c0 Err code: 0 Target addr: 0 ** The stack trace showed: 0x2c62e14: 0x141278 waitForReboot+0x 90 0x2c62e34: 0x141501 fault_dump+0x67 0x2c62e54: 0x13a398 handle_fatal_fau lt+0xa 0x2c62e90: 0x1246211 _ZN14vbbCreateFi les21createDesti nationPathEPcm 0x58545f20: 0x1245c3d _ZN14vbbCreateFi les14createBaseT reeEv+0x463 0x58545fa0: 0x1244d84 _ZN14vbbCreateFi
Fix Summary The panic occurred when VBB used the returned NULL pointer. The fix checks the return code and returns failure if the function call fails, instead of panicking the Data Mover.
Tracking Number 416911 / 40908030
172
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description les3runEv+0xde 0x58545fc0: 0x125422b _ZN24vbbProjectM ember_Initial5st artEv+0x1f 0x58545fe0: 0x145c26 Sthread_startThr ead_internal+0xe If the residual data and actual read data of a file during an NDMP restore did not agree, the restore failed because the data read from tape was not what was expected. ERROR MESSAGE: Tape Silvering restore with error of extent length not multiple of 8K. While doing a VBB DAR full restore, the attributes of directories were not reset to the values saved in the archive. This issue did not exist in non-DAR restore.
Fix Summary
Tracking Number
1
This tape read error was fixed. Code now ignores residual data if the read returns OK.
409887 / 39959654
2
The attributes of directories are now reset to the values saved in the archive.
411250
EMC Celerra Network Server Version 6.0.70.4 Release Notes
173
Version 6.0.51.6
Severity 1
Symptom Description When backing up a file system to tape, a media error caused the server_archive command to fail and the Data Mover to panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000ca204c Err code: 0 Target addr: 0x0000000004 **. The stack trace showed: 08e0: 0xce0520 _ZN7DskDump9chec kdumpEPKc+0x7a 0920: 0xce072e dump_return+0x4 0930: 0x14d9f2 record_panic_tic ks+0x242 0950: 0x14e07d fault_dump+0x89 0970: 0x141da5 handle_fault+0x1 d 6ebd30: 0xca26ee _ZN14pax_cmd_t hread9wr_rdfileE P5ARCHDiP11MsgL 6ebdd0: 0xc99f78 _ZN14pax_cmd_t hread10wr_archiv eEv+0x1f42 6ebf40: 0xc9afbb _ZN14backup_In itial8appStartEv +0x243 When using NDMP, during reverse block level object (BLO) recovery, the cmd
Fix Summary NULL pointer protection was enhanced to prevent this panic.
Tracking Number 422966
2
The reverse BLO (block level object) recovery code was modified to work
424158 / 41813810
174
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description service could not find the task context and logged the following error to the server log every 5 minutes filling up the server log : 2011-06-26 04:07:23: CMD: 3: handleTaskObject: INDICATION action TASK Context not found for Task = APM0011100197720 07_2_53:0 2011-0701 12:23:18: CMD: 3: handleTaskObject: INDICATION action TASK Context not found for Task = APM0011100197720 07_2_53:0 After using NDMP/PAX to backup deduplicated files, and then attempting to restore them on a new file system (or a different file system that did not have dedupe enabled), user saw the following errors in the server_log... 011-1028 18:23:44: DEDUPE: 3: builder: RDE not initialized fsid 178 2011-10-28 18:23:44: DEDUPE: 3: writeStream: failed to get ssp and none of the deduplicated files will be restored on that file system.
Fix Summary correctly and no longer log these errors to the server_log.
Tracking Number
1
Fixed the deduplication initialization for this case.
435544
EMC Celerra Network Server Version 6.0.70.4 Release Notes
175
Version 6.0.51.6
Severity 1
Symptom Description While using NDMP, user was unable to run 8 simultaneous server_archive sessions, but could only get 4 to run in parallel, even on 8gb memory systems. NDMP failed to restore files with error message SHADOW: 6: setName: Cannot translate xxx in unicode (5=InvalidArgument) but did not return an error to the NDMP server When user tried to do a DUMP backup and set file or directory filters, if the filters matched a symbolic link and the symbolic link could not be stat()-ed, the backup job failed. Here are some cases when stat() will fail: 1. destination of the symbolic link doesn't exist. 2. the depth of the symbolic link is higher than 8(File_MaxSymLinksI nPath).
Fix Summary Code was fixed to enable user to run 8 server_archive on 8gb memory systems.
Tracking Number 429341
1
Code now sends this warning message to FMA/DMA.
424257
2
Code was changed to correctly account for symbolic links.
423990
176
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Network
The following entries describe fixed Network-related problems
Severity 2 Symptom Description FTP behavior was not working as expected, the -a option on the ftp ls command did not return . and .. This caused an infinite recursive list loop. Eventually, the list command was truncated, resulting in 'command not understood'. Fix Summary Updated the FTP server in order to allow .and.. to be shown in ftp ls -a command This race condition has been eliminated. Tracking Number 412323 / 40150774
1
If the peer server disconnected while the HTTP client completed the initialization of the TCP connection, a race condition led to a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000213187 Err code: 0 Target addr: 0x0000000000 **. The stack trace showed: 0760: 0xc1ac70 _ZN7DskDump9checkdumpEPKc+0x7a 07a0: 0xc1ae7e dump_return+0x4 07b0: 0x14c7d2 _ZN12Memory_Frame13setProtectionE m19Memory_Pro t 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x14162d handle_fault+0x1d 280d30: 0x41c2c0 readmsg+0xe0 280db0: 0x41c324 GetsocknameStream+0xe 280dd0: 0x10c064a _ZN14http_clientCnx11_tcpConnectE v+0xd84
421919 / 41579878
EMC Celerra Network Server Version 6.0.70.4 Release Notes
177
Version 6.0.51.6
Severity 1
Symptom Description When there were lots of fragments (typically UDP traffic) and empty network buffers at the head of the fragment chain, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x00003ce5e0 Err code: 0 Target addr: 0x0000000000 ** . The stack trace showed: 0760: 0xc1ad80 _ZN7DskDump9checkdumpEPKc+0x7a 07a0: 0xc1af8e dump_return+0x4 07b0: 0x14c7d2 _ZN12Memory_Frame13setProtectionE m19Memory_Pro t 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x14162d handle_fault+0x1d ed90: 0x3c3897 ip_mxrput+0x10d edd0: 0x37d8cd _Z7bcmreadPvP4msgbtjj+0x4df eee0: 0x38e1c2 _Z20LM_ServiceInterruptsP16_LM_DE VICE_BLOCK+0x 3 If a user made an SNMP bulk request with a repeat value of "0", the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 378 : malloc() - incorrect size - not a positive value. The stack trace showed: 825ce0: 0x24a290 malloc+0x5c 825d20: 0x10e49ef _Z28netsnmp_create_subtree_cacheP 23netsnmp_a 825d70: 0x10e4eb0 _Z10handle_pduP23netsnmp_agent_se ssion_s+0xc 825da0: 0x10e5d3e _Z22netsnmp_handle_requestP23nets nmp_agent_s 825dd0: 0x10e67c6 _Z18handle_snmp_packetiP12snmp_se ssioniP8snm 825df0: 0x1116a2d _Z12snmpv3_parseP8snmp_pduPhPmPS1 _P12snmp_se 825e40: 0x111868f _Z10_sess_readPvP8LinkListR12Sele ctStream+0x 825eb0: 0x11187de _Z14snmp_sess_readPvP8LinkListR12 SelectStrea When a BCM NIC was hung, it was immediately
Fix Summary Code no longer panics when empty blocks are removed via MSG_TRIM().
Tracking Number 422980 / 41687604
1
Code now checks for non-null 'r' repeat values before allocating memory using 'r' value.
426298 / 42091892
1
178
Changes
434044
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description reset.
Fix Summary were made to allow specifying that the BCM NIC device not be reset when it appears to be hung. This only impacts some troubleshootin g scenarios. Code has been fixed to properly allocate/relea se the required amount of memory used by TFTP stack, and prevent this Data Mover panic.
Tracking Number
1
Heavy usage of the TFTP protocol on a Data Mover caused memory overflow and led to a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x000089863e Err code: 3 Target addr: 0x0000b42f72 **. The stack trace showed: 1760: 0xbffe60 _ZN7DskDump9checkdumpEPKc+0x7a 17a0: 0xc0006e dump_return+0x4 17b0: 0x14c7d2 _ZN12Memory_Frame13setProtectionE m19Memory_Pro t 17d0: 0x14ce5d fault_dump+0x89 17f0: 0x13378d handle_fault+0x1d 2bdd00: 0x8b9261 _ZN14UFS_CgBlkEntry11postFreeBlkE PvS0_+0x5d1
434697 / 41576516
EMC Celerra Network Server Version 6.0.70.4 Release Notes
179
Version 6.0.51.6
Severity 1
Symptom Description When using a Domain name in an export list for NFS client, if the DNS server was unavailable, a Data Mover panicked while the NFS client was trying to mount the export. DART panic message: *** Page Fault Interrupt. Virt ADDRESS: 0x00004a4c77 Err code: 0 Target addr: 0x0000000048 **. The stack trace showed: 0x36d0760: 0xc1ad80 _ZN7DskDump9checkdumpEPKc+0x7a 0x36d07a0: 0xc1af8e dump_return+0x4 0x36d07b0: 0x14c7d2 _ZN12Memory_Frame13setProtectionEm 19Memory_Prot 0x36d07d0: 0x14ce5d fault_dump+0x89 0x36d07f0: 0x14162d handle_fault+0x1d 0x834ecb0: 0x40db4e udp_input+0x9fa 0x834ee00: 0x3ce997 ip_input+0x1425 0x834eeb0: 0x3c3897 ip_mxrput+0x10d
Fix Summary If the DNS server is unavailable, code will no longer panic under this condition, but will post the following message to the server_log: Portmap session is already removed because of timeout.
Tracking Number 437344 / 43462506
180
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Replication
The following replication-related bugs that were fixed in this release:
Severity 2 Symptom Description User could not create a checkpoint for a nested mounted file system for NDMP backup Unisphere did not display current transfer information of a replication session if the size was greater than 2TB. Fix Summary The fix will allow checkpoint creation from nested mounted file system during NDMP backup. Unisphere now displays the transfer information of a replication session regardless of the size. Tracking Number 407727
2
413467 / 40439600
EMC Celerra Network Server Version 6.0.70.4 Release Notes
181
Version 6.0.51.6
1
If an ongoing replication copy was disturbed (for example: unmounting the base line checkpoint during a nas_copy session), the Data Mover incorrectly marked the copy as complete although there was an error during the copy. The copied file system remained inconsistent which caused the Data Mover to panic when the file system was accessed. DART/panic fault message: mangled directory entry. The stack trace showed: 0x109412b00: 0x89c445 _ZN14UFS_FileS ystem28markCorru ptedFsForU nmoun 0x109412c70: 0x928411 _ZN9UFS_Vnode6 dirbadEjPKc+0xb1 0x109412d10: 0x90cb57 _ZN20UFS_DirHa shInfo_dir3C1EP1 1Local_Vno de+0x 0x109412db0: 0x93e582 _ZN15UFS_DirOp s_dir311initDirH ashEP9UFS_ Vnode 0x109412df0: 0x933dc9 _ZN15UFS_DirOp s_dir37scandirEP 9UFS_Vnode PK18F 0x109412ec0:
Code now properly checks the status codes for local replication using the loopback interface. This prevents marking a replication session to be completed when one of the base line checkpoints is unmounted, and eliminates the panic.
418323 / 41087088
182
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
0x91b353 _ZN9UFS_Vnode7 scandirEPK18File _NameCompo nentR 0x109412ed0: 0x93083c _ZN9UFS_Vnode1 0lookupCompEPK18 File_NameC ompon 0x109413000: 0x13bd49d _ZN10MGFS_Vnod e10lookupCompEPK 18File_Nam eCom
EMC Celerra Network Server Version 6.0.70.4 Release Notes
183
Version 6.0.51.6
1
When a checkpoint was taken, if the memory was fragmented and there was a lot of DBMS activity, the Data Mover panicked with the following .DART panic/fault message:>>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory. The stack trace showed: 0x5a9a1c30: 0x145966 _ZN19Memory_Ad dressSpace10allo cPagesEl12 Memory 0x5a9a1c70: 0x24a34e malloc+0x11a 0x5a9a1cb0: 0x24ad2f realloc+0x81 0x5a9a1ce0: 0x1285087 __os_realloc+0x3 c 0x5a9a1d10: 0x1283f76 __memp_sync_in t+0x280 0x5a9a1de0: 0x1284822 __memp_sync+0xe6 0x5a9a1e10: 0x12902db __txn_checkpoin t+0x1be 0x5a9a1eb0: 0x1290555 __txn_checkpoi nt_pp+0xf9 Too many duplicated final response for
The fix consists of allocating all the memory when the environment is initialized.
421132 / 41400444
1
Added a defensive mechanism to
398657 / 38323670
184
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
DicEvent were queued in the cmd queue, and each of them consumed about 24KBytes which exhausted the Data Mover memory. DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 289 : Out of memory. The stack trace showed: 0xaae2dda0: 0x14124c waitForReboot+0x 90 0xaae2ddc0: 0x1414d5 fault_dump+0x67 0xaae2dde0: 0x1413bd PANIC+0x29 0xaae2ddf0: 0x13bef6 _ZN19Memory_Addr essSpace10allocP agesEi12Me mory 0xaae2de10: 0x10a0283 _ZN10DpBuf_Pool8 getChunkEiiP16Dp Buf_PoolMe mbe 0xaae2de30: 0x10a0386 _ZN10DpBuf_Pool8 allocateEv+0x5c 0xaae2de50: 0x107fef8 _ZN9DpRequestC1E 15DpRequestMode_ e17DpReque stC 0xaae2de70: 0x10b1776 _ZN21DpService_R eplicaInfo19proc essInfoPar all
eliminate final response duplication.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
185
Version 6.0.51.6
1
In a case where an error occurred during a tape silvering restore process, and the restore was aborted, a RedirectReads(FALS E) call in the code caused the file system to enter a freeze state and be set inactive. Restore from the internal ckpt then failed due to the inactive file system state. The replication delete operation failed to delete the mount option record in the database. Leftover replication file system mount option record may silently change the file system readwrite attribute after a reboot, even without replication on it.
The code was fixed by moving RedirectReads(FALS E) to the end of the tape silver restore, preventing this issue and allowing the restore to complete.
411982 / 39959654
1
Code now checks whether there is replication on the file system, and only saves the mount option record when necessary. In addition, replication code removes duplicated entries during replication start.
421209 / 41475440
Security
The following entries describe fixed security-related issues
Severity 2 Symptom Description When the LDAP Service was reconfigured with no bind DN or password specified, the LDAP connection (and associated 'ldap' thread) were 'orphaned' or no longer used by the LDAP Service. While the connection and thread are usually Fix Summary If the connections are not used (idle for a period of time), they will be disconnected by the LDAP Server. Tracking Number 426901 / 42075838
186
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description cleaned up when (idle) timed out at the LDAP Server, if an LDAP disconnected, and multiple reconfigurations were attempted, the number of connections rapidly grew, consuming memory, threads, and other resources. The following known security vulnerabilities existed prior to this release: Apache HTTP server (CVE-2011-3192) Bzip2 (CVE-20100405), DHCP (CVE-20110997) GnuC libraries ( CVE2011-1071 ) GnuTLS (CVE-20081948, CVE-20081949, CVE-20081950 ) JRE (multiple vulnerabilities) Libpng (CVE-20086218) LibXML (CVE-20092414 and CVE-20092416) Logwatch (CVE-20111018) OpenSSL (CVE-20113207, CVE-20113210, CVE-20110014),
Fix Summary
Tracking Number
2
Several packages were updated to address these known security vulnerabilities. JRE update includes Oracle Java SE Critical Patch Update June 2011.
392895 , 413785, 428931, 428932, 428933, 431357, 433301, 433470, 433475, 433477, 432273 / 42301928 42301296 42301876 42302062
EMC Celerra Network Server Version 6.0.70.4 Release Notes
187
Version 6.0.51.6
Storage
The following entries describe fixed storage-related issues
Severity 1 Symptom Description An environmental issue occurred with HBA 1, during a self-test, which led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG. The stack trace showed: cee0: 0x14dd29 system_watchdog+0x61 cef0: 0x132913 t_nmi+0xc3 cfd0: 0x155263 Sthread_startThread_internal+0 x7 During a FLARE upgrade, when the LUN trespassed from the primary to the secondary processor, the Data Mover panicked. DART panic/fault message: >>PANIC in
file: ../BVolumeIrp.cxx at line: 405 : IO failure on Vol:89, blkNo:42139696, SKey:0x54. The stack trace showed: fed0: 0xa9fad5 _ZN8SgIO_IRP9irpFailedE14BlockIO_ Statusy +0x477 ff20: 0xb3baa1 _ZN14CamDisk_Device8completeEP10c cb_head er+0x14 ff90: 0xabf9c4 _ZN8Scsi_Xpt16processCallbacksEv+ 0xea ffc0: 0xabfa3d _ZN18Scsi_serverInitial5startEv+0 xd ffd0: 0x155263 Sthread_startThread_internal+0x7
Fix Summary Code has been fixed to prevent this environmental issue from panicking the Data Mover.
Tracking Number 408642 / 39774566
1
Fixed a timing window where two threads could each detect that a trespass is required and attempted to trespass to the wrong SP. Code now locks the volume and tests whether the current SP has changed before switching paths and issuing the trespass request.
423378 / 41629776
188
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.43.1
Version 6.0.43.1
Deduplication
The following entries describe fixed issues related to Deduplication.
Severity 1 Symptom Description Following an upgrade to version 6.0.42.0, when reading a deduplicated file for the first time and creating/refreshing a checkpoint, the checkpoint was marked as corrupted and unmounted. This was due to improper lock handling and creating a checkpoint with a dirty flag while there was a superblock update on the production file system. Fix Summary Code has been fixed to properly use the lock and prevent checkpoint from being falsely marked corrupted under this condition. Tracking Number 429662 / 42529536
Version 6.0.43.0
File Systems
The following entries describe fixed issues related to File Systems:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
189
Version 6.0.42.3
1
Celerra Network Server version 6.0.42 added a check for invalid Alternate Data Stream names. When a client tried to query an invalid streamname, the operation was rejected as expected but in the process a lock was left on the file or directory. This led to other threads getting blocked if they needed to gain exclusive access to the object.
The checking for invalid streamnames has been removed.
393688
Version 6.0.42.3
Celerra Management Applications
The following are fixed problems related to the Celerra Management Applications, i.e., Celerra Monitor, Celerra Manager, WebUI, or Celerra MMC Snap-ins:
Severity 2 Symptom Description File system names containing a space could not display correct space information on GUI. The usage bar was empty, indicating there was no space used, even if the file system had consumed some space. When a quota crossed the soft limit and the grace period was unlimited, that quota property was Fix Summary Modified code to correctly return the correct space information for file systems with names containing a space. Tracking Number 35613 / 33731506
2
Code now translates the unlimited value to string "No limit".
401502 / 38626372
190
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
not displayed in the GUI. Code represented an unlimited grace period as -1, and translated that to a string "NoLimit". The code path for the above scenario did not translate the value to a string and it caused an exception trace.
CIFS and NFS
This release addresses a potential vulnerability which could, under certain conditions, be exploited to cause a Denial of Service (DOS) attack.
CIFS
The following entries describe fixed Common Internet File System problems:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
191
Version 6.0.42.3
1
When using emcacl to set the security descriptor on a symbolic link on a Data Mover with the param cifs.acl.extacl=0x28 and CEPA enabled, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: c5f858 Err code: 0 Target addr: 1c ** . The stack trace showed: 0x2c39cb4: 0x14124c waitForReboot+0x90 0x2c39cd4: 0x1414d5 fault_dump+0x67 0x2c39cf4: 0x13a398 handle_fatal_fault+0xa 0x2c39d40: 0xc5228f _ZN8AppLibNT5closeEP8 fid_infoiR11File_St atusP17 0xe4307ec0: 0xae5301 _ZN14smb1_threadCtx10 replyCloseEP4msgb+0 x289 0xe4307f40: 0xbe4b0d _ZN14smb1_threadCtx7p rocessEP4msgbi+0xcd f 0xe4307fa0: 0xbd92ec _ZN13smb_threadCtx5st artEv+0x6d0 0xe4307fe0: 0x145bfa Sthread_startThread_i nternal+0xe A 0% CPU idle condition might be observed on CIFS servers with many existing connections.
The Data Mover panics when trying to get the size of the file when setting the security descriptor if the file is a symbolic link. The code now checks if the target is a symbolic link before attempting to get the file size.
416553 / 40899086
1
Code no longer browses existing connections during authentication. Code was incorrectly using the netBIOS name instead of computer name. This has been fixed.
404202 / 38611382
1
If the computer name of the CIFS server differed from its NetBIOS name, an attempt to join a CIFS Server to a Domain from GUI failed with the error "compname <comp_name> does not exists in CIFS
393863
192
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
configuration. 2 When using Virus Checker with a file system mounted on an NMFS path, the path sent to the virus checker was incomplete. This resulted in the file not being scanned When a CIFS user connects to a share attached to an NMFS, code now puts the mount path of all the file systems mounted on this NMFS in memory to ensure all files are scanned. Code now forces a retrytimeout internally to at least 300ms for the heartbeat. 410012 / 39861830
1
CEPA service was intermittently disconnected when retrytimeout was set to a value less than 200ms. This issue occurred because the TCP heartbeat request time out is always 300ms or greater. CIFS clients saw CIFS request timeouts in environments with high latency and high utilization. If the TCP queue for a CIFS was full, typically because a client makes several parallel requests, or there is high network latency (such as with a WAN), then clients may see the Celerra appear to ignore some CIFS requests. If the queue remained nearly full, eventually clients could see a CIFS timeout. This could also happen with CIFS clients that were not acknowledging TCP data quickly, but making CIFS requests aggressively.
397281 / 37879634
2
When the CIFS TCP queue was full, new responses were placed at the head of the queue instead of at the end. If the TCP queue never fully emptied (because there was always data outstanding on the network), then some CIFS responses might never be transmitted, leading to timeouts. The problem is resolved by appending new replies to the end of the TCP queue instead of the beginning. Code now drops incomplete requests, thus avoiding blocked threads.
391540 / 36829842
1
When running clients which don't fully respect protocol, our CIFS server became unavailable after consuming all internal threads resources due to pending requests.
404372 / 39134760
EMC Celerra Network Server Version 6.0.70.4 Release Notes
193
Version 6.0.42.3
1
The Celerra CIFS client was unable to connect to CIFS servers using NTLMv2. This would mean that if Kerberos authentication was unavailable and NTLMv1 was disallowed, the Celerra would fail to open CIFS connections. This could cause failures for DHSM recalls/file access, for CDMS, FTP authentication with Active Directory, or CIFS authentication of users. During a CDMS migration between 2 Celerras, the migration hung. On the target Celerra, user was getting various errors indicating, that the source wouldn't deliver the data. On the source, user was getting the following error at the exact same time at the fault occurred: 2010-12-11 00:50:42: SMB: 3:[codan] Too many open policies on pipe=cMsrpc_Policy for client=10.194.116.100 (257/256). During the migration, the LSA function opens policies, but was failing to close them properly, eventually leading to the process hanging. If a security descriptor (SD) was not defined, a memory leak condition occurred in shareEnum502 which led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../bufpool.cxx at line: 1309 : couldn't get a free page. The stack trace showed: 0x11757f080: 0x241633 _ZN18Buffer_Cached_Po ol17makeCachedBuffe rsEm+ 0x11757f0d0: 0x14e514 _ZN19Memory_AddressSp ace16makeCachedBuff
Fixed a DNS comparison issue preventing CIFS clients from using NTLMv2 in Active Directory environments.
399722 / 38523654
1
Added a use function to close LSA or SAMR open handles and prevent this error.
398486
1
Prior to the fix, code was building a SD which was automatically deleted. With this fix, the SD is no longer deleted.
393688
194
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
erEv+ 0x11757f110: 0x2416ef _ZN18Buffer_Cached_Po ol12findAfreeBufEmi ii+0x 0x11757f150: 0x2417d7 _ZN18Buffer_Cached_Po ol15getAnyCachedBuf Emiii 0x11757f1a0: 0xcf5d8b _ZN17UnCachedIO_Objec t9readBlockEyj16Blo ckIO_ 0x11757f290: 0xa6fc23 _ZN14UFS_FileSystem9r eadBlockEiiRP13Buff er_Ca 0x11757f300: 0xa647fe _ZN9UFS_Vnode5breadEj R11File_Statusi+0x4 de 0x11757f370: 0xb135f0 _ZN9UFS_Vnode9blkatof f2EyRPvRP13Buffer_C ached 1 When using a Windows backup tool to backup Unix files and data, a Data Mover became unresponsive if extended ACLs are validated. Removed a condition where two shared locks were taken in sequence which led to a deadlock. 403587 / 38982154
EMC Celerra Network Server Version 6.0.70.4 Release Notes
195
Version 6.0.42.3
1
When an NFS connection using Kerberos for authentication failed, a function was called using the wrong variable as a parameter, resulting in a Data Mover panic when it tried to obtain data that did not exist. DART panic/fault message: *** GP exception. Virt ADDRESS: 0x0000732c16. Err code: 0 ***. The stack trace showed: 1760: 0xbf6470 _ZN7DskDump9checkdump EPKc+0x7a 17a0: 0xbf667e dump_return+0x4 17b0: 0x14c7d2 _ZN12Memory_Frame13se tProtectionEm19Memo ry_Prot 17d0: 0x14ce5d fault_dump+0x89 17f0: 0x13378d handle_fault+0x1d 38be40: 0x733648 _ZN10NfsGssInit12proc essReplyERP4msgb+0x 42 38be70: 0x733105 _ZN13NfsGssRequest8di spatchEP4msgbPvP17A ccess 38bf00: 0x6f443a _ZN17nfs_serverInitia l5startEv+0xa5c
Code now uses the correct variable in the function call.
409927 / 39986788
196
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
1
Data Mover panicked when a Windows 7 client connected to a CIFS share through a Riverbed WAN accelerator with the SMB2 protocol enabled. The Riverbed WAN accelerator sent an SMB2_IOCTRL request compounded with SMB2 commands which the Data Mover did not decode correctly. DART panic/fault message: >>PANIC in file: /disks/NSG_Maint8/6.0_ build _nfs_env/Dart/server/sr c/kernel/include/st ream.hxx at line: 1119 : datab::freeRef: db_ref <= 0. The stack trace showed: 245ea0: 0xdcc4fd _ZN14smb2_threadCtx7pro cessEP4msgbi+0xf57 245f20: 0xdee8d6 _ZN13smb_threadCtx5star tEv+0xc4c 245fd0: 0x155263 Sthread_startThread_int ernal+0x7
The Data Mover now properly handles compound SMB2 requests received in this manner.
411721 / 40195472
EMC Celerra Network Server Version 6.0.70.4 Release Notes
197
Version 6.0.42.3
1
During a GPO update, if the connection to the domain controller is lost, an error case was used that deleted an object without setting the associated pointer to NULL. At the end of the update, the pointer was checked and deleted if not NULL. This resulted in a double-free, which caused a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 525 : Memory corrupted. The stack trace showed: 06d720: 0x24aaaf free+0x79 06d760: 0x24ae7d _ZdlPv+0x9 06d770: 0xf2b7fc _ZN9DomainGpo18update VdmGpoServerERK6Str ingi+ 06dd40: 0xf2be4b _ZN9DomainGpo15update GpoServerERK6String i+0x4 06dd80: 0xf250ae _ZN4gpod5startEv+0x117e 06dfd0: 0x155263 Sthread_startThread_i nternal+0x7 When using third party VPN software to connect to a CIFS server from a client that had the same machine name as the domain name, NTLM authentication failed with LOGON_FAILURE error and users were not able to connect to CIFS shares. A Data Mover panicked when trying to connect to a share on a file system which was corrupted. DART panic/fault message: ** Page Fault
Code has been fixed to set the pointer to NULL after the object is deleted to prevent this panic.
409288 / 39850746
2
The code has been fixed to ignore client names that match the domain name of the CIFS server, allowing authentication to succeed.
407582 / 39550232
1
Code has been fixed to check the state of the file system before processing the share request, thus
401537 / 38704210
198
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Interrupt. Virt ADDRESS: 72735b Err code: 0 Target addr: 1e8 ** . The stack trace showed: 0x2c4c914: 0x14124c waitForReboot+0x90 0x2c4c934: 0x1414d5 fault_dump+0x67 0x2c4c954: 0x13a398 handle_fatal_fault+0xa 0x2c4c9f0: 0x83e738 _ZN26passthroughShadowDir ectory16lookupN ameLock 0x28a2fa40: 0x83f662 _ZN26passthroughShadowDir ectory23lookupN ameLoc 0x28a2fa70: 0x836609 _ZN26passthroughShadowDir ectory23interna lLooku 0x28a2fb00: 0x8255a6 _ZN19ShadowDirectoryBase1 5lookupComponen tEPtmP 0x28a2fb30: 0xc6b217 _Z15getnodefromRootP15File _NamingNodePKc RS0_+0 1 When there were many CIFS users accessing a Data Mover, the Data Mover might panic due to 64-bit higher memory footprint. DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory.
preventing this panic.
Code has been fixed to reduce the size of various CIFS objects to improve memory usage.
404117 / 39062050
EMC Celerra Network Server Version 6.0.70.4 Release Notes
199
Version 6.0.42.3
1
Under a heavy CIFS load with many antivirus threads, there were 94 Write I/Os in progress which caused a memory fragmentation issue led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory. The stack trace showed: 06b0: 0x145966 _ZN19Memory_AddressSp ace10allocPagesEl12 Memory_ 06f0: 0x207f2e _ZN20Buffer_Uncached_ Pool10addBuffersEji j+ 0xa60750: 0x2098fd _ZN20Buffer_Uncached_ Pool14getUncachedBu fE15Str 0790: 0x20ae0d allocb+0x67 07c0: 0x20bc47 pullupmsg+0x79 0810: 0x231545 _Z10msgdsize_cPP4msgb+0 x1a7 0860: 0x231770 dlpiWPut+0xea 08d0: 0x3d0276 ip_output+0x432 With the dfsutil CLI from windows 2008R2, user could not reset the SD of a DFS link.
Implemented changes to the NIC driver to reduce the odds of panicking due to memory fragmentation.
400441 / 38611382
2
Code has been fixed to properly reset SD on the DFS link using dfsutil.
403696
Control Station
The following entries describe fixed problems related to the Celerra Control Station.
200
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity 1
Symptom Description The MirrorView activate command caused Control Station to reboot twice, 1st time after adding NBS service on RDF7 to DM2, and 2nd time after adding NBS service on RDF10 to DM2. The command finally succeeded on the 3rd attempt. While using Unisphere and trying to enable user and group quotas in one step, only the user quota was enabled. The group quota enable operation failed. LDAP connection in Unisphere under Settings->Manage LDAP Domain fails with the error 13693354039: CST (Common Security Toolkit) library internal error.
Fix Summary Both RecoverPoint and MirrorView DR features have been fixed and prevents Control Station rebppts during the DR failover at remote NBS volumes setup.
Tracking Number 388624
1
Code hass been fixed to detect the enabling of both user and group quota and use the "-both" command line operation.
404960
2
Code has been fixed to provide more information in the recommended action when this error occurs so that the user can correct it and proceed.
401078 / 38622682
Deduplication
The following entries describe fixed issues pertaining to Celerra Deduplication:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
201
Version 6.0.42.3
Severity 2
Symptom Description If the remote file system ID was a substring of another file system ID, user could not take a nearcopy/farcopy backup snap. This issue occurred if the remote file system size query substring match returned more than one item. If the file system had quota enabled with filesize policy, and there's sparse recalled files in the Deduplication feature, then after an online filesystem quota check finished, the quota block usage report will be incorrect.
Fix Summary Changed a remote query from sub-string match to exact match. The online quota check scanned both stub file and migration file, so the file size was calculated twice. This fix makes the online quota check skip the filesize of stub file if it has a sparse recalled migration file. There is a maximum of 4TB of deduped data. Code will now log a more specific error information in the server_log. Code has been fixed to properly declare all the functions and prevent this condition.
Tracking Number 402256
2
406699
2
If the group quota usage exceeded 4TB, and F-RDE was enabled, migrateFile failed. Only an internal error was logged, there was no specific error information in server_log to help issue diagnosis.
403974 / 38996546
1
When performing NDMP restore of a deduped file system on a highly utilized Data Mover, a buffer overflowed and a subsequential Data Mover panicked with the following DART panic/fault message: *** GP exception. Virt ADDRESS: 0x0000964d88. Err code: 0 ***. The stack trace showed: 0760: 0xbf6490 _ZN7DskDump9checkdumpEPKc+0x7a 07a0: 0xbf669e dump_return+0x4 07b0: 0x14c7d2 _ZN12Memory_Frame13setProtectio nEm19Memo ry_Prot
401349 / 38640140
202
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity
Symptom Description 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x13378d handle_fault+0x1d b7a00: 0x13b584e _ZN10DHSM_Vnode30convertOffline BlockChai nToHo b7a80: 0x13faf8c _ZN19File_OpenDedupeFile10write HolesEjj+ 0x416 b7ad0: 0x13fc10e _ZN19File_OpenDedupeFile21handl eExtentListWri
Fix Summary
Tracking Number
2
Needless error messages similar to the following in the server log for Deduplication enabled filesystems being replicated. 201104-22 13:40:27: DEDUPE: 3: PE: PersistState() returned error: Filesystem is a read-only 2011-04-22 13:40:27: DEDUPE: 3: PE: WriteOperationBegin filesystem fsid 292 is read only. There were no dedupe operations initiated by the Celerra administrator at the DR site.
Code has been fixed to hide these informational messages
413576 / 39901452
FileMover
The following entries describe fixed FileMover related problems:
Severity 1 Symptom Description Incremental backup sometimes picked up unchanged FileMover (DHSM) files Fix Summary Corrected the way offline file size is calculated so it will not cause incremental backup to incorrectly backup unchanged data. Tracking Number 405361/ 37027684
File systems
The following entries describe fixed issues pertaining to File Systems.
Severity Symptom Description Fix Summary Tracking Number
203
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
1
The /nas/sbin/uc_config -verify command returned the incorrect translation settings for the client specified. This could lead to thinking the configuration is not configured correctly.
/nas/sbin/uc_con fig -verify now always gives the correct translation setting being used for the specified client. Code no longer logs a message when the retention period of a locked file in FLR-C file system matches the retention period that is passed in by requests.
387225 / 36947724
1
While using Networker 6.1 to backup the file system via a CIFS connection, the FLR log was flooded with a large amount of repeating log entries. The rapidly expanded FLR log contributed to extending backup times. Following is an example of the repeating FLR entries: <date> Extend Retention Period on worm committed file : Inode No = <Inode> : Generation No = <generation> : UID = <UID> : Gid = <GID> : File Mode = <mode> : Filesize = <size> : Old RP = <date> : New RP = <date> : Failed An NFSv4 state object was recycled prematurely (one reference missing) which caused a Data Mover to panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x000068cfbe Err code: 0 Target addr: 0x00000001c0 **.The stack trace showed: 6760: 0xbf96c0 _ZN7DskDump9checkdumpEPKc+0x7a 67a0: 0xbf98ce dump_return+0x4 67b0: 0x14c7d2 _ZN12Memory_Frame13setProtectio nEm19Memo ry_Prot 67d0: 0x14ce5d fault_dump+0x89 67f0: 0x13378d handle_fault+0x1d f7f9d0: 0x6a3257 _Z17NFSv4_lockExpiredP11NFSv4_S tate+0x15 f7f9f0: 0x6a32c2 _ZN17NFSv4_delegClient9sendBrea kEjiyy+0x 30 f7fa50: 0x874603 _ZN10Delegation9sendBreakEjyy+0 x5f3 A race condition occurred when some of the file system operations resulting in vnode closure occurred while the same file system was being unmounted on the Data Mover,
406428 / 39472638
1
The Data Mover now confirms the release of the object before allowing it to be used.
410524 / 40054076
1
Code has been fixed to properly release all the references on the
356875 / 33841178
204
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
resulting in a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../ufs.hxx at line: 1077 : UFS hashRemove: not on hash chain. The stack trace showed: 0xd1cedbb0: 0x13d890 waitForReboot+0x90 0xd1cedbd0: 0x13da8d fault_dump+0x67 0xd1cedbf0: 0x13d98d PANIC+0x29 0xd1cedc00: 0x61cd54 _ZN9UFS_Vnode4loseEv+0x8e 0xd1cedc20: 0x5f05f1 _ZN9VFS_Vnode7reclaimEv+0x81 0xd1cedc40: 0x5ee99e _ZN9VFS_Vnode11getNewVnodeEv+0x 40 0xd1cedc50: 0x5ee337 _ZN9VFS_VnodenwEmP14VFS_FileSys tem+0x9 0xd1cedc70: 0xb98871 _ZN15DHSM_FileSystem8newVnodeEm 8dirTypes +0x41 While unmounting a file system with a DFS root share on it, a race condition occurred which led to a Data Mover panic with the following DART panic/fault message: >>PANIC: I/O not progressing LastVol touched not Known. The stack trace showed: 0x802fe90: 0x14124c waitForReboot+0x90 0x802feb0: 0x1414d5 fault_dump+0x67 0x802fed0: 0x1413bd PANIC+0x29 0x802fee0: 0x8b3c56 _ZN9RawIO_IRP13checkProgressEv+ 0x11a 0x802ffd0: 0x8b3b32 _ZN17IrpProgressThread5startEv+ 0x6 0x802ffe0: 0x145bfa Sthread_startThread_internal+0x e
file system prior to the unmount, preventing this condition.
This race condition has been eliminated by locking the share only when we are sure it is unmounting file system.
383368 / 38343174
EMC Celerra Network Server Version 6.0.70.4 Release Notes
205
Version 6.0.42.3
2
Memory was consumed and not freed when a wildcard search (ex. dir *.txt) was run from an FTP session with no matches. This led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 289 : Out of memory. The stack trace showed: 0xe10d1c30: 0x14124c waitForReboot+0x90 0xe10d1c50: 0x1414d5 fault_dump+0x67 0xe10d1c70: 0x1413bd PANIC+0x29 0xe10d1c80: 0x13bef6 _ZN19Memory_AddressSpace10alloc PagesEi12 Memory 0xe10d1ca0: 0x1459f1 _ZN7SthreadC1EPcP10DQueueHeadIS _Eji+0x1e b 0xe10d1cc0: 0xfc4f42 _ZN21ckptBlockAllocatorThr6laun chEPKc+0x 6a 0xe10d1cf0: 0xf9f333 _ZN19PoolManagedDeltaVol14start Allocator Ev+0xa 0xe10d1d10: 0xf981a5 _ZN19PoolManagedDeltaVolC1EPKcS 1_P7LVolu meS3_m An NFS client issued two illegal cross file system rename requests to the Data Mover at the same time which led to a deadlock condition and loss of NFS access.
The memory was not correctly freed after each wildcard search within FTP. This has been fixed.
407209 / 39528390
1
Code was modified to reject the illegal request and return the NFS3ERR_XDEV error when it receives cross file system rename request. The fix is a protective measure which takes an exclusive mutex before revoking the locks held by a client. This will avoid the Data Mover panic, but users should upgrade the Linux clients’ kernel to
402586 / 38891086
1
Linux NFSv4 clients exposed to the issue described in
400183
https://bugzilla.redhat.com/show_bug. cgi?id=264721 present the same clientid
string to the Data Mover and therefore can be seen as a single client. This caused a race condition which led to a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 1ce857 Err code: 0 Target addr: c8 **. The stack trace showed:
EMC Celerra Network Server Version 6.0.70.4 Release Notes
206
Version 6.0.42.3
0x2c62c84: 0x141278 waitForReboot+0x90 0x2c62ca4: 0x141501 fault_dump+0x67 0x2c62cc4: 0x13a398 handle_fatal_fault+0xa 0x2c62d00: 0x67ca57 _ZN11NFSv4_State5closeEmP8state id4P8nfs4 _ctxRP4 0xe1059d20: 0x67b0d8 _ZN16NFSv4_StateTable5CloseEmP8 stateid4P 8nfs4_ 0xe1059d50: 0x669c7d _Z10nfs4_closeRP4msgbS1_RPhR8nf s4_ctx+0x 18b 0xe1059d90: 0x665e30 _Z23nfs4_convertNfsv3Statusm+0x 626 0xe1059ee0: 0x647bdd _ZN17nfs_serverInitial5startEv+ 0xad7 2 On write heavy workloads, a lot of the following messages "Run out of staging buffers" were observed indicating that the current static setting of 6 for agregateWork::numbuf was not sufficient during peak write loads.
the versions mentioned in the Bugzilla report to resolve the issue completely.
Introduced a new parameter that can be tuned to get rid of 'Run out of staging buffers' messages. This newly introduced parameter ufs.aggregateC ommitNbufs can be adjusted to handle peak write loads for iSCSI and uncached IO. The parameter values are as follows: 32 (default), 6 (minimum) and 256 (maximum). Refer to the Parameters Guide for more detailed information about this parameter.
409307
EMC Celerra Network Server Version 6.0.70.4 Release Notes
207
Version 6.0.42.3
1
After a file system corruption, FSCK discovered the corruption and fixed most of it by freeing the corrupted inodes. However, fsck could not properly handle some special file types (block type & char type) from the file system perspective. Further access to such file types caused the Data Mover to panic with the following DART panic/fault message: isAValidIndBlkBuf: bad bn. The stack trace showed: 0xdc2bb740: 0x14124c waitForReboot+0x90 0xdc2bb760: 0x1414d5 fault_dump+0x67 0xdc2bb780: 0x1413bd PANIC+0x29 0xdc2bb790: 0x778cb5 _ZN14UFS_FileSystem28markCorrup tedFsForU nmount 0xdc2bb950: 0x78f1aa _ZN14UFS_FileSystem17isAValidIn dBlkBufEP 13Buff 0xdc2bb9d0: 0x785233 _ZN9UFS_Vnode17readIndirectBloc kElRP13Bu ffer_C 0xdc2bba00: 0x7877ae _ZN9UFS_Vnode8bmapListEmmPmPi+0 x228 0xdc2bba70: 0x78a5e6 _ZN9UFS_Vnode15blocksReadAsyncE jiPP13Buf fer_Ca When file system had many corrupted directory entries and FSCK was running on this file system, the Data Mover panicked with the following DART panic/fault message: *** GP exception. Virt ADDRESS: 87db94. Err code: 0 ***. The stack trace showed: 0x2c3be94: 0x14124c waitForReboot+0x90 0x2c3beb4: 0x1414d5 fault_dump+0x67 0x2c3bed4: 0x13a398 handle_fatal_fault+0xa 0x2c3bff0: 0xe8155d31 _end+0xe495d509 0xe8151fc0: 0x87d471 _ZN12Notfoundlist6locateEmP6dir ValRbPPS_ +0x3f 0xe8152040: 0x87d471
FSCK code has been enhanced to properly handle block and char type of the files and prevent this panic.
404536 / 39140368
1
The FSCK code has been changed to handle large number of corrupted directory entries correctly.
391164 / 37582944
208
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
_ZN12Notfoundlist6locateEmP6dir ValRbPPS_ +0x3f 0xe81520c0: 0x87d471 _ZN12Notfoundlist6locateEmP6dir ValRbPPS_ +0x3f 0xe8152140: 0x87d471 _ZN12Notfoundlist6locateEmP6dir ValRbPPS_ +0x3f 1 An out-of-sequence redundant close operation for the same owner led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../include/stream.hxx at line: 989 : datab::freeRef: db_ref <= 0. The stack trace showed: 0xe00dbc30: 0x14124c waitForReboot+0x90 0xe00dbc50: 0x1414d5 fault_dump+0x67 0xe00dbc70: 0x1413bd PANIC+0x29 0xe00dbc80: 0x1be7cc _ZN5datab7freeRefEv+0x32 0xe00dbc90: 0x674ee3 _ZN11NFSv4_Owner9saveReplyERP4m sgb8nfsst at4P8n 0xe00dbcc0: 0x679376 _ZN11NFSv4_State5closeEmP8state id4P8nfs4 _ctxRP 0xe00dbd20: 0x6778fe _ZN16NFSv4_StateTable5CloseEmP8 stateid4P 8nfs4_ 0xe00dbd50: 0x6665e5 _Z10nfs4_closeRP4msgbS1_RPhR8nf s4_ctx+0x 18b Under heavy load, the quota usage value was incorrectly reported as being different between on-disk and in-memory. The error message in the server_log implied that the on-disk value was corrupted. 2011-01-24 18:11:47: 13157531648: UFS: 3:[SNXE-FS-02131] inode usage record is not consistent: on disk 2915057 in memory 2915058 type 1 id 32773 ouId 0 2011-01-24 19:28:15: 13157531648: UFS: 3:[SNXE-FS02131] Corrected the serialization of close operations to prevent this Data Mover panic. 401063 / 38621014
2
This error message will be printed only if there is an actual discrepancy between the ondisk and inmemory quota values.
402598 / 38854944
EMC Celerra Network Server Version 6.0.70.4 Release Notes
209
Version 6.0.42.3
1
During a FSCK of a file system, the Data Mover panicked if a block address contained the bad block flag rather than a real address. DART panic/fault message: >>PANIC in file: ../DeltaIO.cxx at line: 727 : DeltaIO_Read_CB::failed. The stack trace showed: 0xd939f430: 0x14124c waitForReboot+0x90 0xd939f450: 0x1414d5 fault_dump+0x67 0xd939f470: 0x1413bd PANIC+0x29 0xd939f480: 0xf9a361 _ZN15DeltaIO_Read_CB6failedE14B lockIO_St atus+0 0xd939f490: 0x8b3d33 _ZN9RawIO_IRP9irpFailedE14Block IO_Status y+0x21 0xd939f4b0: 0x8c8253 _ZN11HyperVolume17readBlockInte rnalEP10D iskIO_ 0xd939f510: 0xf99d69 _ZN15DeltaIO_Read_CB10doneCommo nEv+0x239 0xd939f570: 0xf99b1e _ZN15DeltaIO_Read_CB14getMappin gDoneEy+0 x82 A Data Mover panicked when a file system was mounted with the options: ntcredential and accesspolicy=NT or SECURE. DART panic/fault message: *** GP exception. Virt ADDRESS: 6d4599. Err code: 0 ***. The stack trace showed: 0x2c67d28: 0x141278 waitForReboot+0x90 0x2c67d48: 0x141501 fault_dump+0x67 0x2c67d68: 0x13a398 handle_fatal_fault+0xa 0x2c67e20: 0x6d440d _ZN15File_NamingNode8accessOKEm P17Access _Creden 0x34509e10: 0x64c1cd _ZN16worldAccess_List16printWor ldAccessE P9secF 0x34509ee0: 0x647afd _ZN17nfs_serverInitial5startEv+ 0xad7 0x34509fe0: 0x145c26 Sthread_startThread_internal+0x
FSCK will now skip the read when the block address has been flagged as bad.
393729 / 37862428
1
Code was performing an extra release on NT credential for UNIX in the case of file system mounted with ntcredential and accesspolicy=NT or SECURE. This has been fixed.
407971
210
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
e 1 When running FSCK on a corrupted file system, FSCK may cause memory corruption and a Data Mover may panic when FSCK tries to repair certain types of corruption. DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 351 : Memory corrupted. The stack trace showed: 0x23a81dc0: 0x14124c waitForReboot+0x90 0x23a81de0: 0x1414d5 fault_dump+0x67 0x23a81e00: 0x1413bd PANIC+0x29 0x23a81e10: 0x1fe8e7 free+0x69 0x23a81e40: 0x1fefb8 _ZdlPv+0x10 0x23a81e50: 0x1fefce _ZdaPv+0xc 0x23a81e60: 0x24880e _ZN13cMemoryBufferD1Ev+0x1e 0x23a81e70: 0x24a2fb _ZN11cMemBufListD1Ev+0x15 Server_2 hung and did not respond to commands, client did not have access to data. A failure to release all TCP streams as NFSv4 connections were being closed eventually resulted in there being no TCP streams available on the Data Mover to service commands or user I/O requests. For Redhat Linux, the nfs.v3xfersize transfer size was limited to 32K. The fix adds additional protection in FSCK code to avoid memory corruption. 399957 / 37582944
2
The fix ensures all TCP streams are released when NFS v4 clients disconnect.
412602 / 40332336
1
In order to have better NFS performance from Linux RedHat client running on VMware, the maximum nfs.v3xfersize is now tunable by using the param nfs v3xfersize=1310 72 command.
373951
EMC Celerra Network Server Version 6.0.70.4 Release Notes
211
Version 6.0.42.3
1
After a client rebooted, a Data Mover panicked with the following DART panic/fault message: Assertion failure: 'refLock.isExclusiveLocked()' on line 1691 in file '../nfsv4_state.cxx' >>PANIC in file: ../assert.c at line: 27 : assert failed. The stack trace showed: 0x11456ba70: 0x2a7d0b _assert+0x63 _ZN15NFSv4_ClientEnt7ConfirmEyP A8_KhR8nf s4_ct 0x11456bb10: 0x7dd53d _Z19nfs4_create_sessionRP4msgbS 1_RPhR8nf s4_ct 0x11456bc70: 0x792ec5 _Z9nfs4_linkRP4msgbS1_RPhR8nfs4 _ctx+0x16 87 0x11456be40: 0x81fea2 _ZN17nfs_serverInitial5startEv+ 0x1e40 0x11456bfd0: 0x16ad51 Sthread_startThread_internal+0x 7 An empty tree quota directory was removed by an unexport operation. This occurred if a name was specified when the export was created.
NFSv4.1create proper client mutex and prevents this condition.
403427
1
When processing an unexport, code now performs the remove operation only when the last component is a symbolic link and not a real directory. The fix allows early revocation of NFSv4 lock owners to free up resources, and avoid hang conditions. Corrected a deadlock that occurred when this param was set to zero and 1) server_export command was run, and 2)
388560 / 37329964
1
Under heavy NFSv4 load, NFSv4 clients may log "Remote I/O error" messages and responses from the Data Mover may be interrupted for 5-10 minutes. The network trace would show NFS4ERR_RESOURCE errors from the Data Mover. The error occurs because the code did not release NFSv4 lock owners correctly. When parameter mount.forceFullShowmount was set to 0, Data Mover MAC threads became blocked. GUI and CLI commands became unresponsive, and the following message was logged in the server_log periodically: ThreadsServicesSupervisor: Service:MAC Pool:MACExec BLOCKED for xxx seconds:
402884
1
403999 / 39088798
212
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Server operations may be impacted.
another process attempted to generate a list of FS mounts (such as the server_mount command) simultaneously. Added a threshold param cifs.acl.sdcompa ctthreshold. If a SD is larger than this parameter's value, code tries to compact it by removing redundant ACEs. The default is 16KB. This is to mitigate allocation problem in a fragmented memory. 404188 / 38935580
1
User set large security descriptors (SD) with redundant ACEs. When code tried to read those SDs after several days, it ran into a memory panic because the SD needed a large chunk of memory. Although there was enough memory, code could not grant memory allocation requests because of memory fragmentation. This caused the Data Mover to panic with the following DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory. The stack trace showed: 40bc10: 0x145966 _ZN19Memory_AddressSpace10alloc PagesEl12 Memory 40bc50: 0x24a34e malloc+0x11a 40bc90: 0x24ae1b _Znwm+0x1b 40bcb0: 0x24ae46 _Znam+0x4 40bcc0: 0xcf9e07 _Z11querySDInfoP13smb_threadCtx R11smb_Ou tMngt 40bdd0: 0xdd00e7 _ZN14smb2_threadCtx16replyQuery SDInfoEP4 msgb+ 40be70: 0xdd0bd0 _ZN14smb2_threadCtx14replyQuery InfoEP4ms gb+0x 40bea0: 0xdcbc33 _ZN14smb2_threadCtx7processEP4msg bi+0x68d Code was accepting non $DATA stream types, which are not compatible with NTFS. When trying to copy *.jpg files from a Celerra CIFS share to a local Window XP client, the following error was displayed: cannot copy xxx: The filename, directory name, or volume label syntax is incorrect.
2
Code no longer accepts non$DATA type ADS.
155357 / 31498524
EMC Celerra Network Server Version 6.0.70.4 Release Notes
213
Version 6.0.42.3
Install/Config/ Upgrades
The following entries describe fixed issues related to installations, configurations and upgrades.
Severity 2 Symptom Description Upgrade process might modify the value of some NDMP parameters as the maximum value of these parameters have been changed in newer releases. A new PUHC check is required to display a warning message regarding this modification. Fix Summary A new check has been included to warn user if any NDMP value would be deprecated during upgrade. With this fix, the /etc/nbs.conf file is updated at the same time as the connection setup. Any reboots after that, will include a successful nbs service restart. Code changes have been made to avoid the PAHC/PUHC failure when the hard drive in the Control Station is not in faulted condition. Tracking Number 394516
2
If a failure occurs during installation, or the system is rebooted at the beginning of the install process, the NBS service does not start properly at the next reboot.
401432 / 38569244
2
PAHCPUHC might fail with a hard drive failure error even when the hard drive is not in faulted condition.
------------Errors------------------Control Station: Check the hard drive in the control station Error HC_CS_14505083128: The 'smartctl -H' command failed. Action : Re-run "/usr/sbin/smartctl -H /dev/hda" to observe the output. Contact your system/network administrator to resolve the issue or escalate this issue through your support organization. -----------------------------------------
381917
The check was relying on the output of the smartctl command instead of the return value.
214
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity
Symptom Description
Fix Summary The fix is to stop audited for the duration of the upgrade, and restart the same with the pre-upgrade settings at the end of the upgrade.
Tracking Number 388554
1
Upgrade is seen to fail in the "Wait for NAS service to start" task. This occurred when a lot of access commands, to the file systems mounted on the Data Mover, were being issued from the Control Station while the upgrade was in progress.
NDMP
The following entries describe fixed Network Data Management Protocol/Backup problems:
Severity 1 Symptom Description VBB file level restore of offline/stub files did not work. When calling setAttr() for those files, VBB set file size which caused the file to be recalled and hence become online. Fix Summary Code was modified to set attributes without changing file size. The checkpoints created automatically by an NDMP backup are now removed Tracking Number 395232 / 37208224
2
Some NDMP backup automatic checkpoints were left behind and had to be removed manually. This occurred if the Celerra was very busy and the checkpoint creation completed after the timeout expired.
377919 / 35499712
Networking
The following entries describe fixed issues related to Networking.
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
215
Version 6.0.42.3
1
During a Data Mover reboot, it was possible that the SLIC did not power up and caused the Data Mover to fail bootup. This could lead to a data unavailability situation. Under rare circumstances, the Data Mover may not reply to all NULL requests from a client. This occurred if the transmit queue for the TCP session was full.
This was fixed by upgrading the firmware on the SLIC. The Data Mover will now reply to every NULL request, even if the transmit queue to that client is full. Code has been changed to remove the limitation of small listen queue. It is now unlimited and can be controlled with the param tcp.backlog. The fix adds support to remove stale IPv6 hosts from routing and neighbor caches and prevent this panic.
406453
1
404522 / 38052726
2
When having many simultaneous FTP connections to a Data Mover, some of the connections failed due to a small listen backlog queue. When this occurred, Data Mover logged the following error in the server log: Too many connections in port 21's listen queue.
405484 / 38895674
2
When using IPv6, the host cache table can monotonically grow as a result of expired host entries. These expired entries can make user route retrieval times exceed the watchdog timer and cause a Data Mover panic with the following DART panic/fault message:>>PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG. The stack trace showed: eee0: 0x14dd29 system_watchdog+0x61 eef0: 0x132913 t_nmi+0xc3 efd0: 0x155263 Sthread_startThread_internal+0x7
405168 / 39158304
NFS
The following entries describe fixed issues pertaining to Network File Systems.
216
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity 1
Symptom Description When a file system was close to 100% full, there was incorrect logic in the space allocation code which caused the Data Mover to allocate more space on the file system for a new writes than was available. This led to a Data Mover panic with DART panic/fault message: alloc failed: counters out of sync. The stack trace showed: 0xe1321840: 0x13fb94 waitForReboot+0x90 0xe1321860: 0x13fda9 fault_dump+0x67 0xe1321880: 0x13fc91 PANIC+0x29 0xe1321890: 0x7286c7 _ZN14UFS_FileSystem28markCorrup tedFsForU nmount 0xe1321a10: 0x737932 _ZN9UFS_Vnode5allocEP20UFS_Upda teDescrip torllj 0xe1321a50: 0x735cfc _ZN9UFS_Vnode10ballocListEP20UF S_UpdateD escrip 0xe1321af0: 0x73979a _ZN9UFS_Vnode15blockChainWriteE iP10Block Chaini 0xe1321bc0: 0x73910d _ZN9UFS_Vnode15blockChainWriteE iP10Block Chain+
Fix Summary The allocation logic code has been fixed to ensure this panic no longer occurs when a file system is close to full.
Tracking Number 159227
EMC Celerra Network Server Version 6.0.70.4 Release Notes
217
Version 6.0.42.3
Severity 1
Symptom Description While deleting a large file, an overflow in a file system data structure caused a memory corruption issue which resulted in a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000957f75 Err code: 2 Target addr: 0x1000000008 ** The stack trace showed: a760:0xbfb6b0 _ZN7DskDump9checkdumpEPKc+0x7a a7a0:0xbfb8be dump_return+0x4 a7b0:0x14c7d2 _ZN12Memory_Frame13setProtectio nEm19Memo ry_Prot a7d0:0x14ce5d fault_dump+0x89 a7f0:0x13378d handle_fault+0x1d 05610:0x886740 _ZN14UFS_FileSystem13getCgBlkEnt ryEj+0x10 05620:0x8a003a _ZN14UFS_FileSystem9hashallocEP 20UFS_Upd ateDes 05690:0x8a0335 _ZN14UFS_FileSystem8allocBlkEP2 0UFS_UpdateDesc When using NFS V4, a bug in the Linux clients made multiple clients look as a single client to the Data Mover. A race condition in this context caused the Data Mover to panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 1ce857 Err code: 0 Target addr: c8 **. The stack trace showed: 0x2c62c84: 0x141278 waitForReboot+0x90 0x2c62ca4: 0x141501 fault_dump+0x67 0x2c62cc4: 0x13a398 handle_fatal_fault+0xa 0x2c62d00: 0x67ca57 _ZN11NFSv4_State5closeEmP8state id4P8nfs4 _ctxRP4 0xe1059d20: 0x67b0d8 _ZN16NFSv4_StateTable5CloseEmP8
Fix Summary Code has been fixed to properly handle this condition and eliminate the panic.
Tracking Number 402066
1
This race condition has been eliminated.
411712
218
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity
Symptom Description stateid4P 8nfs4_ 0xe1059d50: 0x669c7d _Z10nfs4_closeRP4msgbS1_RPhR8nf s4_ctx+0x 18b 0xe1059d90: 0x665e30 _Z23nfs4_convertNfsv3Statusm+0x 626 0xe1059ee0: 0x647bdd _ZN17nfs_serverInitial5startEv+ 0xad7
Fix Summary
Tracking Number
1
NFS lock was not preserved upon a Data Mover failover.
Code now preserves the statd registry when file systems are unmounted by administrator.
407035
Replication
The following replication-related bugs that were fixed in this release:
Severity 2 Symptom Description If the Control Station timed out while waiting on a Replication Stop operation and retried, the operation hung and blocked further operations on the session. Fix Summary When the Data Mover receives a Stop request, it now checks whether the Replication Sessions is already in cleaning-up state and if so returns a TaskInProgress error. Tracking Number 404635
EMC Celerra Network Server Version 6.0.70.4 Release Notes
219
Version 6.0.42.3
Severity 1
Symptom Description When one checkpoint with the same snapVol ID as PFS Vol ID was inactivated, the PFS version was wrongly inactivated. User was unable to refresh or create any checkpoints on their main production file system. The following error was generated: Error: DpRequest_vs_InternalError.
Fix Summary The product version is now skipped when version for inactivated checkpoint is searched in inactivateVersi on. A flaw in index slot relocation process that can cause an inconsistency in checkpoint metadata has been fixed. The delete operation will not be accepted while the Replication session is in a clean-up state and the operation will fail with Abort in progress error to the user. A new mechanism was added to Data Mover code and Replication Manager code in order to handle slower response times under heavy load. Now Replication Manager can
Tracking Number 404437 / 39049562
1
Data Mover panicked with DART panic/fault message: PANIC in file:
../DeltaVol.cxx at line: 7585 : Value in TOC is different from the value in btree. After the failover, some
414584 / 40596048
checkpoints were inactive.
2
An attempt to delete a Replication V2 session hung if the command was issued while a failover operation was in progress.
402024
1
Under heavy load, in configurations with remote iSCSI Replication with many iSCSI LUNs and snaps, Replication Manager job may timeout due to the Data Mover not responding in time to iSCSI snap create request from Replication Manager.
396493
220
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity
Symptom Description
Fix Summary query the status on Data Mover and if Data Mover cannot response fast enough, RM will retry the operation until it succeeds or until configurable time out value is reached. This fix is available in Replication Manager version 5.3.3 or later. Code now waits for the completion of the umount before taking a reference on the checkpoint file system. Code now returns TASK_IN_PRO GRESS directly to the Control Station to prevent the memory and reference leak.
Tracking Number
1
When a SnapSure SavVol becomes full, the oldest checkpoint needs to be unmounted and inactivated in order to free space. If a refresh of the oldest checkpoint occurs during this process it can result a deadlock, blocked threads, or a DART panic with the fault message PANIC: I/O not progressing.
399556
2
While using Replicator V2, if the ACK from the Data Mover failed/delayed to reach the Control Station, the Control Station retried sending the modify request. There was a problem in the code which caused a memory leak and cmd reference leak. Hence, there was an inflight task running on the Data Mover.
403651
Security
The following security-related bugs that were fixed in this release:
EMC Celerra Network Server Version 6.0.70.4 Release Notes
221
Version 6.0.41.4
Severity 1
Symptom Description Netgroups that are nested and begin with a special character were not being parsed.
Fix Summary Nested netgroups that begin with special characters are now fully supported. This issue occurred because the resource file contained a white space followed by the value. White spaces are now removed from the resource file for all account type values.
Tracking Number 402867
2
After creating the LDAP or storage domain user, the account type was displaying as "local" instead of LDAP in the GUI.
406555
Storage
The following entries describe fixed storage-related issues:
Severity 1 Symptom Description When a NAS was installed at location where SAN hosts had in-use FAST pools, Solutions Enabler (SE) code incorrectly parsed some float numbers as integer in the FAST pool data area. This caused SE discover failure and NAS install failure if there were existing SAN hosts with FAST pools. Error IM_14505279983 : Failed to synchronize with new backend storage Fix Summary This issue was fixed in SE code by changing from integer parsing to float parsing for several values. Tracking Number 400965 / 38569244
Version 6.0.41.4
Network File System
The following entries describe fixed issues related to Network File Systems:
222
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.4
Severity 1
Symptom Description When a checkpoint of the replication destination file system became corrupted and user was accessing the destination file system while replication data transfer was in progress, the read requests were redirected to the corrupted checkpoint. This caused a Data Mover panic with the following DART panic/fault message: D2320489_0: DeltaIO_Read_CB failed due to blk 16 corruption. The stack trace showed: 0x21a53f50: 0x14124c waitForReboot+0x90 0x21a53f70: 0x1414d5 fault_dump+0x67 0x21a53f90: 0x1413bd PANIC+0x29 0x21a53fa0: 0xfc2c9b _ZN22CkptCorruptMarkRequest13ha ndleReque stEv+0 0x21a53fc0: 0xfc2b26 _ZN23CkptCorruptMarkerThread5star tEv+0x80 0x21a53fe0: 0x145bfa Sthread_startThread_internal+0x e
Fix Summary Code no longer redirects read requests to a corrupted checkpoint.
Tracking Number 402304
Replication
The following entries describe fixed Replication issues:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
223
Version 6.0.41.4
Severity 1
Symptom Description If certain operations were performed on multiple Replication V2 sessions at the same time, the destination file system became out of sync with the source if a transfer was inprogress when the operation was initiated. The operations that can cause this issue are failover, stop, start -reverse and delete (if the session wasn't previously stopped). When one of these operations caused a transfer to be interrupted, the destination file system was restored to the state before the transfer is started. Only one of these restores could run at a time so while one was still in progress others failed and logged a message in the sys_log similar to: Apr 1 15:00:00 2011:DART:REP:ERROR:8:Slot 2:::1301670000:Destination=100_CK200064 400 000_0000_200_CKM00100200000_0000(ali as=fs1 ), restoring a checkpoint failed: DpRequest_vs_VersionExist. In some cases during bootup, the code might incorrectly decide that not enough space was reserved in order to recover a checkpoint. This can cause the Data Mover to panic and the checkpoints for the file systems to be inactive. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: fde91e Err code: 0 Target addr: 58 **. The stack trace showed: 0x2c62684: 0x141278 waitForReboot+0x90 0x2c626a4: 0x141501 fault_dump+0x67 0x2c626c4: 0x13a398 handle_fatal_fault+0xa 0x2c62700: 0xfbedcd _ZN19PoolManagedDeltaVol13start FlushLogE P10Delt 0x200816d0: 0xfbbabc _ZN19PoolManagedDeltaVol9indexS yncE18ind exSync 0x200817b0: 0xfd36ca _ZN10SnappedVol14notifyChildren EN7LVolum e8noti
Fix Summary The code has been fixed. Multiple file systems can now be restored to their state before the current transfer is started. This allows operations to be safely run in parallel for multiple Replication V2 sessions.
Tracking Number 406391 / 39554202
1
Code has been fixed to ensure checkpoints are recovered correctly.
399188
224
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description 0x200817e0: 0x8cee38 _ZN11HyperVolume14notifyChildre nEN7LVolu me8not 0x20081800: 0x8c7687
Fix Summary
Tracking Number
Version 6.0.41.3
CEE
The following fixed issues are related to Celerra AntiVirus Agent, Celerra Event Enabler, and Celerra Event Publishing Agent:
Severity Symptom Description Fix Summary Tracking Number 386093
1
When using Celerra Event Enabler, CIFS users were not able to access CIFS shares during multiple FS scans.
Created a new parameter, viruschk.fsscanburst, which defines the number of files sent in one burst by the scan process to the Virus Checker engines. The default is 20 files. The file system scanning will be suspended until the number of files reaches the value set. If the parameter is set to 0, the burst will be dynamically set to the number of AV thread times the number of online CAVA servers.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
225
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 388970 / 37306154
1
When stopping a virus checking service on a Data Mover while having a lot of unchecked files in virus checker queue, the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../sched.cxx at line: 846 SYSTEM WATCHDOG. The stack trace showed: 0xe6475ed4: 0x14124c waitForReboot+0x90 0xe6475ef4: 0x1414d5 fault_dump+0x67 0xe6475f14: 0x1413bd PANIC+0x29 0xe6475f24: 0x1432c9 system_watchdog+0x77 0xe6475f34: 0x139f6a _t_nmi+0x68 0xe6475f70: 0x71dcc6 _ZN15File_NamingNode14setChec kStatusEi+0 x34 0xe6475fb0: 0xd10d60 _ZN13check_initial5startEv+0x 19a 0xe6475fe0: 0x145bfa Sthread_startThread_internal+ 0xe When virus checking was enabled on a Data Mover, it caused failover to take additional time to unmount the Data Mover's root file system. This added approximately 1 minute to failover/ shutdown time.
Code has been fixed to properly handle the locks within the code and prevent this panic.
2
Virus checking no longer takes a reference on the root file system, so its unmount is not delayed during shutdown/failover. New CIFS client connections are no longer blocked by other unrelated connection attempts in progress. Code has been fixed and no longer tries to scan a file if it cannot
387885 / 37272266
1
The Data Mover was unable to send viruschecking requests to an AV server using MSRPC when a GPO update was trying to connect to a slow or nonresponsive domain controller.
390569 / 37333686
1
When using the nonotify mount option on a file system for which virus checker was configured, a deadlock condition occurred
395401 / 38038272
226
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
if a client tried to delete a file in the same directory where any other file was waiting to be scanned. 2 Following a manual check of a file using Symantec virus checking engine, the ctime of the file was changed. This caused a differential backup done by NDMP to be larger than expected.
get the path using the triggerset.
Code has been fixed to properly update the access time without changing the ctime, so it no longer affects differential NDMP backups. The Data Mover now removes files from the pending queue when this type of files is added to the exclusion list and virus checker is restarted.
400155 / 36478750
1
When a file type (.gz) was added to the exclusion list of the virus checker config file viruscheck.conf, the Data Mover still continued to scan these files. This issue occurred when .gz files were already in the pending queue to be scanned, and the filter was changed while the virus checker service was stopped. Once the service was restarted the Data Mover incorrectly sent the files in backlog queue to be scanned.
392790
Celerra Management Applications
The following are fixed problems related to the Celerra Management Applications, i.e., Celerra Monitor, Celerra Manager, WebUI, or Celerra MMC Snap-ins:
Severity Symptom Description Fix Summary Tracking Number 378575 / 36073712
2
User could not use Celerra Manager to change checkpoint schedule from a monthly to weekly recurrence. The following error was displayed: Modify schedule <schedule name>. The task scheduler is not responding. Please try again later.
Code was fixed to allow the modification of a monthly schedule to weekly schedule.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
227
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 404632
1
If a Data Mover was renamed (ex. server_2 was renamed to datamove2), the nas_alerterd could not monitor its CPU utilization. A Celerra system can only be managed by a version of Unisphere that has a Celerra plug-in of the same version or later. If Unisphere was downloaded from one of the Storage Processors in the array, an older version of the Celerra plug-in was installed and the 6.0.40-5 Celerra was not be manageable. A message of Unsupported or Newer Version was displayed. Rapid rebooting or restarting a Data Mover caused a process leak in the Control Station.
Fixed monitoring of Data Movers that are renamed.
1
Code has been fixed to allow all in-family management through Unisphere.
395531
2
Implemented a fix to properly close child processes to prevent process handle leaks, This issue occurred because VPFS file system did not have storageMB value and a null check was not performed. If the file system is of type VPFS, no details of the file system will be displayed. Added code to remove the trailing spaces from the serial value.
398424
2
When user queries for the capacity info or the capability info of the file systems, if there were checkpoints, XML API displayed the following error message and not the correct value: Unexpected exception of unknown nature.
384126
2
During an enumerate instance for an association, the SMI-S response received contained trailing spaces. There was no code to trim the APL returned value for serial. The device type 10 gigabit and the speed 10000FD were not added in the code. When APL returned this device type and speed, XML API displayed the value as other and incorrectly sent the files in backlog queue to be scanned.
385415
2
Added the new device type and the new value of speed to the code.
386449
228
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
3
The last boot time on the Data Mover properties page was not localized whereas the current time was localized.
Dates are no longer localized on the Data Mover properties page. The APL query attribute for fetching NTP server info contained an error which has been fixed. This issue was caused by a typo in an XML response name value which resulted in the setting not being correctly interpreted by the Celerra Management MMC. This has been fixed. Spaces in domain mapped users / groups are now supported in Unisphere.
381447
2
User could not see and confirm that an NTP server was configured. When CSA queried about ntpServers attached to a Data Mover, it did not list any, even though one was attached. User received the following error message when using the Celerra CIFS management snapin on the Data Mover: Internal Error: Unable to retrieve GPO setting from server. Please insure that the cifs server is still connected to the domain. Possible solution would be to rejoin the domain.
387693
2
394296
2
User was unable to add Domain Mapped Groups that contained spaces to Unisphere. When trying to setup LDAP with Unisphere to map domain groups, if user tried to add a group with spaces, an error was displayed which states only _ . are valid special characters. When a CIFS share spanned across multiple CIFS servers, XML API identified only one server and sent the request to APL. APL expected all the servers as the key, and returned the following error: Archiver::createDhsmConnections() Failed to query the file system of '<name>' for server '10.10.17.153'. Error Reason = 'XmlApi::recvXmlResponse Error.
396496 / 37607682
2
The code has been modified to check for share name and CIFS server name and return the information correctly.
137720
EMC Celerra Network Server Version 6.0.70.4 Release Notes
229
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 369551
2
On the Checkpoint Extend page, after selecting the Extend from Meta Volume option, the following message was displayed: No volumes are available' even though there were free meta volumes available. This occurred because meta volumes with the same disk type as the file system were listed in the dropdown on the Checkpoint Extend page rather than same disk type as checkpoint. Warning message was not generated and visible on the dashboard of the Unisphere manager or sys_log if FTP configuration was incorrect in /nas/site/automaticcollection.cfg file.
The GUI code has been modified so that filtered meta volumes have same disk type as the checkpoint and not the file system.
2
Added additional logging in the code path to generate the warning correctly if it cannot contact the remote FTP server. The MGFS state is now reported correctly.
403294
2
When the migration file system related migration encountered an error, for example, the migration source NFS server was down, its state was displayed as 'Idle' instead of 'Failed'. User created an nfs export with options using the CLI, for example, server_export server_3 –o root=root_hosts,rw=root_hosts:frhosts,access=root_hosts:fr-host s. When viewing the export through Unisphere, it was displayed as root:hosts instead of root_hosts. When the number of Celerras being managed was more than 5, the Celerra Capacity View Panel in the Unisphere dashboard showed incorrect information in the title and incorrect display order of the Celerras in the chart. Celerra/CUT plugin reported a different set of domain members upon startup than the CLARiiON plugin.
376681
2
Fixed the APL code that caused underscores to be replaced with colons.
399488 / 38452644
2
Changed the order in which the Celerras capacity details objects passed before updating the view.
372223
2
CimomAccess.pm treated the IPs in reverse order as a new entry and added it to domain_list file. The
374045
230
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
CimomAccess.pm file was modified to make sure it identifies IPs in reverse order as well and does not populate the domain_list file. 2 When using Unisphere and clicking on a list page button (e.g. Delete, Properties, etc.) user waited for the expected Confirm Delete Dialog page to pop up. Occasionally, the Confirm dialog popped up for a split second then disappeared behind the Unisphere window. On a slow or busy client, this happened more often. The creation of user quotas from the root of a file system could not be performed from the GUI, but worked fine when using the CLI. The option "none" was missing from the user and group creation pages in Unisphere. The Run CLI command in Unisphere was not localized. There were no leading/trailing character strings for the text and none of the buttons were converted. The checkpoint delete popup in Unisphere came up blank when running with a language pack enabled. Unisphere code was fixed so that the action pop-up window remains in the foreground above the main window. 378218
2
Code was modified to include the "none" option.
379661
2
This has been fixed, and the Run CLI command is now localized.
381117
1
Code has been modified to check for the localized version of the string. This issue occurred because APL did not send the unit as Megabytes (MB). This has been fixed.
394269
2
User could not modify the Maximum Storage Capacity through Unisphere, but could modify it using the CLI. The following error was returned: invalid size specified.
382336
EMC Celerra Network Server Version 6.0.70.4 Release Notes
231
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 379910 / 36289838
1
In Celerra Manager Group Quota page users were unable to search quotas by GID or group name.
Search by GID and group names capability was added to Group Quota page of Celerra Manager.
CIFS
The following entries describe fixed issues related to Common Internet File Systems:
Severity Symptom Description Fix Summary Tracking Number 406316/ 39423800
1
The Data Mover panicked and with a Page Fault Interrupt when a pre-released version (BETA) of MACOS X 10.7 (Lion) client accessed a CIFS share. The stack trace showed:
0x2ab3a08: 0x13fb44 waitForReboot+0x90 0x2ab3a28: 0x13fd59 fault_dump+0x67 0x2ab3a48: 0x138d68 handle_fatal_fault+0xa 0x2ab3a70: 0xb22430 _ZN21cMsRPC_LsarpcSimulate13lsa rProcess2 DEP13sm 0xc3745b60: 0xb29164 _ZN21msrpc_legacyOperationI21cM sRPC_Lsar pcSimu 0xc3745b90: 0xc9ffd1 _ZN22msrpc_interfaceHandler13ca llOperati onEtR1 0xc3745bf0: 0xc9edb5 _ZN15msrpc_serverCnx15_processR equestER2 3msrpc 0xc3745cc0: 0xc9df72 _ZN15msrpc_serverCnx10processMs gEP13smb_ thread
The code was modified to correctly handle the LSA Lookup call 0x2D and avoid this Page Fault condition
1
When accessing NAS files from Windows, the count of hard links was incorrect. When using the perl stat command on a Windows file, the link count on the file was always 1.
Code now returns the correct link count for the file (note: for directories, the link count is always 1).
392782
232
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
2
If a new file filter file was created with Windows XP or Windows 2003, it worked as intended for users with the same systems, but not on Windows 7 or Windows 2008 . For these newer Windows versions, files could be created but could not be deleted nor edited. During a domain SID migration, an invalid SID in a Security Descriptor (SD) caused a NULL referencing condition. This led to a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: bf1d26 Err code: 0 Target addr: 2 **. The stack trace showed: 0x2acfc64: 0x13fb94 waitForReboot+0x90 0x2acfc84: 0x13fda9 fault_dump+0x67 0x2acfca4: 0x138d94 handle_fatal_fault+0xa 0x2acfdb0: 0x7a9656 _ZN11AclDatabase5parseEP23Par seAclDataba seConte 0x56b7ff60: 0x7a8eed _ZN11AclDatabase5parseEP23Par seAclDataba seCont 0x56b7ffb0: 0x7a84c1 _ZN23ParseAclDatabaseContext5s tartEv+0x6 9 0x56b7ffe0: 0x144344 Sthread_startThread_internal+ 0xe
Code has been fixed to deny creation of files based on file filtering rules for Windows7/ Windows 2008 clients.
395700 / 37640396
1
Code now ignores the invalid SDs during a domain SID migration and the migration continues.
391200
EMC Celerra Network Server Version 6.0.70.4 Release Notes
233
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 384678
1
When using DHSM CIFS connection, recall of some files failed with the following: MGFS: 3: error direct read block 7 of entry 4248, status 20 – EndOfFile. The issue occurred because a secondary CIFS server returned a nonzero size for getAttr request, but a zero size for Open request. Copying a file over a DFS link failed from an SMB2 client, reporting there was not enough space, even though there was plenty of space on the target of the DFS link. This is because the size of the DFS root file system was being checked rather than the target of the DFS link.
Code now handles cases where Open and getAttr request sizes do not match.
2
The Data Mover now sets the REPARSE_POINT file attribute in the DFS link node. This attribute directs the SMB2 client to query for file system size of the DFS link target. Use correct CIFSbased SPN to access Domain Controller.
401201 / 38573658
1
Unable to access server when Microsoft security policy in raised Windows 2008 R2 domain. See Microsoft article: 2345886, Description of the update that implements Extended Protection for Authentication in the Server service,
393548 / 37839090
http://support.microsoft.com/defaul t.aspx?scid=kb;en-US;234588 6
2 When migrating SIDs from one domain to another using server_cifs –Migrate command, unknown SIDs were not migrated. The sever_cifs Migrate command has been fixed, and is now will include unknown SIDs. Permission inheritance is now handled correctly for "Files Only". Added support for share information through SRVSVC pipe, granted access to unprotected functions, 387781 / 37493702
1
When setting permissions from a Windows client, the special permission "Apply onto" as "Files only" did not get propagated to child directories. User experienced slow performance with Windows7 client using MSOffice applications when Celerra CIFS server was configured with UNIX security. Access to shares enumeration returned
383935 / 36713302
1
389804 / 36737358
234
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
an internal error status, causing delays on the client side. 2 When using CIFS auditing and only enabling events for Read operations, Data Mover generated logs for ALL events, regardless of configured events. The issue occurred because of an error in the mechanism that checks if an audit event log must be generated. Under certain stress circumstances, the Domain Controller replied to LSARPC requests with an error: STATUS_INSUFFICIENT_RESOURCES. The Data Mover kept attempting to use this DC, never getting any information. Server log contained errors like: LsarOpenPolicy2::Request failed, status=INSUFFICIENT_RESOURCES. From the client standpoint, SID was displayed instead of clear text names when attempting to get ACL content. User experienced experience slow performance with Windows7 client using MS Office applications when Celerra CIFS server was configured with UNIX security.
like shares enumeration. The error has been corrected, and the Data Mover will now generate logs only for specified events. 384108 / 36692600
1
The connection with the overloaded Domain Controller will be dropped. The Data Mover will now reconnect to another Domain Controller.
382103 / 36576304
1
Added support for share information through SRVSVC pipe to accommodate pipe paths with ASCII names. The code has been fixed to allow widelinks when SMB2 is enabled. The code now allows the single SMB thread timeout without holding a lock on all other SMB threads.
383876 / 36737358
2
When SMB2 was enabled on Data Mover, widelinks stopped working, and all links appeared as files in Windows Explorer.
381269
1
When using CEPP/CAVA over MSRPC and a CEPP or CAVA server did not respond to an SMB thread, other SMB threads were locked awaiting the timeout of the CEPP/CAVA client. This appeared as hangs or slow access for users.
379264 / 36219350
EMC Celerra Network Server Version 6.0.70.4 Release Notes
235
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 380397 / 36177168
2
Attempting to copy with EA (NTFS extended attribute) from a Window machine to Celerra using SMB2 (Win7/W2K8-R2 client and SMB2 enabled on Celerra) fails with popup message: The extended attributes are inconsistent or The file or directory is corrupt or unreadable.
Celerra does not support EA but the status being returned on SMB2 was not appropriate (STATUS_NO_EAS_O N_FILE). Celerra now reports STATUS_EAS_NOT_ SUPPORTED. Also, the parameter used to make Celerra ignore EAs and not return an error (cifs set_eas_ok) now applies to SMB2. The timeout value for delete CIFS server request to the Data Mover has been correctly set to 180 seconds.
1
When deleting a CIFS server ,the timeout of the Control Station command must be set to at least 120 seconds. Instead, it was set to 50 seconds. In some cases 50 seconds was insufficient for DART to perform the deletion. As the result, the Control Station timed out and failed to delete a server from netd and boot.cfg. Upon Data Mover reboot, the deleted CIFS server was unexpectedly recreated on the Data Mover. If the server_thread command was used with -all, the Data Mover panicked with the following DART panic/fault message: *** GP exception. Virt ADDRESS: 0x00006f3ed1. Err code: 0 ***. The stack trace showed: 0x3685760: 0xbf93d0 _ZN7DskDump9checkdumpEPKc+0x7 a 0x36857a0: 0xbf95de dump_return+0x4 0x36857b0: 0x14c7d2 _ZN12Memory_Frame13setProtect ionEm19Memo ry_Prot 0x36857d0: 0x14ce5d fault_dump+0x89 0x36857f0: 0x13378d handle_fault+0x1d
374845 / 36219350
2
NFS threads now correctly display the last operation executed.
390668
236
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
0x106585940: 0x15a5bf _ZNK15cThreadActivity33displa yIdleThread Info_ 0x1065859f0: 0x6f3e8e _ZNK17nfs_serverInitial21disp layIdleThre adInf 0x106585a00: 0x15af4a _ZN15cThreadActivity11parseTh readER12cSv cPars 1 User could not access a server from an NT4 machine (or any client like Samba using old NTLM SessionSetup request). This occurred when NTLMV2 was enforced with: LAN Manager Auth Level: Send NTLMv2 response only, refuse LM and NTLM NTLM SSP Minimum Client Security: Require NTLMv2 session security NTLM SSP Minimum Server Security: Require NTLMv2 session security NT4 client setup: LAN Manager Auth Level on NT 4 client configure to “Level 2 – Send NTLM authentication only”. The following entries were observed in server_log: 2010-07-30 10:22:55: 13157007360: SMB: 3: Client=customerClient NTLMV2 authn required 2010-07-30 10:22:55: 26041909248: SMB: 6: Client=customerClient killed by NTLMV2 required 2010-07-30 10:22:55: 26041909248: SMB: 6: SessSetupX failed with RPC_NT_UNSUPPORTED_AUTHN_LE VEL Code now allows authentication requests to be sent to the domain controller even if it is suspected to be NTLMV1, based on size. 386340 / 3620668
EMC Celerra Network Server Version 6.0.70.4 Release Notes
237
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 388629
1
A Data Mover panicked or the viruschecker stop operation hung when viruschecker was restarted multiple times. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000155833 Err code: 2 Target addr: 0x0000000008 **. The stack trace showed: 0x3686760: 0xbf90b0 _ZN7DskDump9checkdumpEPKc+0x7 a 0x36867a0: 0xbf92be dump_return+0x4 0x36867b0: 0x14c7d2 _ZN12Memory_Frame13setProtect ionEm19Memo ry_Prot 0x36867d0: 0x14ce5d fault_dump+0x89 0x36867f0: 0x13378d handle_fault+0x1d 0x1af6be20: 0x2069bd _ZN8RWlockRT10sharedLockEj+0x 65 0x1af6be60: 0xf72de4 _ZN12virusChecker5mountEP14VF S_FileSyste m+0x20 0x1af6bf00: 0xf6bb96 _ZN12check_survey5startEv+0x4 16 With Access Based Enumeration (ABE) enabled, some DFS links may not appear in directories even though the user has the right to see them.
Code now waits until the end of the viruschecker survey thread before restarting viruschecker. Also removed an extra shared lock that led to this deadlock condition.
2
The problem is now fixed and a user can see all DFS links if he has the right to see them. The offline flags of the CIFS share are now properly returned to the SMB2 clients.
392030 / 37302982
2
The offline function on a CIFS share did not work as expected for Vista and Windows 7 clients. When the offline setting was disabled on a CIFS share, a Vista or Windows 7 (SMB2) client still could use offline function on this folder. An out-of-memory panic occurred when using large credentials (containing more
386604 / 36952502
1
Code now allocates less contiguous
388145 /. 37290766
238
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
than 1500 groups from the forest) for NTLM authentication.DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 289 : Out of memory. The stack trace showed: 0xf84bd310: 0x13fb94 waitForReboot+0x90 0xf84bd330: 0x13fda9 fault_dump+0x67 0xf84bd350: 0x13fc91 PANIC+0x29 0xf84bd360: 0x13a8f2 _ZN19Memory_AddressSpace10all ocPagesEi12 Memory 0xf84bd380: 0x1f6c3a _Z11more_memoryjP9free_list+0 x4a 0xf84bd3a0: 0x1f6cf8 malloc+0x98 0xf84bd3d0: 0x1f7485 _Znwm+0x15 0xf84bd3e0: 0x1f74b2 _Znam+0xc
memory to complete user's NTLM authentication.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
239
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 383732
1
A memory corruption in a Kerberos context was used by a function that then returned a NULL pointer. The pointer was not checked before being passed into another function, which then tried to dereference the pointer, resulting in a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 1fa53e Err code: 0 Target addr: 0 **. The stack trace showed: 0x2acb158: 0x13fb94 waitForReboot+0x90 0x2acb178: 0x13fda9 fault_dump+0x67 0x2acb198: 0x138d94 handle_fatal_fault+0xa 0x2acb1c0: 0x56e267 krb5_cc_resolve+0x17 0x1eca92b0: 0x56e609 krb5_cc_default_suffix+0xa3 0x1eca92e0: 0x53e1db _Z26krb5_gss_release_principa lPjPPv+0x46 b 0x1eca93a0: 0x53ed90 _Z25krb5_gss_acquire_cred_ext PjPv22gss_b uffer_ 0x1eca9400: 0x5456ff gss_acquire_cred_ext+0x4b When the .etc of a share's file system was missing, each time the share was mounted for the first time, a message similar to the following was logged to the server_log: Cannot get .etc dir (28=InternalError). When trying to set ACL to a file which had an empty DACL (Discretionary Access Control List) using emcsetsd, the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: /disks/NSG_Root/Maint_4/5.6_bui ld_nfs_env/ Dart/
The fix is to check the return value of the pointer, and return an error if the value is NULL.
3
This message was informational only, as the code was correctly handing the issue. This message is now hidden from the user. Before the fix, code was overwriting the header of the ACL. Code now checks if the file has an empty DACL, and write the new ACL after the ACL header.
393943 / 37873442
1
397384 / 38188396
240
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
server/src/cfs/include/acl.hxx at line: 1522 bad SD storage size. The stack trace showed: 0xde7617e0: 0x141254 waitForReboot+0x90 0xde761800: 0x1414dd fault_dump+0x67 0xde761820: 0x1413c5 PANIC+0x29 0xde761830: 0xe824cc _ZN14set_celerra_sd7processEv +0x336 0xde761910: 0xe826e9 _Z21cifs_set_celerra_sd_1P27c ifs_set_cel erra_s 0xde761990: 0xe85ab9 _ZN29xattrp_IterSvc_Parent_In itialC1EP16 Rpc_Sv 0xde761a40: 0x4a9092 _ZN21Rpc_Svc_Child_Initial13s vc_getreqse tEP7rp 0xde761c30: 0x4ac16f _ZN25Rpc_IterSvc_Child_Initia l7svc_runEv +0x10b
EMC Celerra Network Server Version 6.0.70.4 Release Notes
241
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 374292
2
User could not install CelerraCifsMgmt on a Windows 64-bit OS.
Removed the restriction in the CelerraCifsMgmt install package that prevented user from installing it on 64-bit system. The following message is now displayed when installing CelerraCifsMgmt on a 64-bit system: You are now installing 32-bit CelerraCifsMgmt on a 64-bit Operation System. Celerra UNIX property page extensions are only available in Active Directory Users and Computers on 64-bit Operation System through the following instructions: 1. From the Start menu, select "Run ..." 2. In Open field, input the command: mmc /32 dsa.msc 3. Click the OK button to launch the Active Directory Users and Computers The fix respects the precedence of the DELETE_PENDING on the SHARING_VIOLATIO N status. Added /secforce option that force emcopy to overwrite destination security and prevent this condition.
2
A user application might not work properly when using temporary files on a CIFS share. The Celerra returned a SHARING_VIOLATION instead of DELETE_PENDING.
391399 / 37403078
2
During data migration from Windows 2003 file servers to CIFS servers on Celerra using EMCOPY, the following error appeared: EMCOPY: permission changes either copies or does not copy files from
393429 / 37414066
242
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
source to dest. 2 Using the nonotify mount option did not prevent notifications of changes to a directory from being sent to CIFS clients. Also, the option could interfere with directory renames. The nonotify mount option now disables CIFS notifications and does not interfere with directory renames. 395058 / 37393392
Control Station
The following entries describe fixed problems related to the Celerra Control Station.
Severity Symptom Description Fix Summary Tracking Number 144684 / 29744606
1
When an NBS session (which included failovers) was started on the Control Station, a race condition occurred in the NBS driver that led to a Control Station kernel panic and reboot with the following message: Kernel panic - not syncing: Fatal exception. The server_ldap -lookup command did not query groups containing spaces in the name.
To fix this issue, a locking mechanism was added to prevent this race condition.
2
The Control Station now correctly processes LDAP queries containing spaces in the group name. With the fix the command now rejects incorrect parameters and displays command usage.
374688 / 35603200
3
When a nas_quotas report command was entered with incorrect parameters, the command returned the following error message instead of the command usage: nas_quotas: Warning 17716809847: server_2: NIS Error: UID= <file system>.
376711 / 35626866
EMC Celerra Network Server Version 6.0.70.4 Release Notes
243
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 380335 / 36355420
2
If the interface name of the server_cifs command ended with CRLF which is "carriage return" on Windows , the Eiffel trace and vdm.cfg or netd table was corrupt. Management daemon crashed while attempting to allocate memory in order to service an XML message, causing the core dump to fill up the /nbsnas partition.
Code now truncates the CRLF character before doing transaction.
2
In a rare case where the memory is fragmented to the point that the service of an XML message was unable to complete, a retry mechanism was added to re-attempt to allocate the memory. Memory is now recovered in the background. Code has been fixed to verify the username and password correctly. Code now correctly detects the components present in the enclosure.
386399 / 36932966
2
User was denied permission error if the username exceeded 8 characters while executing the new CLI which needed special admin role privilege. When a VG8 system was configured with 3 Data Movers, a false alert that the power supply was removed from nonexistent Data Mover was reported. 0x330070051 Error: Local emc70pri common DME 1 Power Supply B0: Removed. If user specified a storage system for a file system extension and the storage system was different than what the file system already exists on, they might get storage from the wrong storage system.
377427 / 35919230
1
397608 /; 38148792
2
If a storage system is specified in file system extension, code now only takes storage from that storage system for the extension. When creating/refreshing checkpoint task from APL scheduler fails, only one error will be
378214
3
When creating or refreshing a checkpoint task from APL, if the scheduler failed, both APL task manager and APL scheduler generated the same error: Create checkpoint ckpt_alert_sched_001
378439
244
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
of schedule alert_sched has failed. 3 When running a dbchk -f -q command, the following error was logged in the sys_log: A Celerra database error was detected. An inconsistency in the Celerra database files was detected. Run the $NAS_DB/tools/dbchk tool to identify the exact error. However, when user ran the dbchk tool, the output was clear: [nasadmin@apoh0010cs03 ~]$ /nas/tools/dbchk -pvwx dbchk: done. The Data Mover reported too many errors in server log for software correctable errors and generated SMI (system management interrupt). The system clock got out of sync while the SMI was being processed, which resulted in a Data Mover panic with the following DART/panic fault message: >>PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG. The stack trace showed: 0x3635a40: 0x14c81d _ZN12Memory_Frame13setProtect ionEm19Memo ry_Protection_t+0x23d 0x3635a60: 0x14ce29 fault_dump+0x55 0x3635a80: 0x14cf02 PANIC+0x5a 0x3635aa0: 0x14dd29 system_watchdog+0x61 0x3635ab0: 0x132913 t_nmi+0xc3 0x3635b68: 0x378674 _ZN12bcmnicDevice12bcmInterrupt EPv+0x5a 0x3635b98: 0x1d338f _ZN3ISR6invokeEv+0x2d 0x3635bb8: 0x1d3e91 shared_isr+0x207
displayed. If the nas command finishes within 10 seconds, dbchk will no longer report an error (unless nasdb finds some inconsistency problems). 395926 / 38096334
1
Fixes the PCI correctable error handler for devices on the motherboard so it correctly prints warning messages instead of panicking the system and falsely indicting the CPU module.
374413
EMC Celerra Network Server Version 6.0.70.4 Release Notes
245
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 389936
1
Under heavy system load, a large amount of data frames were dumped, causing dumpsize to exceed 24G and dump_slot not showing the right size. This could result in the dump being corrupted, or important data could be missed from the dump. When each checkpoint schedule was fired off, APL scheduler checked its next execution time. If the next_execution_time + 60 seconds < current time, this schedule was marked as miss, and the checkpoint did not create/refresh. If too many schedules fired off at the same time, some of them did not create/refresh the corresponding checkpoint. The server_ifconfig server_X -a command did not display netname for IPV6 interfaces. Netnames were displayed for IPv4 interfaces only.
Code has been fixed to skip the buffer data frames when not required and calculate the correct data dump size.
2
Increased the timeout value from 60 seconds to 5 minutes to allow more schedules to run at once.
384138 / 35719928
2
Corrected the query sent to the Data Mover ensure it displays both IPv6 and IPv4 netnames. Code now rejects an attempt to change the physical Data Mover's name to the same name as the VDM's. This memory leak condition has been eliminated.
384477
2
User changed the physical Data Mover's name to the same name as the VDM using the server_name command. This caused several server_xxx commands to not function properly. Using the nas_dbtable with a filter on a field which was not the primary key, or intensive use of the nas_dbtable caused the Data Mover to run out of memory
385998 / 37053750
2
394952
CORE
The following entries describe fixed problems related to Celerra Data Movers.
246
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 16063 / 32384702
1
Commands against the Data Mover failed with Error 5: server_4 : Input/output errorTimed out or the Data Mover did not respond to the command at all because the MAC threads were hung in an unmount deadlock. [root@A00DRFL003 log]# server_date ALL server_2 : Tue Dec 1 10:04:11 GMT+1 2009 server_3 : Tue Dec 1 10:03:43 GMT+1 2009 server_4 : Error 5: server_4 : Input/output errorTimed out server_5 : Tue Dec 1 10:04:41 GMT+1 2009
ReclaimFreeNode and giveUp are now called directly without checking for a useCount of zero.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
247
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
A memory leak in the rpcbind_getaddr() function of the code caused a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../bufpool.cxx at line: 389 : couldn't get a free page. The stack trace showed: 0xcd2ab8f0: 0x13f244 waitForReboot+0x90 0xcd2ab910: 0x13f441 fault_dump+0x67 0xcd2ab930: 0x13f341 PANIC+0x29 0xcd2ab940: 0x17d7ef _ZN18Buffer_Cached_Pool17make CachedBuffe rsEi+0 0xcd2ab970: 0x13b4ed _ZN19Memory_AddressSpace16mak eCachedBuff erEv+0 0xcd2ab990: 0x17d37d _ZN18Buffer_Cached_Pool15getA nyCachedBuf Eji+0x 0xcd2ab9c0: 0x6b1ec3 _ZN14UFS_FileSystem8getBlockEliRPv+ 0x41 0xcd2aba00: 0x6b1629 _ZN14UFS_FileSystem15readBloc ksAsyncElii PP13Bu
The code has been changed to avoid the memory leak.
380233
CSA
The following entries describe fixed issues pertaining to the Celerra Start-up Assistant.
Severity Symptom Description Fix Summary Tracking Number 379078 / 36174238
2
While configuring the Control Station through CSA, if user tried to set the CS and blade time to (GMT +5:30) Asia/Calcutta (as shown in the drop-down menu). The "Apply Changes" step in CSA failed with NTP server, time zone and unicode properties set up on primary
Time zone information has been changed in the CSA to reflect the updated spelling of Kolkata.
248
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
blade... - Invalid time zoneccmdEscColon Asia/Calcutta. 1 The site ID that was entered by CSA incorrectly used double quotes and was not visible in Celerra Manager or Unisphere. Code was fixed to ensure users can see the information in Celerra Manager or Unisphere. Code was altered to send these CLARiiON events to the sys_log. 361995
2
CLARiiON events for NX Systems with AX4-5F8 backends were passed to /var/log/messages but were not passed to sys_log as expected. When the user tried to upgrade a Fibre Channel (FC) enabled systems from version 5.6 to 6.0 or tried to install integrated systems (running NAS and FLARE), the Celerra Startup Assistant failed to perform the Celerra and CLARiiON system domain join operation. In the Celerra Startup Assistant (CSA), during Control Station install, the Gateway field was not indicated as a required field.
393090
1
The Celerra Startup Assistant was modified to perform a domain join for FC-enabled systems.
382973
2
The Gateway field is now marked as a required field.
361899
Deduplication
The following entries describe fixed issues pertaining to Celerra Deduplication:
Severity
Symptom Description
Fix Summary
Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
249
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 378549
2
Error messages similar to the following were incorrectly logged as ERROR in the server_log when mounting the FLR file system for the first time: Error 2010STORAGENumpaths option u, E EDT 2010FSTOOLSFLR data popula, Error T 2010FSTOOLSflr_clock = -1, flr_max_time. During a read operation of a file that has recently failed deduplication through the CIFS compression or DHSM API methods (via VMWare plugin), the Data Mover panicked with the following DART panic/fault message: isAValidIndBlkBuf: bad bn.The stack trace showed: 0x271b1450: 0x141254 waitForReboot+0x90 0x271b1470: 0x1414dd fault_dump+0x67 0x271b1490: 0x1413c5 PANIC+0x29 0x271b14a0: 0x77663b _ZN14UFS_FileSystem28markCorr uptedFsForU nmount 0x271b1660: 0x78ccbe _ZN14UFS_FileSystem17isAValid IndBlkBufEP 13Buff 0x271b16e0: 0x782d47 _ZN9UFS_Vnode17readIndirectBloc kElRP13Bu ffer_C 0x271b1710: 0x7852c2 _ZN9UFS_Vnode8bmapListEmmPmPi+0 x228
These errors were informational only, and are no longer displayed.
1
Code was updated to ensure data is properly flushed to disk before deduplication is started via the CIFS compression method or the DHSM API method (via VMWare plugin) to prevent this panic.
378009 / 36040202
1
There is a deadlock situation in the FRDE re-ingest code that may be hit if a Data Mover panics and reboots or a file system is re-mounted while having entries in the re-ingest queue. The deadlock will prevent files in the re-ingest queue from being processed. This may cause the reingest queue to become filled and reject new entries.
Code has been fixed to prevent this deadlock condition.
379193
250
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
1
When using NDMP backup on file systems that have deduplication enabled, NDMP backups could take more space than necessary, and NDMP jobs could fail with the following error messages in the log DEDUPE: 3: EDRS: RP CSX COMPRESSOR: internal error decompression failed: 2 DEDUPE: 3: EDRS: TRANSFORMING FILEDATA UNIXFSIS: readBlob: transformOut failed at offset 3244032: RDE_STATUS_DECOMPRESSION_ERROR. Because of a limitation in the compressor module of code a failure could occur if this code receives more than 2 requests for decompression at a time. When using Deduplication, if many CIFS clients were performing directory scanning on F-RDE enabled file systems, CPU utilization on Data Mover reached 100% and impacted performance. The issue occurred because code in deduplication policy engine (PE) was not optimized. There was a race condition between removal of deduplicated files and file system unmount operations. Under certain conditions, deduplicated data space was not reclaimed from the file system when files were removed. A message was displayed in the server log similar to: 2010-05-09 09:40:57: 13163036672: DEDUPE: 3: Cannot find file system with id 15, cannot cleanup object <dart://rde ...
Code now ensures that compressor module code does not receive more than 2 requests for decompression at a time.
381922 / 36502216
1
PE code has been optimized and will not cause 100% CPU utilization under these conditions.
385163 / 36929424
2
If you attempt an unmount during removal of deduplicated files, the code now waits for threads to complete before performing the unmount.
367398
EMC Celerra Network Server Version 6.0.70.4 Release Notes
251
Version 6.0.41.3
File system
The following entries describe fixed file system issues:
Severity
Symptom Description
Fix Summary
Tracking Number 387997
1
When a file system was mounted on a VDM, if the tree root’s inode number was the same as the inode number of the mount point to which that VDM was mounted, renaming such tree quota from the client resulted in “/..” path reported. The cylinder group bitmap initialization process is started asynchronously for each file system as they are mounted on the Data Mover. If a file system was unmounted or the Data Mover was rebooted, failed over, or failed back before this operation completed, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 7831ec Err code: 0 Target addr: 2c **. The stack trace showed: 0x2ba5e34: 0x140cec waitForReboot+0x90 0x2ba5e54: 0x140f71 fault_dump+0x67 0x2ba5e74: 0x139eec handle_fatal_fault+0xa 0x2ba5ee0: 0x1c35ab _ZN17SvcThread_InitialIS_I15St hread_Init ialEE5s 0xe6ca1fe0: 0x1455d4 Sthread_startThread_internal+0xe
Code has been fixed to properly handle this condition so the rename of quota tree will be completed as expected. The cgbitmap initialization is now aborted if the file system state is unmounting, unmounted, frozen, or corrupt. It will be reinitialized when the file system is remounted.
1
357994
1
Data Mover denied write I/O with NFS4ERR_ACCESS instead of NFS4ERR_LOCKED in when a file is locked. When a file was being truncated through setAttr or setLength calls and any panic or unexpected reboot happened during the
NFS threads now return the correct error code.
392703 / 37513228
1
Code has been fixed to properly handle SetAttr/setLenght logic
382208
252
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
same time, the Data Mover panicked when attempting to access that file. DART panic/fault message: isAValidIndBlkBuf: bad bn Panicking Processor: 3. The stack trace showed: 0x79b313f0: 0xa6bb4c _ZN14UFS_FileSystem28markCorru ptedFsForU nmount 0x79b31560: 0xa927dc _ZN14UFS_FileSystem17isAValidI ndBlkBufEP 13Buff 0x79b315e0: 0xa83009 _ZN9UFS_Vnode17readIndirectBlo ckEiRP13Bu ffer_C 0x79b31630: 0xa8407d _ZN9UFS_Vnode4bmapEiRiPiS1_+0x1c 1 0x79b316a0: 0x18c38a1 _ZN10DHSM_Vnode14itruncInterna lEP20UFS_U pdate 0x79b31710: 0xa7fbd1 _ZN9UFS_Vnode6itruncEP20UFS_Up dateDescri ptoryi 0x79b317a0: 0xa80208 _ZN9UFS_Vnode8inactiveEPvi11Fi le_Statusi i+0x47 0x79b31820: 0x1893362 _ZN10MGFS_Vnode8inactiveEPvi11 File_Statu sii+0 1 A Secure NFS mount from a client could caused a Data Mover panic with the following panic header: Virt ADDRESS: 0x00006dd730 Err code: 0 Target addr: 0x000000000c.
to prevent the panic under this condition.
Code was updated to prevent the panic and ensure Secure NFS mounts succeed.
398127
EMC Celerra Network Server Version 6.0.70.4 Release Notes
253
Version 6.0.41.3
1
During file system operations, code improperly passed a handle to a different function. Later the handle has changed, but the function that this handle was previously passed to did not realize this, and used the old handle which led to memory corruption. This resulted in the panic with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 525 : Memory corrupted. The stack trace showed: 4056c0: 0x24aa1f free+0x79 405700: 0x24aded _ZdlPv+0x9 405710: 0x9682b9 _ZN20UFS_UpdateDescriptor8comp leteE11Fil e_Sta 405820: 0x8a1c9f _ZN21aggregateCommitExtent6doWor kEP9VFS_Vn ode 405870: 0x930b47 _ZN13aggregateWork9writeListEP10 gatherList P9V 4058f0: 0x92f6d4 _ZN17VersionFile_Vnode15commitAl locListEP1 1ex 405940: 0x8aafa5 _ZN9UFS_Vnode18uncachedWriteChun kEyjRP4msg bRi 405b30: 0x8ab627 _ZN9UFS_Vnode13uncachedWriteEyjP 4msgbiRj+0 x1a
Code now properly passes the handle by reference to other functions.
397129
1
After a Data Mover failover/failback, certain client requests (allocating writes and file creates) were blocked until all cylinder group (CG) bitmaps were initialized. Depending on the configuration, background initialization took a few minutes or panicked after 6 minutes with the following DART panic message: >>PANIC in file: ../sched.cxx at line: 1543 : SYSTEM WATCHDOG. The stack trace showed: Panicking Processor: 0 0x730cd070: 0x14ee4b system_watchdog+0x61 0x730cd080: 0x140efb t_nmi+0xc3 0x730cd1b0: 0x8d929b _ZN14UFS_FileSystem8allocInoEP 20UFS_Upda teDesc 0x730cd300: 0x9436e3 _ZN9UFS_Vnode12getChildNodeEP2
The block allocate function was modified to only wait until any one cylinder group is found with free blocks prior to continuing the client requests.
399405
254
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
0UFS_Updat eDescr 0x730cd3a0: 0x8e615b _ZN9UFS_Vnode9setLengthEyiiiP1 3Buffer_Ca ched+0 0x730cd410: 0x150ceb6 _ZN10DHSM_Vnode9setLengthEyiii P13Buffer_ Cache 0x730cd460: 0x7ee121 2 In the xlt.cfg file, when the subnet address was defined in the format “x:y:z:w, a:b:c:d”, it failed to lookup the host address and ignored the subnet entry. When FSCK was run on a file system that had bad sectors, a memory leak occurred which caused the Data Mover to reboot. Code was fixed by adding delimiter support for "," along with the existing "/". The allocated bad inode objects and bad sector objects are now deleted. Full clone operations are now suspended during auto extend. If file system Auto Extension is enabled and the HWM is exceeded, code pauses the block copy loop until the extension completes (usually around 30 seconds). In the event we really do run out of pool space and can't complete the extend we only allow 5 delay tries during the full clone operation. The file system reference is no longer stored until the resume event completes. If there is a request to unmount the file system, it can now be processed immediately. 375781
2
114569
1
User got Out-of-Space errors on virtually provisioned file systems that started off at a small size (1 GB) and had Auto Extension enabled during a full clone operation.
383532
1
A resume event was not processed when all asynchronous input-output threads were busy deleting large sparse files. The file system became inaccessible during mount cycles and the user saw the following error: No such file or directory>>File system inaccessible.
378345
EMC Celerra Network Server Version 6.0.70.4 Release Notes
255
Version 6.0.41.3
1
When creating new tree quotas and copying large amounts of data, if there were not enough sequential pages for quotaTreeVnodes array, the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory. The stack trace showed: 562ee0: 0x145966 _ZN19Memory_AddressSpace10allocP agesEl12Me mor 562f20: 0x24a3de malloc+0x11a 562f60: 0x24aeab _Znwm+0x1b 562f80: 0x24aed6 _Znam+0x4 562f90: 0x8caddb _ZN14UFS_FileSystem20treeQuota On_Interna lEPcS 563210: 0x8cf8b7 _ZN14UFS_FileSystem16quotaOn_Int ernalEPcS0 _+0 563300: 0x833066 _ZN14VFS_FileSystem7quotaOnE9quo taTypetPcS 1_+ 563350: 0x8293d0 _Z17File_QuotaControlR6OptionP18 VFS_FileSy stem
Code was optimized so that sequential pages are no longer required for this operation.
399030 / 38339946
1
During a Data Mover failover in NAS 5.6, the internal mount command had become case sensitive. File systems with accesspolicy=mixed (lower case upgraded from older NAS code) failed to mount and printed the following message in the cmd_log.err 2009-10-31 15:26:38.777 db:0:31381:E: /nas/sbin/slot_recover 2 -panic 2009-10-31 15:30:54.196 db:0:31381:E: /nas/sbin/slot_recover 2 -panic: replace_storage: replace_mount: eng0:mount: unknown accesspolicy: mixed. home0:mount: unknown accesspolicy: mixed. corp0:mount: unknown accesspolicy: mixed. derived0:mount: unknown accesspolicy: mixed. This failure caused the system to abort the
Reinstated case insensitive behavior for the server mount accesspolicy options during failover.
158137 / 32026672
256
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
1
failover and restore all services on the original Data Mover. The Data Mover panicked because of the UFS Log recovery mechanism attempted to invoke FSCK resulting in NULL pointer dereference. DART panic/fault message: *** Invalid Opcode exception. Virt ADDRESS: c ***. The stack trace showed: 0x2c297e8: 0x141254 waitForReboot+0x90 0x2c29808: 0x1414dd fault_dump+0x67 0x2c29828: 0x13a3a0 handle_fatal_fault+0xa 0x2c29880: 0x8504b1 _ZN7Fsck_SB4readEP11NewFsckArgp+ 0x51 0x507b9e0: 0x84f8f7 _Z8check_sbP11NewFsckArgpP7Fsck_ SB+0x23 0x507ba70: 0x85c139 _Z9doNewFsckP9IO_ObjectiiiiRPci+ 0xed 0x507bac0: 0x85d8b8 _Z20kickOffFsckInRecoverP15fsRe coverStru ct+0x84 0x507baf0: 0x81a850 _Z20handleRecoverCommandR6Option +0x60a
The code now checks for a valid file system object for FSCK kickoff during ufs log recovery.
376227 / 35807576
2
User was unable to delete rep v2 iSCSI replication session when its production file system was full. The task hung and the following error was displayed in the server_log: VCS: 3: Failed to write attribute file, fsid=31, path=fs31_T1_LUN2_CK200082600835_0 000 name=.opaque status=NoSpace The Data Mover incorrectly returned NFS3ERR_NOENT (No such file or directory) when the checkpoint (.ckpt) directory was being looked up.
Code was modified to ensure the session delete completes under this condition.
359609 / 34143216
1
The fix eliminates the race by no longer caching checkpoint mountpoint names.
162103 / 32681254
EMC Celerra Network Server Version 6.0.70.4 Release Notes
257
Version 6.0.41.3
1
A Data Mover panicked when moving a directory to new parent which was being deleted. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 6d6db8 Err code: 0 Target addr: 44 **. The stack trace showed: 0x2c2a524: 0x141254 waitForReboot+0x90 0x2c2a544: 0x1414dd fault_dump+0x67 0x2c2a564: 0x13a3a0 handle_fatal_fault+0xa 0x2c2ab30: 0x81cfe9 _ZN19ShadowDirectoryBase6renam eEPtmPS_S0 _P17Acc 0xdd52bcf0: 0xc515d0 _ZN8AppLibNT6renameEPtS0_lR11F ile_Status R10ext 0xdd52bed0: 0xc427c2 _ZN14smb1_threadCtx11replyRena meEP4msgb+ 0x11e 0xdd52bf40: 0xbe1033 _ZN14smb1_threadCtx7processEP4ms gbi+0xcdf 0xdd52bfa0: 0xbd5820 _ZN13smb_threadCtx5startEv+0x6d0
If a parent directory to which the directory is being moved has been deleted, code now returns "no such file" and no longer panics.
371168 / 35185054
1
User reported poor write response time with VMware ESX using Celerra through NFS with uncached file systems.
In this case, the Data Mover serialized the writes and sent the replies after the last write storage. Code was changed to no longer serialize incoming NFS write requests. The cylinder group bitmap which is used to quickly find free blocks is now initialized at bootup.
404209
1
When writes were performed on a large (>2TB) file system which was nearly full (>80%), threads become blocked. This was observed by slow user access and messages like the following in the server log: ThreadsServicesSupervisor: Service:NFSD Pool:NFSD_Exec BLOCKED for 419 seconds. A panic is also possible with the message: PANIC in file: ../ufslog.cxx at line: 6895 : sync: no
386406 / 37135080
258
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
progress reducing dirty list. 1 Using IP addresses with an invalid format (separated by a dot instead of a comma, for example, 102.30.121.131.102.30.109.8) as part of the export command caused memory corruption. The Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 7cd814 Err code: 0 Target addr: 1d3 **. The stack trace showed: 0x2c29dd4: 0x141254 waitForReboot+0x90 0x2c29df4: 0x1414dd fault_dump+0x67 0x2c29e14: 0x13a3a0 handle_fatal_fault+0xa 0x2c29e70: 0x7cd63f _ZN14UFS_FileSystem17startMetada taSyncEv+0 xf 0xb278ff40: 0x7cd4dc _ZN14UFS_FileSystem4syncEi+0x260 0xb278ff90: 0x7cc395 _ZN21UFS_SyncThreadInitial5start Ev+0x2c7 0xb278ffe0: 0x145bd6 Sthread_startThread_internal+0xe 1 A user file system reported corruption of the gidmap file which prevented it from being mounted on the Data Mover. The problem occurred if specific bytes of a gidmap file were zero ( 2th, 3th and 512th bytes), resulting in the checksum being calculated incorrectly and the check during mount failing. The data type for the calculation was corrected and the checksum is now calculated correctly. 377385 / 35859262 Code now limits the buffer overflow that can potentially happen with an incorrect IP address format. 370824 / 35147654
FileMover
The following entries describe fixed issues related to the Celerra FileMover:
EMC Celerra Network Server Version 6.0.70.4 Release Notes
259
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 385097 / 36947738
1
While using FileMover, high memory utilization used in EDRS handles caused CIFS threads to hang and rendered the data unavailable.
The high memory utilization was because the memory used in EDRS read handles was not being freed quickly enough as they were getting blocked by another EDRS process. This issue was resolved by putting the close/free operations into their own thread pool. Code has been fixed to check all existing tree IDs resulting in proper path display after a directory rename from the client.
1
When having non-sequential tree IDs on the file system, renaming one of the directories representing one of the quota tree path (having an ID after the gap in the list) from the client did not reflect the change in the quota tree list using the nas_quotas command. The issue occurred because code was checking the sequential IDs only to the first gap. DHSM connections (using Kerberos for authentication) tried to re-use Kerberos tickets in the credential cache. The code used to check to see whether a ticket was available in the cache did not check to see if the ticket was expired, so the specified password was deleted. Subsequent code tried to use the expired ticket, then was unable to get a new ticket because the password had already been deleted, resulting in DHSM reconnection failure. For XML API clients, a query was made to get Shares with a filter of name="value", but the value field was specified as caseinsensitive. Code enforced that values had to be case -sensitive, which caused older queries by XML API clients to fail if they did not specify the value correctly.
386777 / 37133416
1
Added code (in Kerberos) to check for expired ticket in credential cache, and to delete the ticket if it is expired.
375592 / 35601816
2
The C3 code was modified to accept case in-sensitive values for share names in order to remain backward compatible.
384012
260
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 401210 / 37512728
2
User could not setup a NetApp server as a secondary server for DHSM. Netapp always returned unspecific error .
This has been fixed, and the CIFS client can now connect to NetApp server.
Installation/Configuration/ Upgrades
The following entries describe fixed issues related to installations, configurations and upgrades:
Severity Symptom Description Fix Summary Tracking Number 385933 / 36553396
2
The sys_log showed the following repeatedly: Enclosure 0 failed to respond to "ping" at management switch A. These alerts were due to slow response from system Management Switches to Control Station ping requests. If the partition table was corrupted, the PUHC did not capture the error and the upgrade failed at step "Create partitions for upgrade".
Code was fixed to prevent false reboot messages and to harden some boxmonitor communication. A new PUHC check was added to ensure that the disks are consistent so that the issue could be captured prior to the start of the upgrade. Added new check for secondary CS to verify whether root file system have enough space prior to upgrade.
1
391103 / 37606144
1
Control Station upgrade failed due to insufficient space on the "root" file system. In PUHC, there was no check available for the secondary Control Station (CS1) to verify the "available space" on "root" file system before upgrading.
375108
EMC Celerra Network Server Version 6.0.70.4 Release Notes
261
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 381208
2
Post Upgrade health check failed with error: Error HC_SYS_14505213994: "nas_storage -check" command timed out.
Code has been modified to avoid this timeout condition. The checks "cs::check_if_Symapi_ data_is_pres ent" and "cs::check_if_Symapi_i s_synced_with_Storag e_System" are shifted to the end of the "cs" checks in checks.txt file to gain some time. This error is no longer displayed.
1
An install of version 6.0.36.4 in an NSG2 system failed with the following: cmd: /nasmcd/sbin/cs_dhcpd_monitor single_execution output: Unable to start cs_dhcpd_monitor -single_execution, already running! If user's PATH variable was missing and they tried to run clariion_mgmt, the iptables file was not updated. This caused the following error during PAHC/PUHC: Storage System : Check SPA SPB communication Error HC_BE_14505017473: At least one iptables rule is missing. Action : Run "spa_spb_comm -setup" to fix the problem. While performing an out-of-family upgrade from 6.0.x to 7.0.x, a VG8 system was incorrectly reported as VG2. During an in-family upgrade of a dual Control Station system, on the secondary CS, the task 'create IDE cache' was not run , therefore the files were copied directly to the common cache and wbem directories, thus filling up the Control Station root. In an out-of-family upgrade from 5.6 to 6.0, if the NAS service failed to stop
381753
2
Adding full path to commands in clariion_mgmt using iptables-save to prevent this condition.
384764
1
The system now reports the correct model. The task 'create IDE cache' is now run during upgrades to prevent this condition.
402851
2
380619
1
A fix has been provided to ensure that
389477
262
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
properly and some NAS file systems were left mounted at the end of the task. Upon Control Station reboot, the upgrade could not complete since the nas file systems were mounted to restore pre-reboot system state. Error - Partition table on /dev/hda is inconsistent. 2 After installing new language packs on top of an existing language pack in Unisphere, error messages were returned in English. When the file /nas/sys/nas_install_version did not exist, the upgrade failed with the following error: /nas/sys/nas_install_version: No such file or directory. There was no reference to which version of NAS code the system was originally installed with.
no NAS related file systems are mounted by the upgrade for state restoration once the NAS service has been stopped during the upgrade. With this fix, error messages are returned in the installed language. Added /nas/sys/nas_install_ve rsion into nasdb_backup code. 395453
2
387041
iSCSI
The following entries describe fixed iSCSI-related issues:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
263
Version 6.0.41.3
1
In a cluster environment, the iSCSI server hung and stopped responding. Host A had heavy I/O traffic to a LUN, and Host B tried to reserve the same LUN, using the SCSI-3 Persistent Reservation command. The reserve command hung, causing a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../dskdump.cxx at line: 1836 : ESX iSCSI access issue. The stack trace showed: 0x1f055ae0: 0x140ce4 waitForReboot+0x90 0x1f055b00: 0x140f69 fault_dump+0x67 0x1f055b20: 0x140e51 PANIC+0x29 0x1f055b30: 0xa2da32 _Z11paniccfgcmdR6Option+0x17e 0x1f055c70: 0x1c9f67 _ZN11cfgcmd_desc6invokeEPKcPP8CC MD_Msgii +0x1a7 0x1f055ce0: 0x2ae4a2 _Z10xml_cfgcmdP7macDataRP8CCMD_Msg +0x392 0x1f055d30: 0x2a9418 _ZN17macCCMDCommand_DB6invokeEPK cP7macDa taRP8C 0x1f055d60: 0x2a991e _Z17mac_CCMDDocParserP7macDataRP 8CCMD_Ms g+0x20
The problem was caused by coding bug which created a small unprotected windows where an I/O completion callback was skipped. This was corrected and the unprotected window has been eliminated.
377382
NDMP
The following entries describe fixed issues related to Network Data Management Protocol and/or Backup:
Severity Symptom Description Fix Summary Tracking Number 387236 / 36692786
2
McAfee antivirus applications reset the last accessed time (atime) of a file after a clean scan. The Data Mover code updated the ctime when the atime was reset which impacted other features such as incremental NDMP backups. A race condition could occur during NDMP backup, when abort is called ,
The ctime is no longer updated as a result of an antivirus client resetting the atime.
1
Code has been fixed by using mutex lock to
395387 / 38052144
264
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
which results in a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 1d32a4 Err code: 0 Target addr: 0 *. DART stack trace: 0x2c4ba44: 0x14124c waitForReboot+0x90 0x2c4ba64: 0x1414d5 fault_dump+0x67 0x2c4ba84: 0x13a398 handle_fatal_fault+0xa 0x2c4baf0: 0x11fa5cd _ZN9NdmpdData7abortOpEv+0x253 0xf80e7ae0: 0x11fb42c _ZN16NdmpdDataAbortMH14messag eHandlerEPv +0x94 0xf80e7b10: 0x1223aec _ZN14NdmpConnection15processM essagesEi+0 x2c6 0xf80e7b40: 0x122275c _ZN14NdmpConnection15processRe questsEv+0 x14 0xf80e7b60: 0x11f29c3 _ZN21ConnectionFileHandler11h andleEventE v+0x1
protect the Data Mover from panicking under this condition
EMC Celerra Network Server Version 6.0.70.4 Release Notes
265
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 380146 / 36341620
1
At the end of NDMP backup/restore, two threads accessed the same object during cleanup. One thread was reset while the other was using it. This led to a Data Mover panic with the following Dart/panic fault message: *** Page Fault Interrupt. Virt ADDRESS: 9eed24 Err code: 2 Target addr: ce **. The stack trace showed: 0x2c3ab24: 0x14124c waitForReboot+0x90 0x2c3ab44: 0x1414d5 fault_dump+0x67 0x2c3ab64: 0x13a398 handle_fatal_fault+0xa 0x2c3ac50: 0x9ee354 _ZN13PaxDataModuleD0Ev+0x4c 0x37a91c30: 0x11f3192 _ZN22NdmpdDataModuleManager10de allocateEP1 5Nd 0x37a91c40: 0x11f145e _ZN9NdmpdData16deallocateModule EP15NdmpdDa taM 0x37a91c50: 0x11ede7e _ZN9NdmpdDataD1Ev+0x34 0x37a91c70: 0x12020e9 _ZN12NdmpdSessionD1Ev+0x63
Code now protects the cleanup during NDMP backup/restore to prevent this condition.
2
Running security scanner against NDMP connections using port 10000 and having connection closed not cleanly, left that connection open on Data Mover. Eventually, when 2048 such connection will be left open, that will result in the following server error: NDMP: 3: Active NDMP session number reaches the limit 2048, No more NDMP sessions can be started. Contact your Customer Support representative and any NDMP backup will fail from this point. A directory read by an NDMP backup triggered blocks to be flushed to the file system, which in turn caused the changes to be snapped to the SavVol. The oldest
Code has been fix to properly close the NDMP connection when network connection closed abnormally eliminating the problem.
380377 / 36311850
1
Code now allows the NDMP readdir operation to be aborted
375343 / 35697814
266
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
checkpoint had to be inactivated because the SavVol was full. However, the lock held by the NDMP backup on the file system prevented the unmount of the oldest checkpoint. This deadlock persisted until the panic was forced by the pending I/O. DART panic/fault message: >>PANIC: I/O not progressing LastVol touched Sn188 Kind 5 (ptr=0xace4c404). The stack trace showed: 0x6db1e90: 0x13fab4 waitForReboot+0x90 0x6db1eb0: 0x13fcb1 fault_dump+0x67 0x6db1ed0: 0x13fbb1 PANIC+0x29 0x6db1ee0: 0x822a30 _ZN9RawIO_IRP13checkProgressEv+ 0x11a 0x6db1fd0: 0x82290c _ZN17IrpProgressThread5startEv+ 0x6 0x6db1fe0: 0x14424c Sthread_startThread_internal+0x e
and retried when the checkpoint operation has completed.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
267
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 376319
1
During NDMP backup, in some rare conditions, a tape error led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../include/stream.hxx at line: 979 : datab::freeRef: db_ref <= 0 . The stack trace showed: 0xbccc5e50: 0x13f288 waitForReboot+0x90 0xbccc5e70: 0x13f485 fault_dump+0x67 0xbccc5e90: 0x13f385 PANIC+0x29 0xbccc5ea0: 0x1b407a _ZN5datab7freeRefEv+0x32 0xbccc5eb0: 0x96dc41 _ZN11MsgLinkList5EmptyEv+0x71 0xbccc5ed0: 0x970a76 _ZN9rw_thread5startEv+0xcb0 0xbccc5fe0: 0x143a08 Sthread_startThread_internal+0x e
Code was modified to properly handle media errors and prevent this panic.
1
If the user created an NDMP account with text format, and put letter "A" into the account comment part, when DMA sent NDMP_CONNECT message with MD5 format, the decryption algorithm caused a Data Mover panic because the dividend became 0. DART panic/fault message: *** Divide exception. Virt ADDRESS: 11c036b ***. The stack trace showed: 0x2baf8ec: 0x140cec waitForReboot+0x90 0x2baf90c: 0x140f71 fault_dump+0x67 0x2baf92c: 0x139eec handle_fatal_fault+0xa 0x2baf9a0: 0x11c006e _ZN18NdmpdConnectAuthMH14mess ageHandlerE Pv+0x4 0xef84bb10: 0x11ec6f2 _ZN14NdmpConnection15processM
This is a wrong configuration. Code now handles this exception correctly and returns INVALID_ARGUMENT error.
390481 / 37563732
268
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
essagesEi+0 x2ba 0xef84bb40: 0x11eb388 _ZN14NdmpConnection15processR equestsEv+0 x14 0xef84bb60: 0x11bbee3 _ZN21ConnectionFileHandler11h andleEventE v+0x1 0xef84bb80: 0x11ed956 _ZN19NdmpEventDispatcher8disp atchEim+0x7 ce
Network
The following entries describe fixed network-related issues:
Severity Symptom Description Fix Summary Tracking Number 385326 / 36986690
1
A PCI bus failure caused the Neterion 10G NIC device to stop working. Undetected hardware failure of the Neterion 10G device resulted in an unusable port.
Code was modified to periodically check the hardware state of the Neterion device, and issue restart command if the device stops responding to I/O requests.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
269
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 382462 / 36615128
1
When using virtual ethernet comprised of multiple devices (in this case configured an FSN over two 4-port trunks), memory allocation triggered a watchdog panic during power up. DART panic/fault message: >>PANIC in file: ../sched.cxx at line: 840 : SYSTEM WATCHDOG. The stack trace showed: 0x20141ae4: 0x13fb94 waitForReboot+0x90 0x20141b04: 0x13fda9 fault_dump+0x67 0x20141b24: 0x13fc91 PANIC+0x29 0x20141b34: 0x141b41 system_watchdog+0x77 0x20141b44: 0x138966 _t_nmi+0x68 0x20141ba0: 0x30a072 _Z7LM_HaltP16_LM_DEVICE_BLOCK+0 x1ea 0x20141bd0: 0x307b20 _Z15LM_ResetAdapterP16_LM_DE VICE_BLOCK+0 x48 0x20141c00: 0x306392 _Z20LM_InitializeAdapterP16_L M_DEVICE_BL OCK+0x
Code now separates the start into individual threads, causing the RT lock to be released between each device's start processing, thus avoiding the watchdog panic.
2
User reported slow response time when mapping a Data Mover CIFS share or when mounting an NFS export. This issue only occurred when the Data Mover DNS client was explicitly configured with the TCP protocol (server_dns configures UDP by default). The DNS client used for discovering Domain Controllers and query for hostname addresses did not properly interpret two DNS responses included in the same TCP packet. In some instances, the device "opaque" fields (ones that the device is supposed to copy unmodified) experienced single bit corruptions. DART panic/fault message: *** Page Fault Interrupt. Virt
The DNS client code was modified to properly handle two DNS responses received in a same TCP packet.
399644 / 38312216
1
The fix is to replicate the information so that it can be quickly verified when received.
377651 / 36005286
270
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
ADDRESS: 322e53 Err code: 0 Target addr: c **. The stack trace showed: 0x2ba3de4: 0x140ce4 waitForReboot+0x90 0x2ba3e04: 0x140f69 fault_dump+0x67 0x2ba3e24: 0x139ee4 handle_fatal_fault+0xa 0x2ba3e90: 0x31c7a8 _Z20LM_ServiceInterruptsP16_L M_DEVICE_BL OCK+0x1 0x4e81f80: 0x30fecf _ZN12bcmnicDevice5bcmRTEiPv+0x4 5 0x4e81fb0: 0x14226f _ZN15SchedRT_Initial5startEv+0x f3 0x4e81fe0: 0x1455cc Sthread_startThread_internal+0x e 1 Under heavy load, the Neterion 10G NIC hung, which led to loss of CIFS/NFS access, and logged many errors similar to the following in the server log: 2010-07-07 21:43:00: DRIVERS: 6: fxg-1-0 : Rescheduling RT. Enhanced recovery mechanism to fix the issue on Neterion 10G with faulted hardware. 375769 / 35590150
EMC Celerra Network Server Version 6.0.70.4 Release Notes
271
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 386273
1
In a rare case, during the disconnection of an SSL session, a certain sequence of network packets in the SSL module stream caused a Data Mover panic with the following: DART panic/fault message: Assertion failure: '_RDp != __null' on line 695 in file '../../dart_ssl/ssl_str.cxx'>>P ANIC in file: ../assert.c at line: 26 : assert failed. The stack trace showed: 0x77ddea0: 0x14a464 waitForReboot+0x90 0x77ddec0: 0x14a703 fault_dump+0x67 0x77ddee0: 0x14a5d5 PANIC+0x29 0x77ddef0: 0x23de29 _assert+0x51 0x77ddf10: 0x11c9a4b _ZN10ssl_strCtx7wputSslEv+0x37 0x77ddf60: 0x11c9e99 _Z11sslstr_wsrvP5queue+0x4d 0x77ddf80: 0x20b593 _Z10STRserviceiPv+0x1af 0x77ddfa0: 0x14bba6 _ZN15SchedRT_Initial5startEv+0x 132
SSL module stream function now handles the sequence of network packets correctly.
NFS
The following entries describe fixed issues related to Network File Systems:
Severity Symptom Description Fix Summary Tracking Number 388304
2
Write performance was reduced when file system free space was limited.
Enhanced the search for free space by selecting cylinder groups by their availability instead of a predefined order.
272
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 385026
2
When a file system was mounted with option “ntcredential”, the UNIX command chgrp failed on objects with owners that were members of more than 16 groups. The Data Mover took into account only groups that were part of NFS credentials, ignoring other groups specified in NT credentials.
The Data Mover now takes into account both NFS credentials and NT credentials when file system is mounted with option “ntcredential” and parameter cifs.acl.extendExtraGid is set. The limit of total number of groups is defined by parameter security.maxgroups. Modified the way files are allocated to optimize performance. Made the removal of these locks faster, so that the system watchdog is no longer triggered under this condition.
1
File system performance dropped when the files in the file system were not allocated correctly. When an NFSv3 client notifies the Data Mover that it has just rebooted, the Data Mover must revoke all the locks owned by the previous instance of this client. The removal of these locks triggered a system watchdog panic when the count of locks was huge (30K). DART panic/fault message: >>PANIC in file: ../sched.cxx at line: 846 : SYSTEM WATCHDOG. The stack trace showed: 0x5085f24: 0x141254 waitForReboot+0x90 0x5085f44: 0x1414dd fault_dump+0x67 0x5085f64: 0x1413c5 PANIC+0x29 0x5085f74: 0x1432d1 system_watchdog+0x77 0x5085f84: 0x139f72 _t_nmi+0x68 0x5085fe0: 0x145bd6 Sthread_startThread_internal+0x e
397730
1
401962 / 38796804
EMC Celerra Network Server Version 6.0.70.4 Release Notes
273
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 393021
1
If an NFS export was created with readonly host ip as 99.99.99.99, creating the NFS export failed with the error message: Modify NFS export server_2:/sanity_f123. server_2 : Invalid argument Export Error: Unknown option details. When the same process was done for Access hosts, it created the NFS export successfully. User could not set permissions via NFSv4 clients on files and folders with ACL containing ACE of Windows local SYSTEM account. The following error appeared in server_log: NFS: 3: nfs4_checkSpecialAce: SID S-1-5-12 cannot be translated to NFSV4 : ACL can't be modified.
Code was modified to ensure that exporting a path as read-only to specified hosts work correctly.
2
A new parameter nfsv4.ntAllSpecialUser s was added. When this parameter is set to 0 (default), Windows special users that are not listed in the RFC 3530 are returned as 'nobody' and changing permissions on objects with ACL containing local SYSTEM account will fail. When this parameter is set to 1, Windows special users are translated to a unique identifier, this allows modification of ACL for such objects from NFSv4 clients. Code has been modified and a better mutex protection has been added to export database initialization.
387582
1
During NFSv4 export database initialization, the server panicked with the following DART panic message: *** Page Fault Interrupt. Virt ADDRESS: 0x000179fb2e Err code: 0 Target addr: 0x0000000078 **
396771
1
An NFSv4 client was sending a request to confirm the client ID to the server. In some cases, the Data Mover had already confirmed the ID, and didn’t know how to handle this situation. The Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 631ae3
Code has been fixed to handle this request by returning the following error message to the client: ID already in use and the Data Mover no longer panics in this situation.
384098 / 36843082
274
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
Err code: 0 Target addr: 28 **. The stack trace showed: 0x2acfb34: 0x13fb94 waitForReboot+0x90 0x2acfb54: 0x13fda9 fault_dump+0x67 0x2acfb74: 0x138d94 handle_fatal_fault+0xa 0x2acfbc0: 0x62f76c _ZN16NFSv4_ClientList7Confirm EyPA8_KhR8n fs4_ctx 0x433fdcb0: 0x62822b _Z24nfs4_setclientid_confirmR P4msgbS1_RP hR8nfs 0x433fdd10: 0x622de0 _Z23nfs4_convertNfsv3Statusm+0x 656 0x433fde90: 0x657949 _ZN13NfsGssRequest8dispatchEP 4msgbPvP17A ccess_ 0x433fdee0: 0x605e16 _ZN17nfs_serverInitial5startEv+ 0x5d0
EMC Celerra Network Server Version 6.0.70.4 Release Notes
275
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 385131 / 36960608
1
During nfsv3_lookup, if the path name length field in nfs header was corrupted, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 5e5e39 Err code: 0 Target addr: ffee8693 **.The stack trace showed: 0x28a6c64: 0x13fa98 waitForReboot+0x90 0x28a6c84: 0x13fc95 fault_dump+0x67 0x28a6ca4: 0x138cfc handle_fatal_fault+0xa 0x28a6e20: 0x5e2436 _ZN17nfs_serverInitial5startEv+ 0xac2 0xdefd1fe0: 0x144230 Sthread_startThread_internal+0x e
To prevent this panic, string length verification was added to verify the length field of NFS header of the request received from the network.
Replication
The following entries describe fixed problems encountered while using Celerra Replicator, SnapSure or Checkpoints:
Severity Symptom Description Fix Summary Tracking Number 387170 / 37192564
1
A RepV2 session could not be stopped when the transfer was waiting for the destination SavVol to be extended.
Added a check for the stop request to continue during SavVol extension. This will allow a stop to proceed when the session is waiting for the destination SavVol to be extended. Code was fixed to prevent the checkpoint on the destination file system from becoming
1
Repv2 suspends the writes to the destination file system when the SavVol is being extended. However, if the current SavVol usage value was equal to the
388664 / 37379978
276
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
HWM (high watermark), writes were not suspended as expected. This caused the destination checkpoints to become inactive and if the file system on the destination was being accessed, an inconsistent view of the file system led to a Data Mover panic with the following DART panic/fault message: mangled directory ent. The stack trace showed: 0xcfdd56c0: 0x14124c waitForReboot+0x90 0xcfdd56e0: 0x1414d5 fault_dump+0x67 0xcfdd5700: 0x1413bd PANIC+0x29 0xcfdd5710: 0x778757 _ZN14UFS_FileSystem28markCorr uptedFsForU nmount 0xcfdd58d0: 0x7e272c _ZN9UFS_Vnode6dirbadEmPKc+0xea 0xcfdd5990: 0x7c0ee7 _ZN20UFS_DirHashInfo_dir3C1EP 11Local_Vno de+0x5 0xcfdd5a10: 0x7e5b04 _ZN15UFS_DirOps_dir311initDir HashEP9UFS_ Vnode+ 0xcfdd5a40: 0x7d6ef6 _ZN15UFS_DirOps_dir37scandirE P9UFS_Vnode PK18Fi 1 When using NFSv4, users were unable to copy directories on Data Mover shares from AIX clients using relative pathnames (for example, using “cp –r” command). The following error was reported by the client: No such file or directory. The issue occurred due to incorrect MOUNT_FILEID value returned to AIX client in NFSv4 request.
inactive under this condition.
Data Mover now returns the correct MOUNT_FILEID value to NFSv4 clients allowing successful directory copies.
389170 / 37129124
EMC Celerra Network Server Version 6.0.70.4 Release Notes
277
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 394547 / 37882060
1
Replication recovery tasks were not marked as complete if the destination Data Mover could not communicate with the source Data Mover during a VDM replication failover. A subsequent reboot of the Data Mover failed to mount the root file system of the VDM due to the presence of the incomplete tasks.
The code now checks whether a remount of the file system is needed during taskManager recovery, remounts the required file systems only, and correctly updates the status of the tasks. Code has been modified to handle the un-aligned 8KB Read/Write I/Os to writable checkpoints correctly.
1
When using writable checkpoints and when un-cached I/O was used (in cases such as with VMWare/VAAI), Data Mover panicked with the following DART panic/fault message: >>PANIC: I/O not progressing LastVol touched Sh148108 Kind 8 (ptr=0x57d42204). The stack trace showed: 0x7c3de90: 0x14124c waitForReboot+0x90 0x7c3deb0: 0x1414d5 fault_dump+0x67 0x7c3ded0: 0x1413bd PANIC+0x29 0x7c3dee0: 0x8ad1b6 _ZN9RawIO_IRP13checkProgressEv+ 0x11a 0x7c3dfd0: 0x8ad092 _ZN17IrpProgressThread5startEv+ 0x6 0x7c3dfe0: 0x145bfa Sthread_startThread_internal+0x e
386603
1
When the virtually provisioning options were modified on a file system replicated with Repv2, the replication database was not updated with the new settings. Therefore, after the Data Mover was rebooted, the old settings were applied to the file system. This resulted in the virtually provisioned file system not being extended.
The nas_fs - modify command now modifies the mount options correctly.
384264 / 36813726
278
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 115393
2
The nas_replicate -stop and -delete operations required the user to enter a "mode" (set to both, source, or destination).
The nas_replicate command was enhanced so that the stop and -delete options default to 'both'. Code has been fixed to use the correct schedule for mounting the checkpoint.
2
When creating a checkpoint schedule, if a user changed the relative checkpoint from time based to named and then tried to mount the checkpoint, the checkpoint was mounted with the time based schedule instead on the named schedule.
378004
EMC Celerra Network Server Version 6.0.70.4 Release Notes
279
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 391982 / 37718380
1
During the merge of an empty checkpoint into an older checkpoint, after all the needed mappings were moved and the stableEntry was 2048*N. After a reboot, the wrong table of contents (TOC) buffer was flushed and the TOC became corrupt. This led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../DeltaVol.cxx at line: 7550 : Value in TOC is different from the value in btree. The stack trace showed: 0x9cb85df0: 0x14124c waitForReboot+0x90 0x9cb85e10: 0x1414d5 fault_dump+0x67 0x9cb85e30: 0x1413bd PANIC+0x29 0x9cb85e40: 0xfb1e7c _ZN19PoolManagedDeltaVol14verifyTO CValue Emy+0x 0x9cb85ea0: 0xfb1805 _ZN19PoolManagedDeltaVol10ver ifyTOCsEym1 1Verif 0x9cb85f00: 0xfb14fe _ZN19PoolManagedDeltaVol5merg eERP5Event+ 0xcf2 0x9cb85f80: 0xfc6d3c _ZN11mergeThread5startEv+0x1b4 0x9cb85fe0: 0x145bfa Sthread_startThread_internal+0x e
Code has been fixed to avoid the panic and to make the TOC pointer point to the right TOC buffer to be flushed.
3
The nas_replicate-info command always shows the "Next Transfer Size" field to have the same value as "Current Transfer Size" field.
The Next Transfer Size (KB) field in the output of command nas_replicate -l was removed. Code now skips the empty signatures check during one-tomany replication session to avoid this
370770 / 34923460
1
During 1-to-Many replication sessions, empty signatures check prevented the source VDM from being changed to mounted state.
386277
280
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
issue. 1 A deadlock condition occurred when RDE layer queries for SavVol stats while a checkpoint restore is in progress. The deadlock occurs during the dereference process in the DpLayer. Code no longer requires the RDE thread to need a version set lock during the dereference process, hence avoiding a deadlock. When unknown clients attempt to connect to port 8888, the sessions are now properly closed and cleaned up. 395215
1
If an unknown client (e.g. a port scanner) connected to RCP 8888 port too fast, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0 Err code: 0 Target addr: 0 ** . The stack trace showed: 0x2ba3d10: 0x140ce4 waitForReboot+0x90 0x2ba3d30: 0x140f69 fault_dump+0x67 0x2ba3d50: 0x139ee4 handle_fatal_fault+0xa 0x2ba3db0: 0x1d10d4 CloseStream+0x1b6 0x1ff3deb0: 0x12ce0b8 _ZN18RCP_SessionManager6accep tEPvP11sock addr_ 0x1ff3df40: 0x12ca690 _ZN10RCP_Server5startEv+0x3e6 0x1ff3dfe0: 0x1455cc Sthread_startThread_internal+0x e
384423 / 36899914
1
Replication transfer did not make progress as the SavVol on the destination could not be extended. User did not have enough information to understand why replication transfer did not progress.
An event is now posted to inform the user of the SavVol extension progress.
380166 / 36340862
EMC Celerra Network Server Version 6.0.70.4 Release Notes
281
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 373743
1
A checkpoint was stuck at Frozen state when the following three conditions occurred simultaneously: a PFS was being extended, checkpoint was being refreshed and SavVol was full. This led to blocking of snap IOs and triggered the following: DART panic/fault message: >>PANIC: I/O not progressing LastVol touched not Known The stack trace showed: 0xdfe5df0: 0xc6c7bd _ZN9RawIO_IRP13checkProgressEv+ 0x3e9 0xdfe5fc0: 0xc6cb3e _ZN17IrpProgressThread5startEv+ 0x4 0xdfe5fd0: 0x16ad51 Sthread_startThread_internal+0x
Code was modified to ensure that when there are pending unmounting events for a checkpoint, the initFS allows mounting. This prevents the failure of checkpoint thaw for potential ongoing checkpoint refresh operation.
2
There was potential for slower than expected network transmit of RepV2 data when manually setting RCP TCP params.
Corrected an arithmetic error, and the decision logic for when to change the network transmit window. The "Max Out of Sync Time" setting is now preserved and will be applied to the session when it is restarted after a failover/switchover. In the HTTPS server of the Data Mover, there was no timeout during the establishment of a SSL connection (between the acceptance of the TCP connection and the first packet sent by the client). The Data Mover was waiting for
381470 / 35658014
1
When a failover/switchover is executed on a replicationV2 session followed by a start in the reverse direction, the session will be set for manual refresh regardless of original "Max Out of Sync Time" setting.
389459
1
While replicating one Celerra to two other Celerras, the Data Mover on one of the destination Celerras did not clean up the connections to port 5085 and eventually the server_log became full of Too many connections on ports 5085 queue. The Data Mover on the destination file system panicked with the following DART panic/fault message: >>PANIC in file: ../dskdump.cxx at line: 1836 : Too many connections on
381594 / 36464168
282
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
port 5085. The stack trace showed: 0x1facbae0: 0x141254 waitForReboot+0x90 0x1facbb00: 0x1414dd fault_dump+0x67 0x1facbb20: 0x1413c5 PANIC+0x29 0x1facbb30: 0xa44c58 _Z11paniccfgcmdR6Option+0x178 0x1facbc70: 0x1cbb8b _ZN11cfgcmd_desc6invokeEPKcPP 8CCMD_Msgii +0x1a7 0x1facbce0: 0x2b2b3e _Z10xml_cfgcmdP7macDataRP8CCMD_ Msg+0x392 0x1facbd30: 0x2adaf0 _ZN17macCCMDCommand_DB6invoke EPKcP7macDa taRP8C 0x1facbd60: 0x2adff6 _Z17mac_CCMDDocParserP7macDat aRP8CCMD_Ms g+0x20 1 After rebooting, the Data Mover would hang during the checkpoint recovery process and remain in "Loaded" state. Data Mover failovers would also fail.
this first packet (the SSL clientHello), and until this packet was received, the HTTPS didn't process any new connection request. The HTTPS server was blocked and unavailable. This has been fixed.
A deadlock occurred while trying to correct a rare form of checkpoint corruption. The code now unlocks locked memory correctly, avoiding the deadlock.
378406 / 35911666
EMC Celerra Network Server Version 6.0.70.4 Release Notes
283
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 383291 / 36431708
1
A function in the code was consuming stack memory, which caused a stack overflow and led to a Data Mover panic during replication. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 1fa53e Err code: 0 Target addr: 0 **. The stack trace showed: 0x2acb158: 0x13fb94 waitForReboot+0x90 0x2acb178: 0x13fda9 fault_dump+0x67 0x2acb198: 0x138d94 handle_fatal_fault+0xa 0x2acb1c0: 0x56e267 krb5_cc_resolve+0x17 0x1eca92b0: 0x56e609 krb5_cc_default_suffix+0xa3 0x1eca92e0: 0x53e1db _Z26krb5_gss_release_principalP jPPv+0x46b 0x1eca93a0: 0x53ed90 _Z25krb5_gss_acquire_cred_ext PjPv22gss_b uffer_ 0x1eca9400: 0x5456ff gss_acquire_cred_ext+0x4b
Code was modified to reduce memory consumption on the stack by using heap memory instead, preventing this condition.
1
When a checkpoint was being refreshed or deleted the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../DeltaVol.cxx at line: 7550 : Value in TOC is different from the value in btree. The stack trace showed: 0x2971dd70: 0x141254 waitForReboot+0x90 0x2971dd90: 0x1414dd fault_dump+0x67 0x2971ddb0: 0x1413c5 PANIC+0x29 0x2971ddc0: 0xfaae5e _ZN19PoolManagedDeltaVol14ver ifyTOCValue Emy+0x
The checkpoint merge process now correctly handles the case when the merge has been interrupted by a Data Mover reboot.
381357
284
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
0x2971de20: 0xfaa7e7 _ZN19PoolManagedDeltaVol10ver ifyTOCsEym1 1Verif 0x2971de80: 0xfac73f _ZN19PoolManagedDeltaVol11cha ngeToTOCEv+ 0x8b 0x2971deb0: 0xfac110 _ZN19PoolManagedDeltaVol13fin dFirstFreeE RyRP5E 0x2971df00: 0xfa9c62 _ZN19PoolManagedDeltaVol5mergeE RP5Event+0x 474
EMC Celerra Network Server Version 6.0.70.4 Release Notes
285
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 377960 / 35997088
1
There was a race between SnapSure smart snap and delete command. Smartsnap traversed from the latest ckpt to oldest ckpt to decide whether a block is a smartsnap block, while at the same time, the oldest ckpt was deleted by user command. During smart snap, code needed to find mapping from the oldest ckpt and only found it was destructed, then panicked the Data Mover with the following DART panic/fault message: *** GP exception. Virt ADDRESS: b877dd. Err code: 0 *** . The stack trace showed: 0x172f458: 0x13db94 waitForReboot+0x90 0x172f478: 0x13dd91 fault_dump+0x67 0x172f498: 0x137d54 handle_fatal_fault+0xa 0x172f4e0: 0xb91b91 _ZN15snapshotContext10BlockIn UseEP8Delta VolyP8S 0x242e9be0: 0xb7d6dc _ZN19PoolManagedDeltaVol18wri teBlockFilt eredEP 0x242e9c90: 0xb8d232 _ZN10SnappedVol18writeBlockFi lteredEP10D iskIO_ 0x242e9cb0: 0xb8d1ed _ZN10SnappedVol18writeBlockIn ternalEP10D iskIO_ 0x242e9cd0: 0x731731 _ZN11HyperVolume18writeBlockI nternalEP10 DiskIO
This fix is to close this race by suspending smart snap if there is a delete operation in progress.
1
The Data Mover experienced repeated Watchdog panics when an attempt was made to create the first ReplicationV2 session. This was because the Control Station to DART interconnect did not exist. DART Server183ic/fault message: >>Server183IC in file: ../sched.cxx at line: 846 : SYSTEM WATCHDOG. The stack
During a Data Mover reboot, a check will be performed to verify that the Control Station to DART interconnect is present. If the interconnect is missing it will be added by running "nas_cel -
371904 / 35235954
286
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
trace showed: 0x5085f24: 0x14124c waitForReboot+0x90 0x5085f44: 0x1414d5 fault_dump+0x67 0x5085f64: 0x1413bd Server183IC+0x29 0x5085f74: 0x1432c9 system_watchdog+0x77 0x5085f84: 0x13IP_ADDR_nmi+0x68 0x5085fe0: 0x145bfa Sthread_startThread_internal+0x e
update id=0".
EMC Celerra Network Server Version 6.0.70.4 Release Notes
287
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 370442
1
A SavVol was almost full, and the most recent ckpt was about to be inactivated to save space. A snap request on this ckpt happened just before the deactivation, which caused an invalid function in the code and led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../DeltaVol.cxx at line: 3792 : doSavVolIO failure. The stack trace showed: 0x236025e60: 0x1309df4 _ZN19PoolManagedDeltaVol10doS avVolIOEyjP ci16 0x236025eb0: 0x130ed59 _ZN19PoolManagedDeltaVol22Flus hSnapShotL ogLo 0x236025ef0: 0x130f0c7 ZN19PoolManagedDeltaVol16Flus hSnapShotLo gE1 0x236025f20: 0x12fcdf5 _ZN12IntentLog_CB8allocateEv+0x 15 0x236025f40: 0x1348c65 _ZN21ckptBlockAllocatorThr5st artEv+0x375 0x236025fd0: 0x15519f Sthread_startThread_internal+0x 7
This issue was fixed by adding a hdr check condition to FlushSnapShotLogLoc ked, thus avoiding a panic.
1
There was a race condition between ckpt restore complete routine and ckpt delete command. SavVol was full during ckpt restore process, it erased all the ckpts on the SavVol, while a ckpt on this SavVol was being deleted by user's command, it double-destructed the ckpt which led to a panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: f8d2d0 Err code: 0 Target addr: 0 **. The stack trace showed: 0x2ba3bd4: 0x140ce4 waitForReboot+0x90
A lock mechanism was added to avoid the race condition.
159333
288
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
0x2ba3bf4: 0x140f69 fault_dump+0x67 0x2ba3c14: 0x139ee4 handle_fatal_fault+0xa 0x2ba3c70: 0xf8cf26 _ZN19PoolManagedDeltaVol18Rel easeChunkBi tMapEv+ 0x1f453d60: 0xf7f332 _ZN19PoolManagedDeltaVol15Rel easeSnapsho tEv+0x 0x1f453dc0: 0xf7aa74 _ZN19PoolManagedDeltaVol19SetSnap ShotInA ctiveE 0x1f453e40: 0xf94af2 _ZN15snapshotContext17BroadCa stInactiveE PKcP8D 0x1f453e80: 0xf802a8 _ZN19PoolManagedDeltaVol11Res toreStopEv+ 0xa4 1 A replication stop/delete/failover operation task hung on the source side of replication, if the destination side was already stopped or deleted. The task indicated the following error: Error 13160415328: Registration request rejected. Code has been modified to ensure replication commands such as stop/delete/failover no longer hang when the destination side is stopped or deleted. 380942
EMC Celerra Network Server Version 6.0.70.4 Release Notes
289
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 388310 / 37319436
1
A Data Mover hung when an SMB thread tried to close and compress a file on a file system to which a replication thread was trying to unmount. DART panic/fault message: >>PANIC in file: ../dskdump.cxx at line: 1836 : cifs block. The stack trace showed: 0x2128bae0: 0x14124c waitForReboot+0x90 0x2128bb00: 0x1414d5 fault_dump+0x67 0x2128bb20: 0x1413bd PANIC+0x29 0x2128bb30: 0xa450c8 _Z11paniccfgcmdR6Option+0x178 0x2128bc70: 0x1cbdab _ZN11cfgcmd_desc6invokeEPKcPP 8CCMD_Msgii +0x1a7 0x2128bce0: 0x2b205a _Z10xml_cfgcmdP7macDataRP8CCMD_ Msg+0x392 0x2128bd30: 0x2ad00c _ZN17macCCMDCommand_DB6invoke EPKcP7macDa taRP8C 0x2128bd60: 0x2ad512 _Z17mac_CCMDDocParserP7macData RP8CCMD_Ms g+0x20
Replication code now verifies that the file system is not locked by any other threads under such conditions before trying to unmount a file system.
1
A bug which allowed a null value to be returned to SnapSure led to the following Data Mover panic: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: fa59c6 Err code: 0 Target addr: 4 **. The stack trace showed: 0x2c3b184: 0x14124c waitForReboot+0x90 0x2c3b1a4: 0x1414d5 fault_dump+0x67 0x2c3b1c4: 0x13a398 handle_fatal_fault+0xa 0x2c3b260: 0xfbabe0 _ZN10SnappedVol18writeBlockFi
The code has been fixed so that a NULL value is no longer returned, and no panic results.
382916 / 36634308
290
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
lteredEP10D iskIO_I 0xe9871240: 0xfbb091 _ZN10SnappedVol16aioInternalW riteEP10Dis kIO_IR 0xe98712a0: 0xfbaca8 _ZN10SnappedVol11aioInternalE P10DiskIO_I RPP8ai 0xe98712d0: 0x8c2361 _ZN11HyperVolume11aioInternal EP10DiskIO_ IRPP8a 0xe98713b0: 0x8d4007 _ZN17UnCachedIO_Object10lio_listi oE7LioM odePP8
EMC Celerra Network Server Version 6.0.70.4 Release Notes
291
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 383407 / 36730552
1
If an application label containing spaces was used to create an iscsi LUN replication, when user issued a reverse command, the Data Mover panicked with the following Assertion failure: 'dicSessionp != NULL' on line 931 in file '../DpTaskManager.cxx' >>PANIC in file: ../assert.c at line: 26 : assert failed. The stack trace showed: 0xe0031e30: 0x14124c waitForReboot+0x90 0xe0031e50: 0x1414d5 fault_dump+0x67 0xe0031e70: 0x1413bd PANIC+0x29 0xe0031e80: 0x1ff975 _assert+0x51 0xe0031ea0: 0x10c63c8 _ZN13DpTaskManager20createSyn cDicSession ERP15 0xe0031ee0: 0x10cba57 _ZN33DpTaskManagerCopyReverse Initiator8s tartS 0xe0031f30: 0x10caa07 _ZN33DpTaskManagerCopyReverse Initiator14 start 0xe0031f50: 0x10c844e _ZN33DpTaskManagerCopyReverse Initiator14 Start
The application label is now validated and rejected if it contains spaces.
Security
The following entries describe fixed Security related issues:
Severity Symptom Description Fix Summary Tracking Number 393878
1
When simultaneous access by two threads to the same SSL certificate was attempted, the Data Mover panicked with the following DART panic/fault message:
Added a mutex lock to protect the access to the SSL certificate and
292
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
>>PANIC in file: ../malloc.cxx at line: 351 : Memory corrupted. The stack trace showed: 0xe5fefc40: 0x140ce4 waitForReboot+0x90 0xe5fefc60: 0x140f69 fault_dump+0x67 0xe5fefc80: 0x140e51 PANIC+0x29 0xe5fefc90: 0x1fbdb7 free+0x69 0xe5fefcc0: 0xda5b38 CRYPTO_free+0xc 0xe5fefcd0: 0xd3fe2d BUF_MEM_free+0x2c 0xe5fefcf0: 0xd918ba X509_NAME_dup+0xe5 0xe5fefd10: 0xd8c7a1 ASN1_item_ex_free+0x171 2 The server_ldap command gave no indication whether the LDAP Service configuration succeeded. User needed to do additional queries or check the server logs to find out.
prevent this condition.
A new alert-level event was added for LDAP Service configuration failures. If a configuration fails, the following message will be in the sys_log and (if configured) emailed to the user: An attempt to configure the LDAP Service has failed. See the server log for applicable error messages. Correct the error, then retry the configuration command..
149215 / 29904638
EMC Celerra Network Server Version 6.0.70.4 Release Notes
293
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 385248
1
In version 6.0, code was added to log an event if an LDAP service configuration attempt failed. However, the code was added to several error loops that included 20 second delays before retrying the configuration attempt. the result was that the 20 second delay is always run whether there is an error or not, causing a delay in the service configuration. A race condition between periodic check for reconfiguration and manual reconfiguration may lead to an LDAP deadlock.
Fixed the error while loops to only run the 20 second delay if an error occurs.
1
This issue occurred when performing continuous LDAP reconfigurations in a short period of time. The LDAP timer no longer signals the event if already signaled.
383944
SRDF
The following entries describe fixed Symmetrix Remote Data Facility related issues:
Severity Symptom Description Fix Summary Tracking Number 371759
3
The fs_timefinder -M off command running against fs_group failed with the following error message. Error 13431668739: File system group has a file system which is backup of a Primary File System (PFS) which is actively mirroring to file system <fsname>. During a nas_rdf -restore, a gatekeeper lock condition should have forced the command to exit, the command was allowed to continue due to a unique condition. This caused the nas_rdf restore command to hang.
Removed the check which caused this issue. The fs_timefinder -M off command with fs_group now works correctly. The nas_rdf command will now exit when excessive lock contention is experienced. If a gatekeeper lock will not release, identify the lock and manually
1
382490 / 36173416
294
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
release it. Identify the pid holding the lock with the following command: /usr/symcli/bin/symcf g list -sem -v Release the lock with the following command: ipcrm -s <pid> Then rerun the nas_rdf -restore command.
Storage
The following entries describe fixed storage-related issues:
Severity Symptom Description Fix Summary Tracking Number 382625 / 36666666
1
When the source storage system was modified after the destination had been activated, the nas_mview -restore command failed.
Enhancements to the nas_mview command have been applied to detect storage system changes allowing the restore to complete successfully.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
295
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 376685
1
When paths to a device failed, and some (administrative) action caused a disk signature to be verified, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000a43da9 Err code: 0 Target addr: 0x0000000010 **. The stack trace showed: 0x3680760: 0xbf6450 _ZN7DskDump9checkdumpEPKc+0x7a 0x36807a0: 0xbf665e dump_return+0x4 0x36807b0: 0x14c7d2 _ZN12Memory_Frame13setProtect ionEm19Memo ry_Prot 0x36807d0: 0x14ce5d fault_dump+0x89 0x36807f0: 0x13378d handle_fault+0x1d 0x689fd20: 0xa4aec7 _ZN11BasicVolume21insertToIoQ ueueCommonE P8SgIO_ 0x689fd60: 0xa53518 _ZN11BasicVolume15IOBlockInternalEP 10Dis kIO_IRP 0x689fdf0: 0xa57ab9 _ZN11BasicVolume18writeBlockI nternalEP10 DiskIO_
Fixed a locking bug which caused stale queue data to be referenced.
1
When using UDP-based applications with large xfer sizes or using TCP and network environmental conditions caused fragmentation to occur, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x00003cb317 Err code: 0 Target addr: 0x0000000006 **. The stack trace showed: 0x3680760: 0xbf6490 _ZN7DskDump9checkdumpEPKc+0x7a 0x36807a0: 0xbf669e dump_return+0x4
Updated code to correct issues with fragment list management and prevent this pan
388143 / 37288224
296
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
0x36807b0: 0x14c7d2 _ZN12Memory_Frame13setProtect ionEm19Memo ry_Prot 0x36807d0: 0x14ce5d fault_dump+0x89 0x36807f0: 0x13378d handle_fault+0x1d 0x6880d90: 0x3c08e7 ip_mxrput+0x10d 0x6880dd0: 0x37b4bd _Z7bcmreadPvP4msgbtjj+0x4e7 0x6880ee0: 0x38b2b2 _Z20LM_ServiceInterruptsP16_L M_DEVICE_BL OCK+0x3 1 Periodically the Data Mover NIS client was unable to connect to the NIS server and switched to another NIS server. The server log reported an error like: The NIS client is unable to connect to the NIS server w.x.y.z : RPC failure. In some instances the NIS server can be slow to respond so the NIS RPC retry timeout is increased to 5 seconds and the NIS RPC request timeout increased to 15 seconds. This will prevent switching to another NIS server when there is a brief delay in the response from the NIS server. Checks were added in the code and error messages enhanced to determine what needs to be done to diskmark different configurations. 375293 / 34806928
1
User removed 5 LUNs from Celerra to change HLU’s (host LUN unit) and was unable to import them back even though no errors were logged. LUNs were part of a MirrorView/S consistency group and the command nas_mview -init needed to be used to diskmark them, but the Celerra error message gave no indication of this.
126624 / 25702698
EMC Celerra Network Server Version 6.0.70.4 Release Notes
297
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 372043 / 35302478
1
The Data Mover received incorrect inquiry data from a backup tape device, which later triggered a panic due to dereferencing of a null volume pointer. The Data Mover panicked with the following DART panic/fault message: ** Page Fault Interrupt. Virt ADDRESS: 8b341f Err code: 0 Target addr: 0 **. The stack trace showed: 0x2c2b494: 0x141254 waitForReboot+0x90 0x2c2b4b4: 0x1414dd fault_dump+0x67 0x2c2b4d4: 0x13a3a0 handle_fatal_fault+0xa 0x2c2ba60: 0x296c17 _Z16macthr_sysconfigP4msgb+0x1b 5f 0x21e25fa0: 0x27c52a _ZN11Mac_Initial5startEv+0x3e0
The fix avoids nullpointer dereferencing when invalid Inquiry data is returned from a device, preventing the panic.
1
An attempt to specify more than 8 paths on a volume disk command resulted in a Data Mover panic with the following DART panic/fault message: *** GP exception. Virt ADDRESS: 13990eb. Err code: 0 ***. The stack trace showed: 0x2c4bcb8: 0x14124c waitForReboot+0x90 0x2c4bcd8: 0x1414d5 fault_dump+0x67 0x2c4bcf8: 0x13a398 handle_fatal_fault+0xa 0x2c4bd20: 0x1399061 _ZN14XmlElementNodeD0Ev+0x1d 0x25429cf0: 0x13991ca _ZN14XmlElementNode14removeCh ildrenEv+0x 2c 0x25429d10: 0x1399059 _ZN14XmlElementNodeD0Ev+0x15 0x25429d20: 0x13991ca
An attempt to specify more than 8 paths did not fail but caused memory corruption and a Data Mover panic. The code that parses volume disk commands has been modified to explicitly check and fail with the following error if more than 8 paths are specified: logIO::logmsg (LOG_STORAGE, LOG_ERR, 5825 logIO::logmsg (LOG_STORAGE, LOG_ERR "Cannot create Basic Volume %s, numpaths parameter invalid %d\n .
378239
298
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
_ZN14XmlElementNode14removeCh ildrenEv+0x 2c 0x25429d40: 0x1398fc1 _ZN14XmlElementNodeD2Ev+0x15 1 When performing a PFS restore from the oldest checkpoint, if the SavVol became full leading to oldest checkpoint inactivation, the Data Mover panicked with the following DART panic/fault message: >>PANIC: I/O not progressing LastVol touched D6853119_0 Kind 8 (ptr=0x115ed0038. The stack trace showed: 0xc3addf0: 0xc566bd _ZN9RawIO_IRP13checkProgressEv+ 0x3e9 0xc3adfc0: 0xc56a3e _ZN17IrpProgressThread5startEv+ 0x4 0xc3adfd0: 0x16b4b1 Sthread_startThread_internal+0x 7 Code has been fixed to prevent the panic for this condition. 370887
Impact Level Symptom Desc.
Severity 1 Celerra kept reporting the trespassing of HDDs from SPA to SPB. Server log showed many CAM errors like this: 2010-01-11 13:11:42: 81874518019: CAM: 3: 3: Warning: The SCSI HBA 16 is inaccessible. 2010-01-11 13:11:42: 13155041280: CAM: 3: I/O Error: c16t0l0 Irp 0x57c25f04 CamStatus 0x05 ScsiStatus 0x08 Sense 0x05/0x04/0x00 2010-01-11 13:11:42: 13155041280: CAM: 3: I/O Error: c16t9l15 Irp 0x3c641104 CamStatus 0x05 ScsiStatus 0x08 Sense 0x05/0x04/0x00 and sys_log showed errors like this: Dec 8 23:40:28 2009:CS_PLATFORM:NaviEventMonitor:CRITICAL:101::::1260312028: Disk d8 has been trespassed. Dec 8 23:40:28
EMC Celerra Network Server Version 6.0.70.4 Release Notes 299
Version 6.0.41.3
Fix Service Request No.
2009:CS_PLATFORM:NaviEventMonitor:CRITICAL:101::::1260312028: Disk d11 has been trespassed. Dec 8 23:40:28 2009:CS_PLATFORM:NaviEventMonitor:CRITICAL:101::::1260312028: Disk d13 has been trespassed. Dec 8 23:40:28 2009:CS_PLATFORM:NaviEventMonitor:CRITICAL:101::::1260312028: Disk d14 has been trespassed. Enhanced the error status handling code to handle busy luns without forcing a trespass. 31963984
UFS
The following entries describe fixed UNIX File System issues:
Severity Symptom Description Fix Summary Tracking Number 382495 / 36659580
1
During the UTF-8 conversion process for a file system, if a Data Mover attempted to convert a directory owned by a user who was exceeding quotas, the Data Mover panicked with the following DART panic/fault message: Cannot read unnamed stream inode for container. The stack trace showed: 0xdf591310: 0x14124c waitForReboot+0x90 0xdf591330: 0x1414d5 fault_dump+0x67 0xdf591350: 0x1413bd PANIC+0x29 0xdf591360: 0x778b27 _ZN14UFS_FileSystem28markCorr uptedFsForU nmount 0xdf591520: 0x7cabba _ZN9UFS_Vnode18setUnnamedVnodeP trEv+0x192 0xdf591570: 0x777d74 _ZN14UFS_FileSystem16findExis tingNodeEmR 11File 0xdf5915c0: 0x777976 _ZN14UFS_FileSystem8findNodeE
Code was updated to be more tolerant of this error, and will no longer panic under this condition. The following error is returned if the conversion fails: UFS_Vnode::setUnna medVn odePtr -- fs <fsid> inum <inode number> Quota Exceeded\n where 'Quota Exceeded' can be user, group, tree quota, or just general file system out of space. The node being accessed will not be initialized, and the user will not be able to read the affected file/directory.
300
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
mR11File_St atust8 0xdf591610: 0x7e12ea _ZN15UFS_DirOps_dir38getVnode EP9UFS_Vnod emPP9V.
Version 6.0.40.8
Celerra Management Applications
The following are fixed problems related to the Celerra Management Applications, i.e., Unisphere, Celerra Startup Assistant, or Celerra MMC Snap-ins:
Severity Symptom Description Fix Summary Tracking Number 395097
1
User was unable to add or manage newer versions of Celerra systems to legacy domains in Unisphere due to restrictive version check.
Version check was modified to allow Celerra systems with newer version to be added to or managed in the domains of current version systems using Unisphere. Added 10 gigabit network devices for inclusion in Celerra Manager network graphs.
2
Network statistics were unavailable through Celerra Manager when Celerra had 10 gigabit network devices. Graph always showed "Data Loading..." and graph options menu didn't include 10 gigabit network devices.
379658 / 36124794
EMC Celerra Network Server Version 6.0.70.4 Release Notes
301
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 374949 / 35327164
2
When a Celerra was in production mode, Unisphere took a considerable time to load. The loading symbol (rotating arrows) was persistently shown near the IP address of the Control Station in the Unisphere left pane.
An underlying command was used to find the model of the Celerra system which caused serious latency for the loading of the Celerra pages in Unisphere. The command model –x was modified to include caching of the model of the system which speeds up the Unisphere pages. Code has been fixed to ensure that Jserver parsing code never fails on unexpected output lines and ensures that “server_cifs” processes spawned by Jserver always complete and prevent excessive memory utilization from occurring.
2
Jserver fails to parse output of the server_cifs server_x -o audit command due to some extraneous lines coming from Data Mover response resulting in the process hanging on Control Station. Over time, lots of hanging server_cifs processes consumed all the memory on the Control Station and the Control Station had to be rebooted.
375192 / 35638732
CIFS
The following entries describe fixed Common Internet File System problems:
Severity Symptom Description Fix Summary Tracking Number 384136 / 36849782
1
The Windows group Policy was maintained on the domain controller in unicode files. If a user replaced the unicode file with a non-unicode security template, then the file was not read properly and if large enough (greater than 54Kbytes) the Data Mover panicked with the following DART panic/fault message: NIC in file: ../malloc.cxx at line: 525 :
Updated the GPO code to check if the file is unicode (first two bytes of the file are 0xFFFE), then process the file either as unicode or nonunicode, retrieving and maintaining all settings
302
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Memory corrupted..
that are applicable to the Data Mover. Code now uses the FQDN of the first interface of the default CIFS server for the CIFS server. Code now frees the configuration and the streams used to send the heartbeat when VC is stopped. 365843 / 34585998
2
When implementing CAVA using a CIFS server with disjoint DNS and Windows domains, CAVA would not go online unless the default CIFS server was specified in viruschecker.conf. When doing a lot of CAVA /CEPA configurations while the CAVA service was stopped, the configuration was not freed and the streams used to send the heartbeat were not freed. This led to a memory leak which could result in a Data Mover panic.
2
381971 / 36571200
EMC Celerra Network Server Version 6.0.70.4 Release Notes
303
Version 6.0.40.8
1
Under heavy load, when the CEPP feature was enabled, Data Mover panicked with the following due to a race condition between two CIFS threads: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: c601c2 Err code: 0 Target addr: 0 **. The stack trace showed: 0x2c3bc54: 0x14124c waitForReboot+0x90 0x2c3bc74: 0x1414d5 fault_dump+0x67 0x2c3bc94: 0x13a398 handle_fatal_fault+0xa 0x2c3bcd0: 0xc5b803 _ZN8fid_info15attachCloseCepp EP9CIFS_CEP P+0x59 0xfa6ffca0: 0xc60ada _ZN9CIFS_CEPP17setDeferredActio nEv+0x2a 0xfa6ffcc0: 0xc5262a _ZN8AppLibNT5closeEP8fid_info iR11File_St atusP1 0xfa6ffe80: 0xbf8302 _ZN13smb_streamCtx16closeOpen edFilesEt+0 xd2 0xfa6ffed0: 0xb09df2 _ZN13smb_streamCtx10credRemov eER11cRefer encerI
This race condition has been eliminated.
385161 / 36978078
1
After upgrading to 6.0 release, when using NTLM authentication, users from trusted domains might no longer be able to access resources (server, share, file, directory) granted through Universal group membership. DHSM connections (using Kerberos for authentication) try to re-use Kerberos tickets in the credential cache. The code used to check to see whether a ticket was available in the cache did not check to see if the ticket was expired, so the specified password was deleted. Subsequent code tried to use the expired ticket, then was unable to get a new ticket because the password had already been deleted, resulting in DHSM reconnection failure.
Code now correctly processes extra SIDs returned by the Domain Controller.
385626 / 36963164
1
Added code (in Kerberos) to check for expired ticket in credential cache, and to delete the ticket if it is expired.
375592 / 35601816
304
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
DART panic/fault message: >>PANIC in file: ../dskdump.cxx at line: 1836 : DHSM recall issues. The stack trace showed: 0x21201ae0: 0x14124c waitForReboot+0x90 0x21201b00: 0x1414d5 fault_dump+0x67 0x21201b20: 0x1413bd PANIC+0x29 0x21201b30: 0xa44e08 _Z11paniccfgcmdR6Option+0 x178 0x21201c70: 0x1cbdab _ZN11cfgcmd_desc6invoke EPKcPP8CCMD_Msgii +0x1a7 0x21201ce0: 0x2b205a _Z10xml_cfgcmdP7macDataRP 8CCMD_Msg+0x392 0x21201d30: 0x2ad00c _ZN17macCCMDCommand_DB6in vokeEPKcP7macData RP8C 0x21201d60: 0x2ad512 _Z17mac_CCMDDocParserP7 macDataRP8CCMD_Ms g+0x20
1
When users in Active Directory were in "lastname, firstname" format, the resulting distinguished name included an escaped comma; i.e., "\,". The Data Mover search filter processing for LDAP did not process escaped commas correctly, and returned incorrect search results.
Code was modified to allow an escaped comma to be included in the search filter string.
369461 / 3480317
EMC Celerra Network Server Version 6.0.70.4 Release Notes
305
Version 6.0.40.8
2
Data Mover stopped parsing nsswitch.conf file when encountering an invalid entry in it, the rest of the configuration lines were rejected and no error was reported about it which might prevent certain functionality from working, such as LDAP for CIFS UID/GID mapping. User tried to disable file filtering temporarily through the cifs.enableFileFiltering parameter. The parameter description says it applies immediately but it was found that changing the value had no effect if the connection was still opened. An attempt by client to create file "\nasadmin\New Microsoft Word Document.doc" failed with STATUS_ACCESS_DENIED. The Data Mover allowed the write after the Data Mover was rebooted.
Code has been fixed to parse the whole file and ignore invalid entries, but report an error message in the server_log about the invalid entry. The parameter was modified to not require any user action (rebooting), even for connection active when the parameter has been changed.
372383 / 35287884
1
373375 / 35092392
Control Station
The following entries describe fixed problems related to the Celerra Control Station:
Severity Symptom Description Fix Summary Tracking Number 370795 / 35139496
2
The Data Mover did not failover after the panic. This occurred because the code did not notify all the processes to release the process locks, and a checkpoint related process was holding a lock which prevented the failover. Users observed the following error in the GUI alert log: Parsing error. Invalid XML. The message did not contain any specific information about the parsing error.
Code now ensures that all processes are appropriately notified to exit and release the locks.
2
The function that logs the messages containing the reason for the parsing failure was altered to log the entire message with the specific reason of the parsing error in /nas/log/nas_log.al. mgmtd file. This
361847 / 33292418
306
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
message will contain the entire xml string, including in this case, the invalid characters. 2 When NTP services were started on a Data Mover with the interval greater than 1440 minutes specified, it was set on the Data Mover but then the Data Mover panicked with the following DART panic/fault message: *** Divide exception. Virt
ADDRESS: a2edaf ***. The stack trace showed:
Code now prevents a value greater than 1440 minutes (or 24 hours) for NTP sync interval.
361893 / 34311052
0x2ba3d0c: 0x140ce4 waitForReboot+0x90 0x2ba3d2c: 0x140f69 fault_dump+0x67 0x2ba3d4c: 0x139ee4 handle_fatal_fault+0xa 0x2ba3ed0: 0xa2fa0c _ZN8TimeSync5startEv+0xb2 0xfdd37fe0: 0x1455cc Sthread_startThread_internal+0xe
CORE
The following entries describe fixed problems related to Celerra Data Movers:
Severity
Symptom Description
Fix Summary
Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
307
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 374681 / 35598996
1
A rolling panic occurred when a local netgroup file contained a nested netgroup. Users had CIFS access, but no NFS access. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 14dd7a Err code: 2 Target addr: 8f **. The stack trace showed: 0x2c2cce4: 0x141254 waitForReboot+0x90 0x2c2cd04: 0x1414dd fault_dump+0x67 0x2c2cd24: 0x13a3a0 handle_fatal_fault+0xa 0x2c2cd50: 0x1c54a3 _Z13CollectStreamPvS_+0x77 0x5081e30: 0x1d577a ReadPutq+0x2b2 0x5081e60: 0x6350eb nfs_rput+0xa47 0x5081eb0: 0x634b8c nfs_rput+0x4e8
Code now processes only ten levels of nesting in a netgroups. If there exist netgroups nested to a depth greater than ten, the following error message is now printed in the server log: NETGROUP netgElem::IsMember Looping on <netgroup name> (loop in nested netgroups) \n.
1
When a Data Mover was mistakenly zoned with more than 32 other initiators, the Data Mover was unable to see the storage and failed to boot.
The Data Mover will now detect when too many devices are incorrectly zoned into the Data Mover and will log diagnostic information that will help quickly identify the issue. The following messages will appear in the server log under such conditions: Failed to add device, no resources and Failed to add device, unsupported login type
366770 / 34764834
308
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Deduplication
The following entries describe fixed issues pertaining to Celerra Deduplication:
Severity Symptom Description Fix Summary Tracking Number 375789 / 35737678
1
On a system with heavy deduplication activity, the Data Mover did not release memory in the EDRS (EMC Data Reduction Services) portion of the code and caused a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 289 : Out of memory. The stack trace showed: 0xde21b0a0: 0x14124c waitForReboot+0x90 0xde21b0c0: 0x1414d5 fault_dump+0x67 0xde21b0e0: 0x1413bd PANIC+0x29 0xde21b0f0: 0x13bef6 _ZN19Memory_AddressSpace10all ocPagesEi12 Memory 0xde21b110: 0x1fe726 _Z11more_memoryjP9free_list+0x4 a 0xde21b130: 0x1fe7ec malloc+0xa0 0xde21b160: 0x152dbc2 csx_rt_os_mem_raw_alloc_maybe+0 x14 0xde21b180: 0x152bc16 csx_rt_mem_raw_alloc_maybe+0x14
Code was fixed to correctly release memory and prevent this panic.
1
Under rare condition, removing a deduplicated file, creating a checkpoint and performing a restore on the same file system at the same time, resulted in a deadlock due to lock interdependencies resulting in blocked threads on the Data Mover and eventually causing access issues.
Code has been fixed to properly handle interdependencies lock so no deadlock would occur under this condition.
372616 / 35375114
EMC Celerra Network Server Version 6.0.70.4 Release Notes
309
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 374310 / 35478060
1
File systems with deduplication enabled (including NDMP backup checkpoint file systems) failed to unmount. Messages similar to the following were logged to the server log: 2010-07-28 19:24:11: DEDUPE: 3: openFileData failed: fsid 483 status RDE_STATUS_DECOMPRESSION_ER ROR. When a read of a dedupe file system resulted in sending a larger request than the fibre driver could process, the Data Mover panicked with following DART panic/fault message: >>PANIC in file: ../BVolumeIrp.cxx at line: 169 : Problem building SG list. Panicking Processor: 2. The stack trace showed: ee2a90: 0xa9e735 _ZN8SgIO_IRP10initializeEP10Dis kIO_IRPP11B asi ee2ad0: 0xa54b3c _ZN11BasicVolume11aioInternalEP 10DiskIO_IR PP8 ee2ba0: 0xa55cd7 _ZN11HyperVolume11aioInternalEP 10DiskIO_IR PP8 ee2c70: 0xa5f213 _ZN11SliceVolume11aioInternalEP1 0DiskIO_IR PP8 ee2ca0: 0xa55cd7 _ZN11HyperVolume11aioInternalEP 10DiskIO_IR PP8 ee2d70: 0xa55cd7 _ZN11HyperVolume11aioInternalEP 10DiskIO_IR PP8 ee2e40: 0x133ccb5 _ZN10SnappedVol11aioInternalEP1 0DiskIO_IRP P8 ee2e70: 0xa55cd7 _ZN11HyperVolume11aioInternalEP 10DiskIO_IR PP8 This occurred was because of a limitation in the fibre driver code to support up to 64
Code is updated to ensure files are closed when a DECOMPRESSION ERROR is detected.
1
Code now breaks larger requests into chunks of 64 entries to prevent this sort of panic.
379947 / 36247340
310
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
request blocks only.
File System
The following entries describe fixed bugs pertaining to File Systems:
Severity Symptom Description Fix Summary Tracking Number 165384 / 33310518
1
If a frozen file system was unmounted while the system was out of vnodes, the naming node became inconsistent which led to a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 6d8dd7 Err code: 0 Target addr: 1d4 **. The stack trace showed: 0x2acb734: 0x13fb94 waitForReboot+0x90 0x2acb754: 0x13fda9 fault_dump+0x67 0x2acb774: 0x138d94 handle_fatal_fault+0xa 0x2acb7b0: 0x7dc2c4 _Z17MPDShadowCallBackP15File_ NamingNodeP K18File 0x24a45b00: 0x7cf46a _Z14ShadowCallBackP15File_Nam ingNodePK18 File_N 0x24a45c00: 0x7777a6 _ZN9UFS_Vnode19doDirChangeCal lBackEPK18F ile_Na 0x24a45c30: 0x694d7f _ZN12File_DirNode21lookupComp onentLocked EPK18F 0x24a45d00: 0x69b627 _ZN16File_RootDirNode21lookup ComponentLo ckedEP
The naming node is now appropriately replaced under this condition to avoid the inconsistency.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
311
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 370855 / 35172362
1
When user tried to mount a root file system in read-only mode on a stand-by Data Mover, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 794170 Err code: 0 Target addr: 40 . The stack trace showed: 0x2898554: 0x13fa6c waitForReboot+0x90 0x2898574: 0x13fc69 fault_dump+0x67 0x2898594: 0x138cfc handle_fatal_fault+0xa 0x28985e0: 0x75723c _ZN9UFS_Vnode10lookupCompEPK 18File_Name Componen 0xc48a77d0: 0xec4db4 _ZN10MGFS_Vnode10lookupCompE PK18File_Na meCompo 0xc48a7810: 0x6f7de3 _ZN18UFS_FileSystemKind6init FSEPKcP9IO_ Objecti 0xc48a7900: 0xec1f89 _ZN19DHSM_FileSystemKind6ini tFSEPKciR6O ptionP1 0xc48a7980: 0x6cc267 _Z25File_MountCommandInterna lR6OptionP1 8VFS_Fi
During a mount, code now checks whether the default log device is initialized or not. If not initialized, the following error is returned: Log device not initialized.
1
Issuing a quota check command at the same time as a quota operation from a Windows client resulted in a deadlock condition and rendered the file system inaccessible until the Data Mover was rebooted. 1. During any version 5.6 to 6.0.36.4 upgrade, if the quota policy was set to filesize (which is not the default blocks configuration), the task for upgrading Data Mover could timeout because the Data Mover takes too long to reboot due to an automatic tree quota check. 2. While running version 6.0.36.4,
The fix has the quota command from Windows abort requiring the quota lock while the FS is inactive. Code no longer forces a quota check during the reboot procedure, following a quota policy change. See Primus emc254145 for more details.
381651 / 36551080
1
388085 / 37305372
312
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
changing the quota policy and rebooting the Data Mover could cause a long reboot, due to the automatic tree quota check.
FileMover (DHSM)
The following entries describe FileMover-related fixed bugs:
Severity Symptom Description Fix Summary Tracking Number 378918 / 36077544
1
User saw incorrect data when reading stub files on loopback connections. This issue occurred because of the way that offsets and read sizes were calculated on read operations which led to read timeouts. A Data Mover exhibited high CPU usage when DHSM was enabled and in-use on one or more of the file systems. Errors similar to the following were displayed in the server log. 1. "error direct read block xxx of entry xxx, status 20 - EndOfFile", and/or 2. "File size mismatch: Stub file xxxx bytes, migration vnode xxxx bytes,ino=xxxx status=EndOfFile"
Code now correctly adjusts the read offset and size on error recovery read operations.
1
Fixed a logic error in the DHSM I/O (read/recall) path to prevent this infinite loop.
372101 / 34673002
Install/Config/ Upgrades
The following entries describe fixed issues related to installations, configurations and upgrades:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
313
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 372086
1
During an NS4-120 installation, the "setup_enclosure detectSingleDMSystem" failed with the following error: Discover 1 Mgmtswitches in Primary subnet A, 0 Mgmtswitches in Secondary subnet B Please check cables and Mgmtswitches on Secondary subnet B Error: detectSingleDMSystem FAIL (EMISMATCHCOUNT) Remove Dynamic IP range in DHCPD_CFG ... OK Restart DHCPD with updated DHCPD_CFG ... OK Error: UNDEFINED retval = -10 (EMISMATCHCOUNT). While performing one of the following Upgrade/Conversion procedures, setup_slot failed while adding a new I/O module in single Data Mover system: -Upgrade from one blade to two blades -Add a second SFP for second Tape support -Add an I/O module (to all blades simultaneously) -Changing to the faster I/O Module of the same type -Add a second Control Station. During Express install, the "set_up_Data_Mover" process took 40 minutes to complete on a 4 Blade setup. This was due to a built-in delay during a reboot.
Code was modified so that the install will not fail if it does not discover the switch on subnet B in a single Data Mover system. It will be discovered by the setup_enclosure initSystem command which is run later during the install process.
2
The wrong reason code was being monitored for these procedures, and because the system never reached that reason code (5), the command returned an error. This has been fixed.
374063
1
The fix eliminates the delay when issuing a reboot with NAS services down.
388839
iSCSI
The following entries describe fixed iSCSI-related issues:
Severity Symptom Description Fix Summary Tracking Number 377716 / 33424708
2
When client demoted an iSCSI snap LUN using the force flag, the iSCSI remove LUN operation logged an error instead of
The server log message for this case is not a warning. An
314
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
a warning message in the server log. Users are likely to see this when using EMC Replication Manager because ERM internally demotes snap with force flag on its replication destination Celerra.
error message is only logged when the remove LUN operation actually fails.
NDMP
The following entries describe fixed issues related to Network Data Management Protocol and/or Backup:
Severity Symptom Description Fix Summary Tracking Number 371282 / 35158192
1
An uncooperative peer closed the TCP receive window during an NDMP backup. This led to the following panic because the NDMP was waiting on a TCP stream completion that never came. DART panic/fault message: >>PANIC: I/O not progressing LastVol touched Sn128 Kind 5 (ptr=0xa26d6104). The stack trace showed: 0x79dbe90: 0x13fb94 waitForReboot+0x90 0x79dbeb0: 0x13fda9 fault_dump+0x67 0x79dbed0: 0x13fc91 PANIC+0x29 0x79dbee0: 0x8478e0 _ZN9RawIO_IRP13checkProgressEv+0 x11a 0x79dbfd0: 0x8477bc _ZN17IrpProgressThread5startEv+ 0x6 0x79dbfe0: 0x144344 Sthread_startThread_internal+0x e
This was fixed by setting the abort flag for 3-way VBB operation, should an unmount or internal error require abortion of the backup/ restore operation.
Network
The following entries describe fixed network-related issues:
EMC Celerra Network Server Version 6.0.70.4 Release Notes 315
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 371067 / 35186226
1
LACP could not be used with Dell Poweredge 6248 network switch and Celerra. The following error message appeared in the server_log: DRIVERS: 3: LAP_Port::receive_LACP len=114 expected_len=110. The verification of the PDU length was too strict, where it should have been verifying the PDU is a least the minimum size of 110. When the client sent a packet greater than the MTU size (between 1518 and 1530 bytes), the Broadcom device hung. The Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../dskdump.cxx at line: 1836 : network issue. The stack trace showed: 0x1f635ae0: 0x14124c waitForReboot+0x90 0x1f635b00: 0x1414d5 fault_dump+0x67 0x1f635b20: 0x1413bd PANIC+0x29 0x1f635b30: 0xa45118 _Z11paniccfgcmdR6Option+0x178 0x1f635c70: 0x1cbdab _ZN11cfgcmd_desc6invokeEPKcPP 8CCMD_Msgii +0x1a7 0x1f635ce0: 0x2b205a _Z10xml_cfgcmdP7macDataRP8CCMD_ Msg+0x392 0x1f635d30: 0x2ad00c _ZN17macCCMDCommand_DB6invoke EPKcP7macDa taRP8C 0x1f635d60: 0x2ad512 _Z17mac_CCMDDocParserP7macDat aRP8CCMD_Ms g+0x20
The PDU length verification has been changed to accept anything not less than the expected PDU length.
1
Fixed the Broadcom 10G hang issue by correcting the MTU (max transmission unit) size passed to the device.
376090 / 35799674
1
An exploit of the NFS mount protocol was identified where an attacker could spoof IP addresses that are normally used between the Celerra Control Station and Data Mover over a private IP network and gain unauthorized access to file systems on the Celerra.
Code has been fixed to reject all mount requests using the Control Station IP addresses that do not come over from a private network that is internal to the Celerra
118547
316
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
cabinet, eliminating this vulnerability.
NFS
The following entries describe fixed issues related to Network File Systems:
Severity Symptom Description Fix Summary Tracking Number 372545 / 35327318
1
When using NFSv4, a missing reference for NFSv4 state object on Data Mover caused a Data Mover to panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 66fdf2 Err code: 2 Target addr: 24 **. The stack trace showed: 0x2c2aba4: 0x141254 waitForReboot+0x90 0x2c2abc4: 0x1414dd fault_dump+0x67 0x2c2abe4: 0x13a3a0 handle_fatal_fault+0xa 0x2c2ac60: 0x66e352 _ZN16NFSv4_StateTable5CloseEm P8stateid4P 8nfs4_c 0xe10b5d50: 0x65d129 _Z10nfs4_closeRP4msgbS1_RPhR8 nfs4_ctx+0x 18b 0xe10b5d90: 0x6592b0 _Z23nfs4_convertNfsv3Statusm+0x 626 0xe10b5ee0: 0x63b64d _ZN17nfs_serverInitial5startEv+ 0xad7 0xe10b5fe0: 0x145bd6 Sthread_startThread_internal+0x e
Code has been fixed to properly handle missing reference on the object and prevent this panic.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
317
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 377431 / 35858230
2
NFSv4 locks could not be reclaimed after a temporary network outage. If an NFSv4 client temporarily lost contact with the Celerra and attempted to reclaim locks after its lease expired (20 seconds by default), the operation would fail. Access to the file from the process that owned the lock could be impacted, though if the process that owned the lock specifically released the lock, access would be restored. This problem was introduced in 5.6.45. NFSv4 code did not check permissions when modifying the ACL based on the mode bits. This can result in access behavior inconsistent with permissions. Unable to write data to a File Level Retention Enterprise (FLR-E) files system if the file system was mounted using NFS V4.
Fixed a miscomputation of the NFSv4 owner sequence number in Open-Claim-Previous operation. Attempts to reclaim locks will now succeed.
1
Code now correctly applies permissions when changing the NFSv4 ACL. Code has been fixed to ensure users can write data to FLR-E file systems mounted using NFS V4. Code has been fixed to properly release allocated memory for dropped requests, preventing Data Mover panics in unhealthy network environments.
378959 / 36091074
2
368246
1
In an unhealthy network, where many network packets were lost, memory on the Data Mover was exhausted by those many dropped requests. This led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../strbufs.cxx at line: 1747 : out of msgb. The stack trace showed:
379200
0xe2d6960: 0x20adcc _ZN15Buffer_MsgbPool3getEv+0 xac
0xe2d6990: 0x20af9a allocb+0x54 0xe2d69c0: 0x6e5ec8 nfs_rput+0x2c0 0xe2d6ab0: 0x40aa3e udp_input+0x9fa 0xe2d6c00: 0x443879 _Z9ip6_inputP10IPDevice_tP4msgb i+0x10cb
EMC Celerra Network Server Version 6.0.70.4 Release Notes
318
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
0xe2d6d60: 0x4316f0 _Z10ip6_mxrputP5queueP4msgb+0x1 3a 0xe2d6dd0: 0x37b4bd _Z7bcmreadPvP4msgbtjj+0x4e7 0xe2d6ee0: 0x38b2b2 _Z20LM_ServiceInterruptsP16_L M_DEVICE_BL OCK+0x3
Replication
The following entries describe fixed problems encountered while using Celerra Replicator, SnapSure or Checkpoints:
Severity Symptom Description Fix Summary Tracking Number 360751 / 34233340
1
The Data Mover on the destination side of a ReplicationV2 session panicked after a failover issued using the session name. This happened because the replication database contained an old entry for an inactive session with the same name and the failover operation used the old entry rather than the new one. When attempting to create an iSCSI replication session, if version set was out of space, the task showed success when it actually failed.
The session ID is now used during the failover since it will always be unique.
2
The iSCSI replication create task status was changed to Severity=error. The following will be displayed in the task status: Version set out of space.
374003
EMC Celerra Network Server Version 6.0.70.4 Release Notes
319
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 366022 / 34665376
1
In a cascade replication scenario, a replication session for the middle hop to the third hop was created before creating a session for the first hop to the middle hop. The code allowed a file system that acted as replication source and destination (middle hop) to be mounted as read-write. This violated replication destination read-only property assumption and caused a corruption on that file system which lead to Data Mover panic with the following DART panic/fault message: >>PANIC in file:
../ufsdirhash.hxx at line: 623 : UFS_DirHashIter_dir3::getEntry: mangled entry. The stack trace showed: 0x2b1f7050: 0x140ce4 waitForReboot+0x90 0x2b1f7070: 0x140f69 fault_dump+0x67 0x2b1f7090: 0x140e51 PANIC+0x29 0x2b1f70a0: 0x7b045a _ZN20UFS_DirHashIter_dir38getEn tryEm14Bu fInten 0x2b1f70e0: 0x7c2e9d _ZN20UFS_DirHashIter_dir34findE mPKcm13Fi le_Nam 0x2b1f7110: 0x7d1dd4 _ZN20UFS_DirHashIter_dir34findE mPK18File _NameC 0x2b1f7150: 0x7c3ad7 _ZN15UFS_DirOps_dir37scandirEP9 UFS_Vnode PK18Fi 0x2b1f71d0: 0x7cd806 _ZN15UFS_DirOps_dir311getSynony msEP9UFS_ VnodeP
A new check was added to the code, which prevents the read-write mount of a file system that acts as replication source and destination in a cascade replication.
2
While trying to perform an out-of-family Rep V2 replication, whenever a server with a lower slot number was made as standby for a server with a higher slot number, the multicast session lost the lower slot number server from the target list. When user invoked the nas_fs -delete command, the following error was generated: Error 5005: DpRequest_Max_VS_SuspendedCheckp ointsReach ed.
When there is a change in the server state, CIC (Celerra Interconnect) multicast target list is now generated from scratch rather than being modified. This solves the problem of a target being missed due to state and relation change of the servers.
144857
320
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
2
While using Replication V2, if a Data Mover was rebooted during a nas_copy process, the data transfer completed successfully, but the command did not complete and hung after the reboot. During merge operations involving file system checkpoints, the TOC (Table Of Contents) became inconsistent with the trailer and map. This inconsistency led to the following panic which required an unmount of the bad checkpoint before the Data Mover could be stabilized. DART panic/fault message: D130832_0: Btree_Insert failed, mark ckpt corrupted. The stack trace contained various messages similar to the following: 0x20643f40: 0x141254 waitForReboot+0x90 0x20643f60: 0x1414dd fault_dump+0x67 0x20643f80: 0x1413c5 PANIC+0x29 0x20643f90: 0xfb36ad _ZN28IndexChainCorruptMarkReq uest13handl eReque 0x20643fc0: 0xfb3312 _ZN23CkptCorruptMarkerThread5st artEv+0x80 0x20643fe0: 0x145bd6 Sthread_startThread_internal+0x e
The nas_copy command will now complete after a Data Mover reboot.
376684
1
Code now verifies the TOC entries against the other 2 references before the merge, during the merge, and after merge completes. Code then corrects the TOC as needed.
366985 / 34775056
1
Network issues might cause RepV2 session hang.
Code now promotes the RCP_CLOSE message priority so it will not cause blocking.
156084
EMC Celerra Network Server Version 6.0.70.4 Release Notes
321
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 383673 / 36791874
1
If the internal checkpoint name exceeded 16 characters, memory corruption occurred which caused a system panic such as I/O not progressing and volume display error. DART panic/fault message:
>>PANIC in file: ../sched.cxx at line: 846 : SYSTEM WATCHDOG. The
stack trace showed:
0x5089f24: 0x14124c waitForReboot+0x90 0x5089f44: 0x1414d5 fault_dump+0x67 0x5089f64: 0x1413bd PANIC+0x29 0x5089f74: 0x1432c9 system_watchdog+0x77 0x5089f84: 0x139f6a _t_nmi+0x68 0x5089fe0: 0x145bfa Sthread_startThread_internal+0xe
Checkpoint internal names whose length exceeded 16 characters were truncated resulting in unexpected behavior for example the name D2354619762_10113 could be truncated to D2354619762_1011 which does not exist. The naming convention was changed to shorten the name. The new format of the name is "D<ckptFsID>_<ckptG enCnt>", which keeps the total length below 16 characters and maintains the uniqueness of the name. This checkpoint name is only used internally for debug/display purposes. It can be seen in the server_log during checkpoint refresh operations for example 2010-09-20 06:42:01: 26042826752: SVFS: 6: D2354619762_1011: Recovered(Stable Index) FsID:23546 state:Active SavVol:29390 FsVol:D2480619762_ 0 rsv=2 h=12322040 c=12322057 t=12322057, in 0 Msec 2010-09-20 06:42:01: 26042826752: SVFS: 6: D2354619762_1011:
322
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
currDel:D235461976 2_101 13 st:Active prevDelta:1573888
1
Replication V2 sessions were not automatically synchronized. The timer being used to check if it is time to transfer was using an overflowed value.
The interval between checks of whether it is time to sync a session now uses a consistent value rather than a calculation which could overflow.
386995 / 37192564
Security
The following entries describe fixed security- related issues:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
323
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 376153 / 35803348
1
The LDAP configuration on a Data Mover was using multiple LDAP servers, and LDAP was configured to use the Active Directory "member" attribute instead of "memberUid". When the network connectivity to all LDAP servers was bad, memory corruption occurred and the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 351 : Memory corrupted . The stack trace showed: 0xdb9477e0: 0x14124c waitForReboot+0x90 0xdb947800: 0x1414d5 fault_dump+0x67 0xdb947820: 0x1413bd PANIC+0x29 0xdb947830: 0x1fe8e7 free+0x69 0xdb947860: 0x1fefb8 _ZdlPv+0x10 0xdb947870: 0x4f6582 _ZN10LdapClientD1Ev+0x32 0xdb947890: 0x53f311 _ZN16LdapDomainClient13getgrb ymemberEPKc PPP12A 0xdb947bd0: 0x69b7f3 _ZN20Access_GroupDataBase15ad dNewExtraGi dsEP18
Code was modified to correctly handle failover between LDAP servers and prevent memory corruption.
1
NTP software on the Control Station was vulnerable to the denial-of-service attack described in US CERT VU568372 when NTP service is running (disabled by default).
Code has been fixed to pickup the nonvulnerable version of NTP software from RedHat which prevents this denial-of-service attack.
364874 / 34476786
Storage
The following entries describe fixed storage-related issues.
324
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 369324 / 35030006
1
A Data Mover watchdog panicked due to FC environmental conditions. This can happen when the processing used to clear some FC conditions takes too long due to unnecessary reads across the PCI bus. Direct attached tape drives connected with cable lengths greater than 50m may not be discovered after upgrade to 5.6.47.11 and above. The issue was caused by a race condition that was triggered by increased latencies using longer cables.
The unnecessary reads on the PCI bus have been removed during FC processing.
1
The code has been fixed to avoid race condition when latency is increased due to long cables.
381342 / 36335402
UFS
The following entries describe fixed UNIX File System issues:
Severity Symptom Description Fix Summary Tracking Number 361139 / 34143216
1
In an environment with VMware ESX server that was using Celerra iSCSI LUNs for its VMFS file systems, the ESX server had aborted an iSCSI session and disconnected from iSCSI LUNs while a large iSCSI snap was being deleted on the Celerra. This caused an outage on the ESX virtual machines. The following messages in server_log indicate the issue: SCT: 6: Task Management Function received: Abort Task, I_T Nexus: [<iqn>] ISCSI: 3: IscsiSession::closeSession: I_T nexus lost, sess=0xe0b6dc04 from [<iqn>]
The Data Mover code has been updated to improve response time during the iSCSI snap delete process. From the ESX server side, VMware updated ESX server code to allow longer timeout before aborting an iSCSI session.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
325
Troubleshooting and getting help
Troubleshooting and getting help
This section provides specific instructions for contacting EMC Customer Support and obtaining additional information about EMC products.
Where to get help
EMC support, product, and licensing information can be obtained as follows. Product information: For documentation, release notes, software updates, or for information about EMC products, licensing, and service, go to EMC Online Support (registration required) at: http://Support.EMC.com Troubleshooting: Go to EMC Online Support. After logging in, locate the appropriate Support by Product page. Technical support: For technical support and service requests, go to EMC Customer Service on EMC Online Support. After logging in, locate the appropriate Support by Product page and choose either Live Chat or Create a service request. To open a service request through EMC Online Support, you must have a valid support agreement. Contact your EMC Sales Representative for details about obtaining a valid support agreement or to answer any questions about your account.
Note: Do not request a specific support representative unless one has been assigned to your particular system problem.
326
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Copyright © 2013 EMC Corporation. All Rights Reserved. Published in the USA. Published February 2013 EMC believes the information in this publication is accurate as of its publication date. The information is subject to change without notice. THE INFORMATION IN THIS PUBLICATION IS PROVIDED "AS IS." EMC CORPORATION MAKES NO REPRESENTATIONS OR WARRANTIES OF ANY KIND WITH RESPECT TO THE INFORMATION IN THIS PUBLICATION, AND SPECIFICALLY DISCLAIMS IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Use, copying, and distribution of any EMC software described in this publication requires an applicable software license. EMC2, EMC, and the EMC logo are registered trademarks or trademarks of EMC Corporation in the United States and other countries. All other trademarks used herein are the property of their respective owners. For the most up-to-date regulatory document for your product line, go to the technical documentation and advisories section on the EMC online support website.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
327
Version 6.0.70.4
Release Notes
February 28, 2013
P/N 300-009-958
REV 18
These release notes contain information on these topics:
Product description ................................................................................... 2 New features and changes ........................................................................ 2 Known problems and limitations .......................................................... 27 Problem prevention ................................................................................. 50 Technical notes ......................................................................................... 56 Environment and system requirements................................................ 78 Documentation ......................................................................................... 79 Installation and upgrades ....................................................................... 81 Fixed problems ......................................................................................... 88 Version 6.0.70.4......................................................................................... 89 Version 6.0.65.2......................................................................................... 98 Version 6.0.60.2....................................................................................... 115 Version 6.0.55.3....................................................................................... 129 Version 6.0.51.6....................................................................................... 141 Version 6.0.43.1....................................................................................... 189 Version 6.0.43.0....................................................................................... 189 Version 6.0.42.3....................................................................................... 190 Version 6.0.41.4....................................................................................... 222 Version 6.0.41.3....................................................................................... 225 Version 6.0.40.8....................................................................................... 301 Troubleshooting and getting help ....................................................... 326
1
Product description
Product description
The EMC Celerra Network Server is a network-attached storage (NAS) server. It enables clients on a network to store and retrieve files over the network. These release notes contain information about version 6.0 of Celerra Network Server. Check EMC Online Support at http://Support.EMC.com for updates to this document. Updates are made when a new version or patch is released, or when new information is discovered.
New features and changes
Version 6.0.70.4
NS gateway support for new Symmetrix Enginuity 5876.159.102. • • NS gateway supports Enginuity 5876.159.102 running on any of the VMAX platforms (10K, 20K, and 40K). NS gateway software features such as TimeFinder® and SRDF®, which are supported on the earlier Symmetrix Enginuity, will be supported on 5876.159.102 Enginuity. Software features such as FAST™ and Federated Tiered Storage (FTS) are not supported with 6.0.70.4. Thin BCV (Business Continuance Volume) disks are supported on NS gateway. NS gateway inherits all hardware and software restrictions imposed by VMAX 10K, 20K and 40K, and Enginuity 5876.159.102. VP Compression is supported on NS gateway. It is recommended that you create a user defined pool for the compressed LUNs so that consistent performance can be provided for the file systems. VP compression must be done from the Solution Enabler host. I/O Quota is supported on NS gateway. It is recommended to create a userdefined pool for the LUNs that have the same I/O Quota configuration so that the user-defined pool can provide consistent performance for file systems. I/O Quota must be configured from the Solutions Enabler host. The recommended configuration (best practice) for NS gateway using I/O Quota with masking view (LUN masking feature) is as follows: Ensure that a parent Storage Group is in a masking view with the Data Mover initiator group and Symmetrix FA port group.
® ™
• • • •
•
•
2
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
Create multiple Child Storage Groups with different performance expectations. For example, extra performance, performance, and capacity. Set an appropriate I/O Quota in each Storage Group. Execute diskmark operation. If the storage groups are all from fibre channel, create user defined storage pools for each storage group. Create file systems from the user defined storage pools.
Version 6.0.65.2
For NS-G2, NS-G8, VG2, VG8 gateways, if you plan to upgrade to VNX File OE 7.0 or later in future, do not install this version.
Versions 6.0.61.0, 6.0.60.2, 6.0.51.6, 6.0.43.1, 6.0.43.0, 6.0.42.3, and 6.0.41.4
There were no new features or changes to functionality in these versions. The Fixed problems section contains changes made in this version.
Version 6.0.55.3
NS-G2/NS-G8 6.x support for VNX Block arrays This release supports installation of NS-G2/NS-G8 Gateways and their connection with VNX Block arrays using the Fibre Channel protocol. New features in the upcoming VNX Operating Environment for Block 5.32 are not supported.
Version 6.0.41.3
Checkpoint scheduling relative name convention As of this release, SnapSure supports checkpoint CVFS names that are dynamically assigned based on the relative order of checkpoints (for example, nightly.001, nightly.002, nightly.003, and so on.) To use this option, you must specify the relative CVFS name prefix, the delimiter, and the starting index. The delimiter and starting index values are recorded in the nas_param database and must be manually edited. The complete CVFS name while using the relative prefix option is a concatenation of the prefix, delimiter and starting index. The prefix of a relative checkpoint must be a PFS-wide unique string and can contain up to 20 ASCII characters. It must not include intervening spaces, colons (:), or slashes (/). The delimiter must be either a period (.)
EMC Celerra Network Server Version 6.0.70.4 Release Notes 3
New features and changes
or an underscore (_). The starting index must be either 0 or 1. This feature is managed through the GUI or the CLI, or both. CPU Utilization Threshold Alerting As of this release, you can enable CPU utilization threshold alerting and receive automated SNMP trap or email notifications when a Data Mover CPU utilization crosses a certain threshold for a set period of minutes and persistence. You can also be notified when the condition is no longer applicable and the CPU utilization subsides to a normal level. The Configuring Celerra Events and Notifications technical module provides more detailed information about this feature. File system replication with common base checkpoints This feature is also known as "Incremental Attach Replication". When you replace source or destination file systems, the replications and internal checkpoints are deleted. You will have to perform a full data copy to be able to create future replications with the original configurations. To avoid performing a time-consuming full data copy over the WAN, you can create a local replica of the file system that you want to replace. You can then create one or more user-specified checkpoints on the file system to be replaced, the local replica, and each of the other file systems involved in the replication. After creating user-specified checkpoints, you must refresh each replication by specifying the user checkpoint on the source and destination file system for that replication. Consequently, the user checkpoint of each destination file system (including local replica) has the same view of the file system as the user checkpoint on the source file system that will be replaced. User checkpoints on the source and destination file systems, which have the same content, are called common base checkpoints. You can now use the local replica for replications without performing a full data copy over the WAN. Implementing this technique is useful in disaster recovery scenarios as well. To avoid a full data copy in the event of a disaster, create one or more user-specified checkpoints on the source and destination file systems and refresh their replication by using these user checkpoints. This process transfers data from the source user checkpoint to the destination file system through the existing replication session. Thus, the destination user checkpoint has the same view of the file system as the
4 EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
source user checkpoint. This process is repeated for each destination in the one-to-many configuration. For a cascading replication, the replication must be refreshed for this first hop and then for the second hop. After this process, even if you delete the replication sessions, you can configure new replication sessions between any two destinations because replication automatically identifies the common base checkpoints and avoid full data copy. You can create user checkpoints for file system and VDM replications. For VDM replications, you can create the user checkpoints on the root file system of the VDM. You cannot delete user checkpoints when they are being used by replication either for transfer or as an active common base. Greater than 4 TB Quotas As of this release, the storage limit per user, group, or tree quota has been increased from 4 TB to 256 TB. You can control the usage of the disk resource and limit the expensive disk space usage by various user or group or trees on a file system. Support for 256 TB quotas does not imply any change in the maximum size of the underlying file system. Caution: If you replicate a quota-enabled file system from a system running Celerra Network Server (CNS) version 6.0.41.0 or later to a destination system running a CNS version earlier than 6.0.41.0, the destination file system can become unusable. Upgrade quota database limits for file systems You can upgrade your quota databases to the new 256 TB limit from the existing 4 TB limit. The nas_quotas command has a new option called quotadb that, when run, performs the upgrade. The file system will be unavailable during the upgrade.
Note: Before the upgrade process runs, the Control Station displays the estimated upgrade time on the file system whose quota database will be upgraded, and also displays a warning message to notify you that the file system will be unavailable during the upgrade process. If you are in interactive mode, a dialog displays letting you choose whether you want to continue. Since the file system will be unavailable during the upgrade, the upgrade can be deferred to a later time. If you are in non-interactive mode, after displaying the estimated upgrade time message and warning message, the upgrade process starts immediately.
The Command Reference Manual contains more detailed information about the -quotadb command.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
5
New features and changes
Online quota update tool Once a tree, user, or group quota is enabled on a directory, in some instances, differences may occur between a reported quota usage and the actual usage. The online quota update tool lets you run a quota check on a tree or file system. The tool updates the tree, group, and user usage statistics, while keeping the directory tree and file system online, and keeping data access available. You can also run a quota check in the offline mode, if desired. If you have file system administrative privileges, you can use the Celerra CLI to run the online quota update tool. If an error is encountered, an error message is generated. The status of the online quota check shows the percentage completed.
Note: 512 MB of free space is needed on the file system to run a tree quota check.
Performance problem collection tool This tool enables users to send all relevant information related to the file system to EMC for diagnosis. The performance problem collection tool is a single CLI tool that gathers all the relevant logs (systat, IO trace, etc.) for a specified file system so that it can be easily transmitted to EMC as part of a support escalation. Prior to this release, users had to manually provide this information. Symmetrix Enginuity version 5875 This release supports the new Symmetrix Enginuity version 5875, which runs on the VMAX platform, and includes a new version of Solutions Enabler v7.2. VNX supports tolerance for Enginuity 5875 release only. It does not leverage any new features included in 5875. User-defined pool usage enhanced The -is_greedy attribute of the nas_pool command has been enhanced to include identifying which member volumes of a user-defined storage pool are used to provide space when creating or extending a file system. If the -is_greedy attribute is set to n (default), the system uses space from the user-defined storage pool's existing member volumes in the order that the volumes were added to the pool to create a new file system or extend an existing file system. If the -is_greedy attribute is set to y, the system uses space from the least-used member volume in the user-defined storage pool to create a new file system. When there is more than one least-used member
6
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
volume available, AVM selects the member volume that contains the most disk volumes. For example, if one member volume contains four disk volumes and another member volume contains eight disk volumes, AVM selects the one with eight disk volumes. If there are two or more member volumes that have the same number of disk volumes, AVM selects the one with the lowest ID. When extending a file system, the -is_greedy attribute is ignored unless there is not enough free space on the existing volumes that the file system is using. Windows event log auto archive With Windows operating system, applications can use the event logging mechanism to log their own events. Celerra currently supports three such event logs that are security, system, and applications. The physical format of these logs use a Microsoft format called 'evt' that has a limitation of 4 GB in size because there are some fields stored on 32 bits integers. Windows 2008 has introduced a new format 'evtx' that does not have this limitation. The event log auto archive feature allows you to automatically archive an event log on a particular trigger and to continue the logging on a new event log without losing any events. The archive is triggered on a time and/or on an event log size basis defined by parameters in the Windows registry. You can specify a retention policy to keep the event log archives before they can be recycled based on the duration and/or the total archive disk size. All parameters are stored in the Windows registry of each VDM. Therefore, each VDM will have its own configuration. The parameters can be viewed and edited with tools like regedit. The Configuring and Managing CIFS on Celerra technical module provides more information about this feature.
Version 6.0.40.8
Celerra Upgrade Tool -localization upgrade support The Celerra Upgrade Tool (CUT) allows you to download language packages and manage your Celerra in your primary language. The Download Software wizard helps you download the language packages from the EMC Online Support website based on your Celerra configuration. The Celerra Upgrade wizard then installs the language
EMC Celerra Network Server Version 6.0.70.4 Release Notes 7
New features and changes
packages on your Celerra. Celerra file services will not be disrupted during language package installation. This feature is supported on EMC Celerra 6.0 systems only. Unisphere upgrade procedure EMC Unisphere Service Manager (USM) is a collection of tools that help you update, install, and maintain your CLARiiON system. It also includes the Celerra Upgrade Tool (CUT) which consists of tools that help you perform software upgrades to Celerra. This procedure explains how to use USM to upgrade Celerra Unified Platform systems. 1. 2. Go to EMC Online Support and select Support by Product > Celerra NS> Downloads. Select the USM installation (executable) file (for example, Unisphere Service Manager (USM) (Windows) 1.2.2.1.0020) and download the file to a known location on your PC. In the folder where you save the USM installation file (for example, UnisphereServiceManager-Win-32-x86-en_US-1.0.0.1.0490-1.exe), double-click the executable (.exe) file, and if necessary, click Run to start the installation wizard. Follow the instructions that appear. After you accept the license agreement, the wizard verifies that the server is running the supported Java Runtime Environment (JRE) version. If the server is running an earlier version of the JRE, select Yes when asked if you want to continue with the installation. The installation wizard prompts you to download and install the later version of the JRE from the Sun website. If the server is running a later version of the JRE, the installation wizard asks if you want to continue with the installation. Select Yes to continue or No to quit the installation. If you select Yes, the USM may not work properly with a later JRE version. If you select No, and manually remove the later version of the JRE, other applications that require this version may not work properly. 5. When the installation is complete, click Done.
3.
4.
8
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
Register your Celerra to use the Celerra Upgrade Tool (CUT) 1. Register your system on EMC Online Support to enable CUT to download software. Go to the Service Center tab and select Register a Product. 2. 3. Enter the Product ID and click Continue. You are now entitled to use CUT.
Run the Unisphere Service Manager The USM will allow you to upgrade the NAS and FLARE software on your system. As a general rule, you should update the NAS version to the latest qualified release (6.0) prior to updating the FLARE version to its latest release (R30). Upgrade NAS 1. Double click the Unisphere Service Manager icon on your desktop.
2.
Click Login and log in to a Celerra. CUT supports version 5.6 system login as root only. The Name field is prefilled with the username root. The root password is required.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
9
New features and changes
3.
In the menu bar, click Software > System Software to run the Celerra Upgrade Tool. Three wizards will guide you, step-by-step, through the process of upgrading the Celerra software, which includes updating the Control Station and the blades. Select the wizards in the order presented on the screen: Download Software Pre-upgrade Health Check Perform Software Update
4.
The Celerra Upgrade Tool online help provides information about using the wizards. 5. Select the CUT files for download. For example: 6.0.36.4_apps.iso 6.0.36.4_emcnas_CD1.iso pkg_6.0.36.4_emcnas.tar
Note: After running the Health Check, be sure to resolve any errors before performing the Software Update.
10
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
Upgrade Flare 1. Double click the Unisphere Service Manager icon on your desktop. 2. Click Login and log in to your storage system. Enter the storage array admin account name and password and verify that Scope is set to Global.
3. 4.
In the menu bar, click Software > System Software and select the Prepare for Installation wizard. In the Prepare for Installation wizard, select the Download software and verify storage environment option, and follow the steps to the Select Software screen. Select the Any supported software option. This sequence of steps allows you to select only the Flare package for downloading, rather than all other associated but optional software items. When the Prepare for Installation wizard is finished, start the Install Software wizard. Follow the instructions to install the desired software packages, for example, Flare 30 release: 04.30.000.5.004. The actual download file is named CX4-Bundle-04.30.000.5.004.pbu.
EMC Celerra Network Server Version 6.0.70.4 Release Notes 11
5. 6.
New features and changes
7.
When the installation is complete, select Notify your service provider with updated storage system configuration information, and click Finish.
Version 6.0.36.4
The following new features are supported as of Celerra Network Server version 6.0.36.4. 256TB capacity/blade With the introduction of the Celerra 6.0 software release, we have increased the capacity/blade of the NSG8 and NS960 platforms to 256TB per blade. This enables users to increase the number of disk drives per blade without needing to acquire an additional blade, thereby reducing the user's TCO.
64-bit OS
The Celerra 6.0 release introduces the 64-bit version of the DART operating system. 64-bit technology enables blades with large memory configurations to effectively utilize system resources in order to provide capacity and scalability enhancements across the board.
Control Station auditing system
The system auditing feature has been enhanced with new trigger mechanisms to detect potential system access violations and to start up automatically on single and dual Control Station configurations. For more information, see the Auditing in the Celerra Control Station Technical Note.
Celerra Upgrade Tool
The Celerra Upgrade Tool (CUT) lets you upgrade your Celerra systems without assistance from EMC or an EMC service provider. CUT communicates over the Internet to EMC Online Support and over a local network to a Celerra system. CUT is part of the Unisphere Service Manager (USM), and you will now be able to use the same tool to upgrade both Celerra and CLARiiON systems. CUT uses three wizards to safely guide you, step-by-step, through the process of upgrading the Celerra software.
12
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
The Celerra Download Wizard selects the latest Celerra software for your system. Software selection is based on your hardware configuration, the software currently installed on the Celerra, and the storage system software running. The wizard can examine your system to obtain this information, or you can provide the information. The Download Celerra Software Wizard gives you one or two choices depending on your configuration, a maintenance release with the latest bug fixes or a new version with new features. The wizard then helps you download the appropriate version of the Celerra software to a central repository on a PC or laptop. The Pre-Upgrade Health Check Wizard examines the health of the system to determine if the Celerra will support a successful software upgrade. The Celerra Upgrade Wizard runs the pre-upgrade health check on the Celerra, upgrades the software, and then runs a post-upgrade health check.
Unified Storage system privileges required to run Celerra Upgrade Tool wizards With the release of NAS 6.0, Unified Storage systems can now participate in a CLARiiON Storage domain as a member system. After becoming a member of the CLARiiON domain, whether the system joined automatically or manually, global user accounts are automatically mapped to the Celerra with Operator (nasadmin) privileges by default. By design, the migrated accounts do not have the necessary privileges to run any of the Unisphere Service Manager's (USM) Celerra Upgrade Tool (CUT) wizards. USM displays only the Celerra Download and the Celerra Health Check wizards. The Celerra Upgrade wizard is displayed only if the fullnas (nasadmin) or root privileges are assigned to the migrated Users. In addition, to use all three CUT wizards, the Control Station shell allowed privilege must be enabled for each user. This right allows users to fully use the Celerra Download and the Celerra Health Check wizards. In order to run the Celerra Upgrade wizard, users must also have fullnas (nasadmin) privileges. To fully use the CUT wizards: 1. 2. Run Unisphere and enter a hostname or IP address. In the login dialog, log in to the Celerra as root.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
13
New features and changes
3. 4. 5.
From the Scope drop-down list, select Local. Select Settings > User Management > Users. Highlight the user account to which you want to assign privileges and click Properties. In properties, from the Group (Role) Membership field, select fullnas (nasadmin). From the Client Access field select Control Station shell allowed.
Once you have set these privileges for users, they can run USM and log in to a Celerra system and upgrade it with CUT.
Celerra VG2 and VG8
Celerra now supports the new low-end and high-end gateway platforms VG2 and VG8. These two platforms are based on the newest Intel Technology which provides greatly increased performance. The low-end platform, VG2, introduces UltraFlex™ technology for the first time in this level of NAS product and improves the system availability with the option of two Control Stations for high availability failover.
CEPA support for NFS audit applications
This feature provides event notifications and context to audit applications that monitor the NFS file system activity on Celerra, which run in the Windows environment. The audit applications require event notifications, from the Data Mover, to organize the access of information that is stored on the Celerra file system. To provide this functionality, the Celerra Event Publishing Agent (CEPA) API was modified to allow the applications to obtain the required Data Mover event information without blocking the CIFS or NFS client applications. To use this feature, the audit applications need to register for post-event notifications through the CEPA API. The CEPA API consists of an IDL file and an XML DTD file. These files contain information that is required by the audit application to interact with the CEPA. The audit application may coexist with the Celerra Event Enabler (CEE) framework on the same client (where Local RPC is used for communication) or on the remote client (where MS-RPC is used for communication). The CEE facilitates the use of selected third-party applications with the Celerra file systems. It provides event values containing the required context as defined by the audit applications for each class. As more vendor applications are added to each class, the events and associated context are modified to accommodate the applications.
14
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
The Celerra Event Publishing Agent technical module provides more information.
CLARiiON SATA EFD disk volumes
CLARiiON SATA EFD disk volumes are now supported, in addition to CLARiiON Fibre Channel disk volumes. Celerra support includes the ability for users to provision storage and create / extend file systems on these drives. The SATA EFD drives are treated identically to FC EFD drives which mean both drive types can be part of the storage pool, one drive type can be replaced with another, and file systems can span FC EFD and SATA EFD drives. The Managing Celerra Volumes and File Systems with AVM book has more details relating to EFD drives.
Control Station Linux upgrade
As of this release, the Celerra Control Station is upgraded to Red Hat Enterprise Linux 5 (RHEL 5). RHEL 5 leverages new security features, updates, and bug fixes in the kernel and applications provided by RedHat.
Dual port 10GbE SLIC for Data Movers
A new dual-port 10GbE UltraFlex I/O Adapter is being made available for the NS960, NSG2 and NSG8 platforms. This I/O Adapter doubles the effective throughput from the prior generation, which consisted of a single port card. In addition, the UltraFlex form factor allows for easy installation and upgrade to any chassis that supports UltraFlex adapters.
Express Install
Express Install introduces a method to reset the Celerra system back to its default factory configuration.
FAST Cache for the Provisioning wizard
As of this release, Celerra supports EMC FAST Cache as an extension to the dynamic random access memory (DRAM). This extended cache can potentially improve performance of the overall system by increasing write cache size at a comparably low cost. FAST Cache can be configured anytime by using Unisphere. The provisioning wizard in CSA and UMPW provides you with an option to
EMC Celerra Network Server Version 6.0.70.4 Release Notes 15
New features and changes
skip the provisioning of the EFD drives in order to configure them as the FAST Cache drives (maximum 6). You can continue to provision the remaining drives for file / block use.
FLARE R30
Celerra Network Server now supports the latest CLARiiON FLARE release, R30 when installed on CLARiiON CX4 storage arrays. It supports all R30 features, but does not actively support Fully Provisioned Pool LUNs, Auto-Tiering and Compression. Refer to the E-Lab Interoperability Navigator at http://Support.EMC.com for more specific FLARE version numbers.
TM
IPv6
IPv6 is fully supported and configurable by means of the CLI, and partially supported (read-only and some limited configuration) by means of Unisphere. CLI commands All CLI commands now support IPv4 and IPv6 addresses. There are some new commands provided specifically to handle IPv6. Some existing commands have updated syntax to support IPv6 addresses. Supported areas include: NFSv2/v3, FTP, TFTP, NIS, DNS, partial LDAP, Kerberos, CIFS, iSCSI, Network, NTP, NFSv4, Secure NFS, UserMapper, ReplicationV2, XLT, CDMS, FileMover, Control Station, SNMPv1 traps, SNMPv3, MirrorView, SRDF, CAVA/VirusChecker, ConnectEMC and Celerra XML API V2. The Configuring andManaging Celerra Networking technical module provides detailed information about IPv6 networking. IPv6 read-only support Unisphere provides IPv6 read-only support with a few exceptions. If IPv6 is configured by means of the CLI, that configuration is visible on the Unisphere. The information about auto-generated link-local interfaces and routes is not visible by means of Unisphere. The IPv6 addresses in drop down lists and checkboxes display IPv6 configuration information and are selectable. IPv6 addresses in hyperlinks navigate to the appropriate destination. Some screens allow the selection of IPv6 address for configuration in CIFS and Replication. The netmask field displays either an IPv4 subnet mask for IPv4 addresses or a non-editable IPv6 prefix for IPv6 addresses.
16
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
The broadcast address is blank for IPv6. The IPv4 network configurations are fully supported. When there are combined configurations of IPv4 and IPv6, changes via Unisphere are prohibited. Unisphere wizards, including the CIFS, Control Station, iSCSI, Network, Replication, and Set up Celerra wizards provide read-only support for IPv6. Lists containing IPv6 address are selectable, but not editable. CBM applications Client-side CBM applications may not work with an IPv6 configured iSCSI LUN. Celerra Monitor Celerra Monitor provides read-only support for IPv6. When IPv4 is configured, the screen displays the IP address, netmask, and broadcast fields. When IPv6 is configured, the screen displays IP address and the subnet prefix length fields. CSA The Celerra Startup Assistant (CSA) does not provide IPv6 support at this time. A new CLI command (nas_cs) provides IPv6 support. IPv4-only or IPv4/IPv6 environment? Should you configure IPv4-only environment or a dual stack IPv4/IPv6 environment? When deciding this, you need to answer the following question: In my environment, do any nodes or peers served by the Celerra use the IPv6 protocol? If you answer yes, the Celerra and associated infrastructure, such as routing, Domain Name System (DNS), and Network Information Service (NIS), should be configured for a dual stack environment. Link-local information The Celerra XML API V2 supports IPv6, however the auto-generated link-local information is not visible for the XML API V2. The application protocols are not supported via link-local interfaces. The link-local related information is supported only for IPv6 core protocols such as Duplicate-Address-Detection, Router Discovery, Neighbor Discovery, and ICMP6 and is exposed only via the server_ping6 command at this time. Only the CLI will have the view of all the IPv6 interfaces. MMC Microsoft Management Console (MMC) Snap-in provides full IPv6
EMC Celerra Network Server Version 6.0.70.4 Release Notes 17
New features and changes
support. That is, MMC accepts both IPv4 and IPv6 addresses that you input. Machines hosting MMC must be running Vista or Windows Server 2008 or later to support IPv6.
Kerberos authentication support for LDAP-based directory services
In this release Celerra adds support for Kerberos authentication to the LDAP-based directory services provided by Active Directory with Microsoft Windows Services for UNIX (SFU) or Identity Management for UNIX (IdMU). When configuring LDAP, the Data Mover’s directory client can now use Kerberos, simple (password), or SSL authentication. Anonymous authentication is not allowed. Kerberos authentication means the Data Mover, configured as a CIFS server, uses a KDC to confirm the Data Mover’s identity when accessing the Active Directory. The Configuring Celerra Naming Services technical module describes how to configure a Data Mover which is configured to use Active Directory with SFU or IdMU to use Kerberos authentication.
MAC OS 10.6
In this release, EMC qualified MAC 0S 10.5 and 10.6 in a CIFS protocol environment. File access protocol using native CIFS protocol stack in MAC-OS client has been verified in term of compatibility with Celerra CIFS implementation. The 'AppleSingle' feature which is using data stream to store resource information instead of a second file associated to the data file has also been validated (with MAC-OS 10.6, this is the default behavior on the client mount option).
MPFS Operating Systems
MPFS now supports Windows 7 clients (x86 and x86_64), Windows 2008 R2 (x86_64 only), CentOS 5.4 (x86 and x86_64), SLES 10 SP3 (x86, x86_64) and SLES 11 (x86 and x86_64 only). CentOS is supported over MPFSiSCSI only. All other operating systems are supported over Fibre Channel and iSCSI. For Windows 7 and Windows 008 R2, MPFS supports SMB1, SMB2 and SMB2.1 versions of the CIFS protocol.
18
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
NDMP
Increased NDMP backup streams The increased NDMP backup concurrent streams feature enables the Celerra Network Server to increase the maximum number of simultaneous concurrent backup sessions from four to eight. It verifies that the system memory is at least 8 GB, and then sets the value to eight. If the system memory is less than 8 GB, the value defaults to four simultaneous sessions. The Celerra Network Server Parameters Guide provides more detailed information about this feature. To support eight concurrent backup or restore streams: The backup client (DMA) must be compatible and configured to run eight concurrent streams. The Data Mover must have at least 8 GB memory.
Increasing the number of concurrent sessions to eight may result in an increase in the utilization of CPU and system memory. The Configuring NDMP Backups on Celerra technical module provides more information. NDMPcopy The NDMPcopy feature is an open source program that allows users to copy file system data within the Data Mover or between different Data Movers that are on the same or different Celerra. The NDMPcopy enables users to directly transfer data from the source to the destination file system, without traversing the NDMPcopy host machine or network (if the source and destination file systems are on the same Data Mover). It uses NDMP (Network Data Management Protocol) to transfer the file system data. With the NDMPcopy program, the NDMP host and network (if the source and destination file systems are on the same Data Mover) are not overloaded, because the data transfers take place directly from the source to the destination. The NDMPcopy program contacts the NDMP server processes on the source and destination file systems, and authenticates the connection. It then initiates data transfers between the source and destination file systems. The NDMP host receives status messages only. The NDMPcopy also provides multiprotocol (NFS/CIFS) support. EMC recommends using the NDMPcopy program for data transfers only. The NDMPcopy may affect the scheduling of backup activities in the same source, as it modifies the BackupDates file on the Data Mover. It uses the text mode for authentication and so the password is sent across the
EMC Celerra Network Server Version 6.0.70.4 Release Notes 19
New features and changes
network in a non-encrypted format. The Configuring NDMP Backups on Celerra technical module provides more information. NDMP port range The NDMP port range feature allows users to configure the port range to use a limited range of TCP ports that correspond to the ports for which the firewall allows inbound connections. EMC recommends setting the NDMP port range using dynamic or private ports (49152-65535) of IANA. The required minimum value of the port width is 32; however, EMC recommends setting the width to at least 100. Setting the port range does not imply that these ports are reserved for NDMP connection. When the port range is set, NDMP selects a random port number from the specified port range as the data connection listening port. If the port range is not set, the default port range is used. NDMP is incompatible with NAT (Network Address Translation) firewalls. It uses the existing server_param management interface and is independent of the DMA. Configuring the NDMP port range allows the Data Mover to use a limited range of TCP ports that correspond to the ports for which the firewall allows inbound connections. The Configuring NDMP Backups on Celerra technical module and the Celerra Network Server Parameters Guide provide more detailed information about NDMP port range.
pNFS
With this release, the Celerra Network Server provides a conforming implementation of NFSv4.1 in order to support pNFS. The pNFS is a part of NFS version 4.1 that allows clients to access storage devices directly and in parallel. In order to support pNFS, the Celerra Network Server fully supports all required NFSv4.1 functionality. The pNFS architecture enhances the scalability and performance of the existing NFSv4 servers. This is achieved by the separation of data and metadata, and moving the metadata server out of the data path.
Quotas
Choose and set a quota policy As of this release, administrators can change the quota limit check policy without turning off (disabling) all of the quotas on the file system. The new policy change will be applied after the Data Mover is rebooted.
20
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
The technical module Using Quotas on Celerra provides additional information about this change.
RSA Common Security Toolkit module
In this release the RSA Common Security Toolkit (CST) module replaces the CAMS Common Sign-On (CSO) module which was previously used to handle remote authentication of Celerra administrative users against an external directory server. CST changes the operation of the directory server settings test function. The Celerra Security Configuration Guide and Unisphere Settings > User Management online help provide more information about this feature.
Samba 4
Celerra supports Samba 4, except for the Samba 4 domain controller. Samba is an open source / free software suite that provides file and print services to all manner of SMB/CIFS clients, including the numerous versions of Microsoft Windows operating systems.
SAN Copy
SAN Copy™ replication is an array based replication technology. The performance and scalability of array based solutions is better for largescale migration than host based applications. SAN Copy is integrated into the Unisphere console and simplifies the configuration and management of LUN migrations through a user friendly GUI interface. The SAN Copy CLI Reference Guide and the on-line help contain additional information about SAN Copy.
SMI-S v1.4
The new SMI-S feature enables generic SMI-S compliant clients to manage a Celerra using CIM (Common Information Model) operations defined for NAS. This feature delivers two components of WBEM (Web Based Enterprise Management) infrastructure; namely a CIM Server and a Provider. Both components are embedded on the Celerra Control Station. SMI-S contains an API that storage management applications can use to provide standardized view of a Celerra. Providing a compliant API in SMI-S terms means supporting profiles and sub-profiles of the NAS domain. For Celerra, this means supporting the NAS Head Profile at a minimum, along with its mandatory subprofiles.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
21
New features and changes
SNMPv3
The Simple Network Management Protocol (SNMP) collects and processes valuable network information. It gathers data by polling the devices on the network from a management station at fixed or random intervals. In this release, a new SNMP agent, the SNMP daemon (SNMPD), which runs on the Data Mover, supports SNMPv1, SNMPv2c, and SNMPv3. SNMPv3 supports IPv4, IPv6, and enhanced security over SNMPv1 and SNMPv2c. SNMPv3 provides important security features such as: Authentication to verify that the message is from a valid source Encryption of packets to prevent snooping by an unauthorized source Message integrity to ensure that a packet is not tampered with in transit
Tolerance for 8 GB FC in Symmetrix
When a Celerra gateway with 4G Fibre Channel (FC) is attached through a switch to a Symmetrix platform incorporating an 8 GB FC card, there is no loss of or change in functionality. The communication rate automatically drops to 4 GB (Symmetrix platform data rate negotiation) and operation continues as expected.
Tools
Windows event log auto archive With Windows operating system, applications can use the event logging mechanism to log their own events. VNX currently supports three such event logs that are security, system, and applications. The physical format of these logs use a Microsoft format called 'evt' that has a limitation of 4 GB in size because there are some fields stored on 32 bits integers. Windows 2008 has introduced a new format 'evtx' that does not have this limitation. The event log auto archive feature allows you to automatically archive an event log on a particular trigger and to continue the logging on a new event log without losing any events. The archive is triggered on a time and/or on an event log size basis defined by parameters in the Windows registry. You can specify a retention policy to keep the event log archives before they can be recycled based on the duration and/or the total archive disk size.
22 EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
All parameters are stored in the Windows registry of each VDM. Therefore, each VDM will have its own configuration. The parameters can be viewed and edited with tools like regedit. The Configuring and Managing CIFS on VNX technical module provides more information about this feature.
Unisphere
Version 6.0.x introduces Unisphere. Unisphere is a unified SAN/NAS application providing uniform and intuitive management and monitoring services. It replaces Celerra Manager, Navisphere Manager, and Navisphere Express. Highlights of the new Unisphere application include: Single sign-on for a user identity common to multiple Celerra and CLARiiON systems. This functionality is facilitated by creating a global domain user who can also manage Celerra systems. This common user only needs to log in to one system in the domain in order to access and manage all systems. Management of up to 50 Celerra and CLARiiON systems within a single Unisphere instance. System summary views in a new dashboard that provides alerts and status summaries for the managed system list populated by the user. Participation in storage domains by both CLARiiON and Celerra systems. Cross-linking capability for LUN, disk group, storage pool, and storage system properties from Celerra to the relevant CLARiiON.
Note: All existing Celerra features, with the exception of message-of-the-day and system monitoring after session timeout, are supported within Unisphere.
The following table lists where each Celerra Manager view can be found in Unisphere. In some cases, Celerra Manager views map only to tasks found in Unisphere’s left-pane task list.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
23
New features and changes
Table 1 Celerra Manager to Unisphere Feature Mapping Celerra Manager Celerra Home > Status Celerra Home > Licenses Celerra Home > Statistics Celerra Home > System Info Celerra Home > Control Station Properties Data Movers Data Movers > Virtual Data Movers Data Movers > Properties > Notifications Storage > Systems Storage > Systems > Properties Storage > Systems > Properties > Disk Groups Storage > Systems > Properties > Spare Disk Devices Storage > Volumes Storage > Pools Storage > Pools > Usage Statistics Unisphere All Systems > Dashboard > Alerts System (System tasks) > Manage Licenses Monitoring (Statistics tasks) > All Performance System > Details System > Details System (System tasks) > Manage Control Stations System > Data Movers Sharing > CIFS Monitoring > Notifications System > Details System > Details Storage > RAID Groups Storage > Hot Spares Storage > Volumes Storage > Storage Pools Storage (Storage Pools tasks) > View Usage Statistics Monitoring (Statistics tasks) > Storage Pool Usage Storage > File Systems Storage > File Systems > Properties Storage > File Systems > Properties Storage > File Systems > Mounts Storage > File Systems > Properties Storage > File Systems (File Systems tasks) > Manage Quota Settings Removed Storage > File Systems Storage (File Systems tasks) > View I/O Statistics Monitoring (Statistics tasks) > File System I/O
File Systems File Systems > Properties File Systems > Properties > Checkpoint Schedules File Systems > Properties > Mounts File Systems > Properties > Checkpoint Storage File Systems > Properties > Quota Settings File Systems > Properties > Notifications File Systems > Mounts File Systems > I/O Statistics
24
EMC Celerra Network Server Version 6.0.70.4 Release Notes
New features and changes
File Systems > Usage Statistics
Storage (File Systems tasks) > View Usage Statistics Monitoring (Statistics tasks) > File System Usage1 File Systems > Properties Storage > File Systems Storage > File Systems (File Systems tasks) > Manage Quota Settings Replicas > Checkpoints Storage > Data Migration Sharing > CIFS Sharing (CIFS tasks) > View Statistics Monitoring (Statistics tasks) > CIFS Sharing > NFS Sharing (NFS tasks) > View Statistics Monitoring (Statistics tasks) > NFS Sharing > iSCSI Sharing (iSCSI tasks) > Manage Settings Replicas > Replications System > Network System (Network tasks) > Manage NIS Settings System (Network tasks) > Run Ping Test System (Network tasks) > View Statistics Monitoring (Statistics tasks) > Network Device Storage > Virtual Tape Monitoring > Event Logs Monitoring > Notifications System > Hardware Monitoring > Background Tasks All task lists Settings > User Management Settings (task list) > Manage Idle Timeout System > Network Settings > Public Key Certificates Monitoring (task list) > Launch Celerra Monitor n/a System (Network tasks)
EMC Celerra Network Server Version 6.0.70.4 Release Notes
File Systems > Deduplication Settings File System Quotas File System Quotas > Settings SnapSure Checkpoints Data Migration CIFS CIFS > Statistics NFS Exports NFS Exports > Statistics iSCSI iSCSI > Configuration Replicator Network Network > NIS Settings Network > Ping Network > Statistics VTLUs Events Notifications Inventory Task Status CLI Commands Security > Administrators Security > Celerra Manager Security > Network Services Security > Public Key Certificates Tools > Celerra Monitor Tools > Launch Navisphere Wizards – Network
25
New features and changes
Wizards – CIFS Wizards - File System Wizards – Replication Wizards – iSCSI Wizards – Celerra Support
Sharing (CIFS tasks) Storage (File Systems tasks) Replicas (Replications tasks) Sharing (iSCSI tasks) System (System tasks) System (Service tasks)
Upgrade media
Dual Compact Disk As of this release, the upgrade media is delivered on dual CDs. An outof-family upgrade, for example, V5.6 to V6.0 requires both CDs. The software upgrade prompts the user to swap CDs when necessary. DVD The upgrade media is available on one DVD for users of Control Stations that have optical drives capable of reading DVD discs.
Users and roles
Version 6.0.x introduces several changes to the Celerra user and rolebased access feature. CLI role-based user access A user account is always associated with a primary group and each group is assigned a role. A role defines the privileges (that is, the operations) the user can perform on a particular Celerra object. In version 6.0.x, Celerra extends support for assigning role-based privileges to a user accessing Celerra through the CLI. The ability to select a predefined role or define a custom role that gives a user the necessary privileges to execute a command is not available for all CLI commands. Certain commands can only be performed by the default user accounts root and nasadmin or, in some cases, by a user account associated with the root and nasadmin roles. You create and manage role-based user access in Unisphere Settings > User Management > Roles. You must be root or a user who has root or security operator privileges to create a user account and to associate it with a group and role. The Celerra Security Configuration Guide, Celerra Network Server Command Reference, and Unisphere Settings> User Management online
26
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
help provide more information about the specific command actions for which privileges can be controlled.
Unisphere storage domain users and roles
To simplify access and privileges for users managing both CLARiiON and Celerra systems in Unisphere, user accounts created to manage CLARiiON storage are mapped to Celerra user accounts, groups, and roles so that CLARiiON storage users can be given privileges to access and control Celerra objects. Groups and roles with the prefix sd_listed in Unisphere Settings> User Management views are mapped from CLARiiON storage domain groups and roles.
Known problems and limitations
CDMS
Connecting to a Solaris secondary server Users are unable to create a CDMS or a FileMover connection to a Solaris secondary server using TCP with IPv6. IPv6 with NFSv3 will not be able to create this connection. If UDP is chosen instead of TCP, this type of connection will work with a Data Mover NFS server. This problem has only been seen with Solaris NFS servers. Workaround Specify UDP as the transport protocol instead of using the default TCP. This option can be set with both the server_cdms and fs_dhsm commands. Celerra Event Enabler The following known problems and limitations pertain to Celerra Event Enabler: CAVA/CEPA servers Each Data Mover should have a CAVA pool consisting of a minimum of two CAVA servers specified in the Data Mover’s “viruschecker.conf” file, or a CEPA pool consisting of a minimum of two CEPA servers specified in the Data Mover’s “cepp.conf” file.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
27
Known problems and limitations
CIFS and NFS clients The CAVA feature of the CEE product is for CIFS and NFS clients. If FTP protocols are used to modify files, the files will not be scanned for viruses. No UNIX antivirus application support is provided. File-Level Retention It is strongly recommended that the antivirus (AV) administrator updates the virus definition files on all resident AV engines in the CAVA pools, and periodically run a full file system scan of the file system to detect infected FLR files. The Using Celerra AntiVirus Agent technical module provides additional information about virus definition updates. Microsoft Access Do not scan Microsoft Access database files in real time. Files inside a database are not scanned. Windows 64-bit operating systems In order to run CEE on Windows 64-bit operating systems, the Celerrato-CEE communications must be over Microsoft RPC (MS-RPC). The version of CEE that runs on Windows 64-bit operating systems is supported with Celerra version 5.6.45 or later and CEE version 4.5.0.4 or later. Celerra iSCSI AIX A known problem with SMIT (System Management Interface Tool) in AIX versions 5.2, 5.3, and 6.1 prevents it from creating and handling compliant IQN names for iSCSI nodes. As a workaround, manually edit the file /etc/iscsi/targets to enter or modify IQN names. Check the IBM support website for additional information. Configurations Celerra iSCSI configurations in the backend replication environment do not support TimeFinder®/FS, NearCopy, and FarCopy. Host connectivity issues Check the following guides for issues related to the Microsoft initiator, QLogic HBAs, and compatibility with the Windows operating system:
28
Host Connectivity Guide for Windows Host Connectivity Guide for Linux
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
These guides are available from EMC Online Support at http://Support.EMC.com. Microsoft iSCSI Software Initiator EMC Celerra iSCSI supports the most current version of the Microsoft iSCSI Software Initiator, which is available as a free download from Microsoft’s website (search for iSCSI initiator). This section contains known issues with the Microsoft iSCSI Software Initiator. Initiator uses only the first network portal of a target An iSCSI target on the Celerra Network Server may have multiple network portals configured for it; however, the Microsoft iSCSI initiator GUI uses only the first network portal of a target.
Note: Applies only to Microsoft iSCSI Software Initiator versions 2.0 and earlier.
Workaround Use the Microsoft initiator CLI to access other portals.
Control Station
Authentication failure Following a dual Control Station upgrade to Celerra Network Server version 6.0, users can log in successful using LDAP on the primary. When the Control Station fails over, the LDAP login no longer works. The following error is displayed: Session timeout. Relogin and try this operation again. This issue only affects LDAP users; local root/nasadmin accounts and global accounts are still available. Workaround Run /nas/site/cst_setup -reset to resync the CST settings. Once those settings are synced, you will be able to failover/failback and successfully be able to login each time using LDAP. Configuring LDAP domain settings When the Control Station is freshly installed and the Custom Active Directory page is used to configure the initial LDAP domain settings, the distinguish name is not set properly. Testing the Domain Settings generates an error similar to the following: Primary server 12.3.45.67 bind error. This issue only occurs during the initial setting. Once the
EMC Celerra Network Server Version 6.0.70.4 Release Notes
29
Known problems and limitations
initial setting is done, the issue never happens again. Workaround Use the 'Default Active Directory' option to configure the initial LDAP domain settings and then use the other options to change the settings. Copying large files to remote over NFS When a user copies very large files from the Control Station to a remote file system over NFS on a slow network, other users of that Control Station may experience slow performance when running CLI commands. For dual Control Stations, this issue can cause the Control Stations to fail over. Over time, the Control Station will correct itself, and performance will return to normal. Domain attributes After a Control Station failover on a Celerra that is a member of a storage domain, the administrator will not be able to use or see domain changes made while the Celerra was in the CS failover state. Workaround Continue using the Celerra with the domain attributes it currently has, or remove the Celerra from the domain and re-add it. Linux automount The /nasmcd/quota and /nasmcd/rootfs directories are used to nfs mount the root FS of the Data Movers using the Linux automount package. Occasionally, automount's attempts to perform this nfs mount fail, resulting in the requested Data Mover file system not being available. This problem happens more frequently when there are multiple automount requests for the same share over a short period of time. When the mount does fail, retrying the attempt after a short delay will usually be successful. One area that may be impacted by a mount failure is quota management, as it relies on access to files under /nasmcd/quota to retrieve quota information. The quota functionality and other operational aspects of the Data Movers are not affected. Security auditing When security auditing is stopped for any reason, for example, due to a lack of disk space, no alert is raised to the administrator.
30
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
Deduplication
Deduplicated data space not reclaimed A race condition occurs between removal of deduplicated files and file system unmount operations. Under certain conditions, deduplicated data space may not be reclaimed from the file system when files are removed. Messages similar to the following are displayed in the server_log when this issue occurs: 2010-05-09 09:40:57: 13163036672: DEDUPE: 3: Cannot find file system with id 15, cannot cleanup object <dart://rde..." Free Space During the deduplication process, the file system must have enough free space available that is equivalent to the size of the original file to be deduplicated plus the size of the compressed version of the file to be stored. An additional 1 percent of the file system must be free, or if autoextension is enabled, an additional 1 percent below the auto-extension threshold must be free. Restoring space-reduced files Restoring space-reduced files from a PAX-based NDMP backup into a file system causes the deduplication state to be set to suspended if the deduplication state is not currently enabled. I/O errors reported or deduplicated data space not reclaimed A race condition occurs during domain re-initialization. Under certain conditions, I/O errors can be reported or deduplicated data space may not be reclaimed from the file system when files are removed. Messages similar to the following are displayed in the server_log when this issue occurs: removeFromSecondaryStorage could not determine ssp
File-Level Retention
Prior to Celerra version 5.6.43.9, FLR protected empty locked files from being deleted. As of version 5.6.43.9, both FLR-E and FLR-C allow the deletion of empty locked files and empty append-only files. Any locked or append-only file that contains data will be protected. This behavior will be changed in a future release such that all locked and append-only files will be protected regardless of whether they contain
EMC Celerra Network Server Version 6.0.70.4 Release Notes 31
Known problems and limitations
data.
File system
Mounting file systems When mounting a file system as read-only on two (or more) Data Movers, the following error message appears on the copy file system pop up: "There are no Data Movers interconnect available on server_3, server_2". The error message is misleading and can be ignored. EMC does not support creating a copy of a file system that is to be mounted to more than one Data Mover. nas_fsck command A race condition occurs when using nas_fsck command on file systems smaller than 2G. This causes the file system to be unavailable to the user. Timestamp mismatch After a restore operation, a timestamp mismatch occurs on the softlinks with file names longer than 83 characters. The timestamp shows current time (i.e. time of restoration) instead of the original timestamp on the link.
Installation and upgrades
Celerra now joined to domain during upgrade The Celerra is now being joined to the CLARiiON domain automatically during an upgrade. You can use Unisphere to elevate the privileges of any user that has been transferred from the Domain to the Celerra. This does not occur for gateway machines, however. For gateways, users will need to join the Celerra to the domain they want it to be a part of. Celerra Startup Assistant After a fresh factory Express Install of an NS40FC system, the CSA fails on the Pre-Configuration apply changes page with the following error: CLARiiON's Domain Join Setup | Command Failed. If this occurs, ignore the error; this step should not have been tried, and failure poses no operational problems.
32
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
CallHome messages The Celerra uses connectEMC 2.6 software to deliver callhome messages to EMC. The outgoing mail functionality of this version issues duplicate HELO/EHLO commands. This is in accordance with RFC 2821, however, some mail servers do not support this (especially older mail servers). If this occurs in your environment, consider updating your mail server. Control LUN requirements enforced There is strict enforcement of the control LUN size and HLU numbers for fresh installations. An exact match between the LUN capacity and the correct HLU value is required. Mismatches will generate a "Control LUN Check Failed" error message during installation. You must correct the control LUNs before installation can complete successfully. Determining NAS version While using CUT to perform an upgrade from 5.6 to 6.0, the CUT tool presented a pop up saying it could not determine the nas_version, after or during the Linux upgrade portion of the upgrade. Retrying the task succeeds. Error message While upgrading from version 5.6 to 6.0, the following error message is displayed: Error - cannot open /usr/sbin/ext2online /dev/hda1. This error can be ignored, as it does not stop the upgrade and the task completes successfully. Gateway and FC-enabled systems During upgrades of gateway and FC-enabled systems, a message stating that the Celerra was joined to the back-end domain is presented. That message is incorrect, and can be ignored. The Celerra can be joined to the domain using CLI commands or via UEM on the domain master system. Incorrect certificate Users may experience an invalid certificate after fresh installing their system. Workaround Set the correct time on the Control Station and regenerate the certificate, or wait for the certificate to become valid and login again.
EMC Celerra Network Server Version 6.0.70.4 Release Notes 33
Known problems and limitations
nasadmin credentials If you pause and resume the upgrade, the "Clariion domain nasadmin Credentials" screen is not displayed. Workaround Use the nas storage command to change the password of nasadmin. nasadmin password When logging in globally and attempting to change the nasadmin password, an authentication error is generated and users are not allowed to change the password. The workaround is to use a nas storage command to change the password of nasadmin. Password confirmation The "Clariion domain nasadmin Credentials" screen in the Celerra Upgrade Tool (CUT) currently does not have a confirm password field. Users can mistype the password without knowing it, and will be unable to log in. PUHC "Check for running tasks" fails While upgrading from version 5.6 to 6.0 using USM, the PUHC "Check for running tasks" fails and offers a set of actions to take. These actions do not work. The first step says to type the following; "<br> * nas_license -list". Neither of the 2 steps offer strings work if typed in as specified and produce errors. Supported hardware When attempting to run USM on unsupported hardware (CNS|CFS|NS350|NS500|NS600|NS700), the Celerra Upgrade Tool will correctly exit but exit code is not passed back to the USM. Therefore, the USM tries to continue the upgrade process but will fail on the next step that interacts with the Celerra system. The error returned at that point does not give a clear indication of what the actual failure was. If you see unclear error messages while using USM, make sure that you are using a supported hardware platform. Upgrade paused During an upgrade with the CLI, and when it was paused during the Control Station reboot, the Upgrade wizard and on the welcome screen tells user that the upgrade is paused and allows user to resume the
34 EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
upgrade. Do not use the wizard to resume an upgrade that is started in the CLI. Multibyte support Using multibyte characters in fields in which they are not supported. If you use multibyte characters in data element fields that do not support multibyte characters (for example, checkpoint schedule descriptions), these fields will not display correctly. If you want these fields to be readable, you will have to recreate or rename the existing data element text using the ASCII character set. Refer to PRIMUS emc224755 for more detailed information about this limitation. Using multibyte characters with ECC Celerra version 5.6.40.3 and later may cause discovery and name display issues for ECC 6.x if certain Celerra object names use non-ASCII characters. The relevant names are: File system name Mount point name CIFS share name NFS export name Quota user name
Discovery issue If non-ASCII characters are used and if the name is more than 63 characters, discovery may not work. Garbled name display issue If these names use non-ASCII characters, the names do not display properly in ECC. However any ASCII characters used in the name do display properly. Furthermore, the name itself and the underlying data on the Celerra system are not affected. Suggested Workarounds Use only ASCII characters in the specified object names when using ECC for Celerra discovery. If multibyte characters must be used, keep all applicable object names under 63 characters. Since ASCII characters display properly, append or prepend ASCII characters to object names so you can use these characters to find the correct name.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
35
Known problems and limitations
Refer to Primus emc229599 for more details about this issue.
NDMP
Incremental NDMP backups and NDMPcopy operations Incremental NDMP backups and NDMPcopy operations use the same last backup timestamp tracking for each data set on a Data Mover. A data set in this context means the point where the backup or copy operation starts (e.g., the root of a file system or a specific sub-directory within a file system). This means that if incremental NDMP backup and NDMPcopy operations of the same data set (i.e., starting from the same point within a file system) are interleaved, then the following will occur: The incremental backup will contain only changes in that data set since the NDMPcopy operation instead of since the last backup operation. The incremental NDMPcopy of that data set will copy only those files changed since the NDMP backup instead of since the last NDMPcopy operation. To avoid this, do not interleave NDMP backup operations with incremental NDMPcopy operations of the same data set; if you do, perform a full NDMP backup of that data set after any NDMPcopy activity. Vendor support Not all NDMP vendor versions support this version of Celerra Network Server. For the latest NDMP Vendors/Versions/Patches and restrictions supported in this version of Celerra Network Server, refer to the Backup section of the E-Lab Interoperability Navigator.
Network
Do not disable IP reflects when using IPv6. System may panic when using IPv6 if IP reflect is disabled.
NFS
Cached files If a client mounts exported parent and child paths of the same file system, it is possible to access the same file from two different mount points. If client side caching is enabled, the client may end up maintaining two different versions of a cached file.
36
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
Celerra Data Migration System Access to online migration file systems while using CDMS is not supported by NFSv4 clients until migration completes. When NFSv4 is used, it triggers directory migration of subdirectories and requires global lock for the entire Data Mover. Therefore, the Data Mover will be inaccessible during the migration. NFS export failure If the export of a corrupted file system fails, the export entry of the file system which was corrupted & cleaned using FSCK is not seen in the export list. However, if user manually exports it again, the entry can properly be seen in the NFS export.
Quotas
Performance statistics Termination summary is not sorted for Set statistics if a statpath name is duplicated in the server_stats command. The following example will not have a sorted termination summary: server_stats server_x monitor store.diskVolume,store.diskVolume -sort writes. This issue does not impact the sorting of the sampled data displayed periodically during the server_stats session, only the termination summary at the end of the session is impacted. Workaround Not using duplicate statpath names when -sort is specified will eliminate this issue. For example: server_stats server_x –monitor store.diskVolume -sort writes.
Replication
nas_copy command When using Replication V2 and the nas_copy command, there is a window where if the X-Blade reboots, results may not be properly reflected. The transfer will complete but the task is not cleaned up and the appropriate status is not reported back to the user. Once completed, the task can be manually aborted.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
37
Known problems and limitations
Security
Changing LDAP settings When running the server_netstat command, or updating the LDAP settings, the following error was displayed: The lockbox stable value threshold was not met because the system fingerprint has changed. Reset the SSV values by running "/nas/sbin/cst_setup -reset" from the CLI to resolve the issue.
SnapSure
Parameter file Do not insert a new-line character at the end of a nas_param file (/nas/site/ or /nas/sys/). The empty line causes fs_ckpt to return an encodeable.extract error. Writable checkpoints operations Writable checkpoints do not support the following operations and features: Checkpoint refresh Checkpoint schedule Celerra replication CDMS File-level retention File system extension iSCSI MirrorView/SRDF TimeFinder/RDF VTLU
Data mover panic (425057) The data mover panicked. The following error was displayed: DART panic/fault message: >>PANIC: I/O not progressing LastVol touched Sn891 Kind 5 (ptr=0x65d74b08) Workaround Contact your service provider.
38
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
SRDF/S
nas_rdf –activate The nas_rdf –activate –reverse command may fail to swap device-pair personalities during the activate. This occurs when the source Celerra has a TimeFinder/FS snapshot/s created on R1BCVs which use TimeFinder Emulation mode (Symm 8 supports Emulation mode only). The error message below is specific to the personality swap and having remotely protected storage, and does not impact the successful activation of the destination Celerra system. Error message in nas_rdf –activate –reverse: An RDF 'Swap Personality' operation execution is in progress for device group '1R2_500_3'. Please wait... Cannot use the device for this function because it is a Copy Session target ERROR 13431799810: Swap operation of device group 1R2_500_3 failed due to the improper backend state or configuration. Run 'nas_message -info 13431799810' to get full description and recommended action. Workaround Canceling the relationship between the R1BCV mirrored volumes and the R1STD volumes is needed to complete the swap operation. After the SRDF restore completes, the SYMAPI will automatically request a full establish during the subsequent fs_timefinder –M refresh command to the snapshots built on the canceled R1BCV volumes.
Table 2 nas_rdf –activate –reverse procedure Step Action
1
To check symm id of source Symmetrix, run /nas/symcli/bin/symcfg list on the destination SRDF system. To check the mirrored volumes, run nas/symcli/bin/symmir –sid <source symm id> list on the destination SRDF system. You can also double check if Emulation mode is used from the output. To find R2 device group, run /nas/symcli/bin/symdglist on the destination SRDF system.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
2
3
39
Known problems and limitations
4
To find R1BCV volumes, run /nas/symcli/bin/symrdf –g <r2 group name> query –bcv on the destination SRDF system. If the same device ID is seen in the output that is the R1BCV device you need to cancel. To prepare for the cancel operation, create a file which has all R1BCV volume pairs to cancel on destination SRDF system. Example of the file: # cat file # <standard device id><bcv device id> 0068 00B8 0069 00B9
5
6
To cancel the R1BCV volumes, run /nas/symcli/bin/symmir –sid <source sid> -f <file name> cancel on the destination SRDF system. To confirm the canceled volume, run /nas/symcli/bin/symmir –sid <source symm id> list on the destination SRDF system. The canceled volumes should not be listed in the output. To compete the activate reverse operation, run /nas/sbin/nas_rdf –activate –reverse on the destination SRDF system.
7
8
nas_rdf –restore The nas_rdf –restore command may fail on SRDF system which uses TimeFinder Emulation mode (Symm 8 supports Emulation mode only) with R2BCV mirrored volumes. Error message in nas_rdf –restore: An RDF 'Failover' operation execution is n progress for device group '1R2_500_3'. Please wait... Cannot use the device for this function because it is a Copy Session target CRITICAL FAULT: SRDF restore failed for 1R2_500_3 Workaround Canceling the relationship between the R2BCV mirrored volumes and the R2STD volumes is needed to complete the restore operation.
40
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
After SRDF restore completes, full establish is required when we do mirror on again to the canceled R2BCV volumes.
Table 3 nas_rdf –activate –restore procedure Step Action
1
To check the mirrored volumes, run /nas/symcli/bin/symmir –sid list on the destination SRDF system. You can also double check if Emulation mode is used from the output. To find R2 device group, run /nas/symcli/bin/symdg list on the destination SRDF system. To find R2BCV volumes, run /nas/symcli/bin/symrdf – g <r2 group name> query –bcv on the destination SRDF system. If the same device ID is seen in the output, that is the R2BCV device you need to cancel. To cancel the R2BCV volumes, run /nas/symcli/bin/symmir –g <r2 group name> cancel -force on the destination SRDF system. To confirm the canceled volume, run /nas/symcli/bin/symmir –sid list on the destination SRDF system. The canceled volumes should not be listed in the output. To complete restore operation, run /nas/sbin/nas_rdf – restore on the destination SRDF system.
2 3
4
5
6
Nas_rdf –init command fails Problem: During the SRDF discover remote storage devices process, the nas_rdf init command fails and returns the following error: "Unable to setup remote devices". Workaround: Delete the device group for R2 devices using /nas/symcli/bin/symdg delete <DgName> -force. For example: /nas/symcli/bin/symdg delete 1R2_500_13 –force. Then run nas_rdf –init, to re-create the R2 device group.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
41
Known problems and limitations
SSD drives not supported SRDF/S does not support solid state disk (SSD) drives. If you attempt to configure an SSD drive with SRDF/S, Celerra health check operations reject the device as an unsupported storage.
Unisphere
The following notes apply to Unisphere. Adding a system to a Celerra-CLARiiON domain When users attempt to add a system to a Celerra-CLARiiON domain and uses a LOCAL scope to authenticate to the remote system, it appears to work. However, the system does not actually get added to the domain. The correct procedure is to authenticate to the system to be added with a GLOBAL scope user that is part of the root group. Adding a system to a Celerra-Symmetrix domain When attempting to log into a Celerra that has a Symmetrix backend and is part of a storage domain, you may see the following message on the console: "Target system already managed." If this occurs, log out, wait 15 minutes, and log in again or close the domain master page and restart your browser session. Adding a system to a second domain Adding a Celerra to one domain followed by adding it to a second domain from a non-master CLARiiON may result in the Celerra appearing to be in both domains at the same time, depending on which system scope you are in. Workaround Remove the Celerra from the first domain before adding it to a different one. If you have already added it to both you can still remove it from the one that you are trying to get it out of. File Type field is Blank in Export to File Dialog with JRE 1.7 The “File Type” field is empty when the “Export to File” dialog is displayed. This happens only with JRE 1.7. No file is created in the destination folder and the dialog closes without an error. Workaround The workaround is to select the file type from the “File Type” drop down.
42
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
JRE install auto link gives an error Unisphere will prompt a message to install JRE 1.6 if JRE is not installed. The JRE install link in this message does not work. Workaround The workaround is to install JRE 1.6 directly from the Oracle website.
Alerts For performance and memory usage reasons, the alerts displayed in Unisphere are now limited to 400. A new alert has been added to notify the administrator if there are more than 400 alerts:
Severity: Critical System: <hostname> Domain: Local Created: <timestamp> Message: Too many alerts on <hostname>. Full Description: This system has more than 400 alerts. Until you delete the alerts, you will only see the top 400 alerts (by date and severity) for this system. Recommended Action: Delete the alerts on this system to clear them from the list of alerts. Event Code: 0x240060005
Do not attempt to delete the "Too many alerts" alert in Unisphere. Unlike most of the other alerts, this alert will clear itself if the condition is resolved and the total number of alerts for a system is reduced to fewer than 400. If you include this new alert in a selected list to be deleted, the alerts will appear to be deleted, but will be reloaded once a new browser session is started. A failure will be posted to /nas/log/webui/apl_tm.log, but no failure will be displayed in Unisphere. In /nas/log/webui/apl_tm.log:
2010-06-24 10:20:21.260 webui-N:[[email protected]]:0:29366:S:402345-Delete alert(s). 2010-06-24 10:20:28.783 webui-N::0:29366:E:402345-Delete
EMC Celerra Network Server Version 6.0.70.4 Release Notes
43
Known problems and limitations
alert(s).(complete=true,failed=true) <- Note failed=true
Browsers EMC requires the use of one of the following browsers to run Celerra Manager: Internet Explorer 6.0 SP2, 6.2.3 through 7.x.; Internet Explorer 7.1 is recommended. Internet Explorer 8.0 is also supported. Firefox 3.0 or later.
Firefox Firefox Password Manager (PM) is enabled by default. This feature records and stores the password that you enter on a website and then automatically fills in the password entry field on your next and subsequent visit. The password field masks the password with asterisks (*). To prevent Firefox from remembering passwords: 1. 2. Click Tools > Options... and select the Security tab. Under Passwords, clear the Remember password for sites checkbox.
Browser session hang Depending upon a client's particular browser configuration (browser type, version, JRE version, native OS, etc.), launching the USM tool from the UEM may result in a browser session becoming hung. If you verify that there are no hidden dialogs awaiting confirmation, terminate the browser sessions and all associated processes. Rebooting the client system may be required to restore full browser operation. To prevent this issue in the future, launch the USM directly (i.e., not via UEM). Changes from other user interfaces Any changes made to the network server using another user interface such as the Celerra CLI are not automatically shown in Unisphere. Refresh the browser to see these changes. Changing default password Changing the CLARiiON array default admin password breaks the ability of the Celerra to communicate to the CLARiiON or make any changes to its database [diskmark, discoveries, etc. will all fail].
44
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
Checkpoint schedule page If the time zone of the Celerra Control Station and the client system running the UEM are different the checkpoint schedule page will display times in UTC (coordinated universal time). Confirmation dialog inactive While using Unisphere, a confirmation dialog appears briefly, and then moves to the background. Users need to manually re-focus the window and bring the dialog box to the foreground. Dashboard screen The data is not presented in the Celerra Capacity View screen in an appropriate amount of time. However, the other dashboard screens are loaded successfully. Workaround Close the Unisphere window and clean out the persistence files. On the next restart of Unisphere - you will not see this issue. To clean out the persistence files, follow these steps: 1. 2. 3. 4. Close the Unisphere application. Go to the Unisphere client log folder; the default should be C:\Documents and Settings\<username>\emc\Unisphere. Delete the persistence.ser file. Re-launch the Unisphere application
Dashboard summary The Dashboard Summary page did not contain view block. This is unexpected, but those blocks can be re-added using the customize button. File system delete confirmation The 'OK' button is disabled on the Unisphere File System Delete Confirmation page for a file system with File-Level Retention enabled at the Compliance level. The table on the File System Delete Confirmation page indicates that protected files still exist on the file system even though the File System Properties page and the 'nas_fs -i' CLI command both show that all file protection has expired for the file system. The user is unable to delete the
EMC Celerra Network Server Version 6.0.70.4 Release Notes 45
Known problems and limitations
file system using Unisphere. Workaround: The File-Level Retention status is now cached for performance reasons. Force the file retention data to be updated by selecting the Unisphere Refresh for the File System List page (in the upper right corner of the File System List panel) or by opening a new browser session. Help sessions While getting a session timeout in a help session, some help text is displayed, but none of the icons are displayed. Clicking OK button has no effect. Workaround: Use the close button of browser to close the window. JRE memory limit For optimal UI performance, EMC recommends that the memory limit allocated to the JRE be increased. The default size of 64 MB is often too small and may result in Out of memory errors. To set the memory size, from the Start menu in Windows, select Settings > Control Panel. Double-click the Java Plugin icon. Select the Advanced tab. In the Java Runtime Parameters, enter the following: -Xmx512m The argument specifies the maximum memory. Typically, about ½ the size of your overall workstation's memory is recommended as the maximum memory value. For instance, if your workstation has 1 GB of memory, then use -Xmx512m for the JRE runtime parameter. If your workstation has 2 GB or more memory, you can adjust the max JRE memory as needed to avoid running out of Java memory and possibly improving overall performance. Logging in after browser crash After a browser crash, the GUI reports "system already managed" when user tries to log back in. Logging in via multiple screens Logging into Celerra via two or more screens may succeed, but logging out of one of the sessions causes the other sessions to lose contact. Login screen When you log into Unisphere via a Celerra system, Unisphere will not display a login screen if you click the “yes” button on the security
EMC Celerra Network Server Version 6.0.70.4 Release Notes
46
Known problems and limitations
warning dialog, without also checking the “Always trust content” checkbox on the same dialog. MMC snap-in MMC installer program may not completely uninstall all components when requested to do so. However, when a new version is installed, the components are overridden and the application runs as expected. Modifying system properties When storage system properties (specifically - security credentials) are modified in the GUI, the following (example) warning message is displayed in addition to the expected icons: Modify CLARiiON CLARiiON CX4-960 APM00093300340. Invalid component information :BackendResume: This error is a result of the component information cron job querying the status while the credentials are in the process of being modified. The correct component information will be refreshed during the next iteration of the cron job (approx. 5 to 10 minutes). Mounts table view The file system column in the Mounts table view gets stuck at "Loading" for multiple minutes. The rest of the table is populated correctly. Workaround: Refresh the page using the green arrow icon. nbsnas/jserver filling up at 100% Unisphere jserver mounted on /nbsnas/jserver filling up at 100% prevents installation. When this occurs, the following error is displayed:
The feature identified has detected a potential capacity issue that may need the attention of the system administrator. The device name and mount point is indicated in the event posting. Recommended Action: Check that the mount point's condition is in fact approaching cautionary levels. Use normal administrative procedures to see if some free space can be returned to the mount point, such as removing unnecessary files or reorganizing areas of this mount point off onto other less full mount points on the system. If no means are available to reduce the space consumption on this mount point then you may want to investigate if there is a possibility that this mount point needs to be expanded to allow for more free space. Contact your Authorized Service Provider if you need assistance with this process or need to discuss acquiring additional storage system capacity.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
47
Known problems and limitations
Provisioning wizard If an error occurs while in the Provisioning Wizard, you are instructed to correct the error and are given a choice of either restarting the provisioning process or canceling the activity. Rolling back intermediate tasks may fail or hang. If you cancel the activity by clicking the NO button, all the previous tasks are rolled back and you will be redirected to the first screen of the Provisioning Wizard. Slice volumes not listed in wizard Occasionally, slice volumes created through the CLI or Unisphere Volume Create page are not listed as available volumes in the New File System Wizard when selecting Manual Volume Management and then attempting to create a new stripe volume from the slice volumes. Storage domain When the Celerra belongs to a storage domain that is destroyed, the storage domain users on the Celerra Control Station are not automatically moved to the "defunct" state. To correct this, manually remove all the storage domain users from the system by running the command: /nas/http/webui/bin/admin_management.pl defunctStorageUsers Then remove the Celerra from the domain. User’s tab displays unrelated messages Unisphere > Settings > User's tab displays messages which are unrelated to that screen. These messages are a result of a query that was initiated by the user on some previous table screen. Workaround Visiting another table screen and then to coming back to the table screen where the messages were seen should remove these messages.
Other restrictions
Caution Do not install Linux non-Celerra applications on the Control Station. Doing so consumes resources and might adversely affect system operation. Celerra Control Station is a dedicated controller and not an applications platform.
48
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Known problems and limitations
Do not manually edit the nas_db database without consulting EMC Customer Service. Any changes to this file could cause problems with your Celerra installation.
Command line scripts The NAS database stores specific information required for each Data Mover. The Celerra Network Server automatically performs a backup of the entire database every hour and saves it to a file named nasdb_backup.1.tar.gz. These backups start one minute after the hour. EMC recommends that you check your scripts to make sure they are not scheduled to start at the same time as the backups. During backup, the database becomes locked or inaccessible, and some commands that rely on the database might fail. Correctable hardware errors Transient correctable hardware errors are currently being handled by the firmware. Although rare, if enough of these errors occur, a watchdog panic can be triggered. In a future release, these errors will be handled by the Celerra operating system instead of the firmware which will improve efficiency and prevent the watchdog panic. server_cpu The server_cpu ALL -halt now command does not complete when using Celerra NSX Network Servers. This is because the NSX needs at least one Data Mover to be up in order to access its NAS partitions through NBS. server_http The server_http command fails if you issue it on a Data Mover while running fsck. Other commands (including FileMover API) may not work until fsck completes. server_stats On a Data Mover with more than 500 mounted file systems, the following statistics fail to display: - store.logicalVolume.metaVolume metaVolumes-std. If any of these statistics are supplied as an argument to server_stats -monitor option, the server_stats will hang. Use Ctrl-C to terminate the session. server_sysconfig If you use the option string of server_sysconfig, the entire string must be
EMC Celerra Network Server Version 6.0.70.4 Release Notes 49
Problem prevention
enclosed by double quotes. Virtual device naming When creating virtual devices, such as Fail-Safe Network (FSN) devices, Ethernet channels, and link aggregations, do not use the following names as the name of the device:
Amgr Blksrv Client Cmfap Dep dfe echo escon http i82596 icmp icmp6 ip ip6 lockd lpfn Mac meshforwarder nd nfs nfsPrimary nwh pax pcnfs pfkey pipe pmap rcp rfa scsicfg smb smpte spstream Ssl sslpipe statd StreamHead tcp tcpnfsmesh tcprpc tcprpc Tftp Toeframechk upd udpnfsmesh
Problem prevention
The EMC Customer Service center has reported a number of calls concerning the following issues. Please take a moment to familiarize yourself with these potential issues.
Administrative activities
Log in as nasadmin As with any UNIX-based operating system, refrain from performing normal administrative activities as root user. Log in using the administrative user account instead. Shutting down a Data Mover or deleting an internal interface Shutting down a Data Mover or deleting an internal interface will prevent you from issuing any server_ commands against that Data Mover. Whenever you issue any command on the Celerra Control Station CLI server, the command is sent to and executed by a Data Mover. If that Data Mover is down or unreachable, the command will fail.
50
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Problem prevention
Celerra AntiVirus Agent (CAVA)
AntiVirus (AV) threads If a large number of files need to be checked at the same time, the CAVA file checking process may be slow depending on number of threads configured, number of CAVA servers, and speed of the CAVA servers. If the CAVA servers cannot process file checks quickly enough, then all AntiVirus (AV) threads will become busy and the AV “collector queue” will increase, indicating that file check requests are queueing up. Increasing the number of AV threads (10 by default) will allow more files to be checked in parallel and may improve performance. However, the number of CAVA engines may need to be increased to satisfy a heavy load. Number of CAVA engines It is possible to configure your EMC Celerra installation with only one Celerra AntiVirus Agent (CAVA) engine. However, this non-highavailability configuration can make your installation susceptible to data unavailability in the event that the CAVA engine goes offline for any reason. The best method for preventing this type of outage is to increase the number of available CAVA engines. Additional information for configuring the Celerra AntiVirus Agent can be found in the Using Celerra Anti Virus Agent technical module. This document describes the parameter setting that will specify shutdown actions to take when no servers are available. Removing AV user from Data Mover password file If you remove an AV user from the Data Mover password file, CAVA will not work. Depending on the setting you have on CIFS, you could use up all the CIFS threads because they are used for virus checking or it will shut down CIFS altogether.
CIFS
Deleting machine account from active directory Deleting a machine account from active directory will result in loss of access to the CIFS server and its resources. Deleting the computer account of any CIFS server, including a Celerra Data Mover, from the domain of which it is a member will prevent that CIFS server from being
EMC Celerra Network Server Version 6.0.70.4 Release Notes 51
Problem prevention
able to authenticate both itself and users attempting to connect to it with the domain. This will mean that users will not be able to access data hosted by the CIFS server. Deleting server out of the domain Deleting a server out of the domain will result in shares associated with the server to be unavailable. The same applies to deleting the machine account. Naming CIFS shares Do not name your CIFS share "all" or you risk deleting all of your shares when trying to delete it. NetBIOS name conflict Instead of destroying the conflicting NetBIOS, simply rename it using the existing server_cifs <server_x> -rename -netbios <oldname> <newname> command. Changing a NetBIOS name using this command will update the share and local group and enable you to reuse it. LGDB and ShareDB information associated with the server are persistent. If the target name of a rename conflicts with LGDB or ShareDB information, re-create the old server and rename it to a name that will not conflict with your naming structure to free the old slot.
Control Station
Applications The Control Station’s primary function is to manage the Celerra. The use of applications that do not come installed on the Control Station is not recommended. The Celerra Control Station is not tested or certified with any of these applications and there could be issues with memory usage or CPU time that may affect normal operation. Dual Control Stations If you have dual Control Stations on either (CFS/CNS/NSxxx) and they concurrently try to access the /nas file systems, the nasdb may become corrupted and the Data Movers will be unable to communicate with the Control Station.
52
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Problem prevention
File System
Deleting files from Data Mover or file system Files that you delete will be permanently lost unless you have a backup copy. Files can be backed up and restored through the Celerra SnapSure, TimeFinder/FS, Replicator or by using standard tape backup methods. Deleting file systems File systems that you delete will be permanently lost unless you have a backup copy. Backups can be created via TimeFinder/FS, Replicator or other backup methods. File system full Insufficient space in a file system will prevent users from writing to that file system. In addition to monitoring file system usage, set an event notification for file system full condition and enable automatic file system extension with a conservative threshold. Refer to the Configuring EMC Celerra Events and Notifications technical module for more information. Mounting file system to /tmp Some variants of UNIX or Linux periodically delete some or all of the files under the /tmp directory. If you mount a file system under /tmp, you risk having the contents of that file system deleted by the automatic cleanup process. Splitting BCVs When a production file system (PFS) is mounted in a read/write fashion to a Data Mover and has BCVs established, the command to split (or mirror -off) the BCVs from the PFS volumes must be initiated using the Celerra fs_timefinder command. The fs_timefinder command ensures consistency of the BCV snapshot by managing file system access and Data Mover log operations when the volumes are split. These additional considerations are not available directly from Solution Enabler commands (symmir).
Name Resolution SunOne/iPlanet
When using SunOne/iPlanet, make sure that the defaultServerList
EMC Celerra Network Server Version 6.0.70.4 Release Notes 53
Problem prevention
attribute on the LDAP server is set to use IP addresses and not host names. If hostnames are specified and the Celerra is configured to use SunOne/iPlanet for host resolution, a deadlock can occur.
Networking
Changing an IP address Changing an IP address on a Data Mover can result in a naming (DNS/NIS) mismatch and render the IP address inaccessible. Changing or removing DNS settings When you make changes to the DNS server, such as IP addresses make sure you reflect those changes in the Data Mover. Failure to do so may result in accessibility issues. Removing the DNS server will prevent the CIFS server from communicating with the domain. Any CIFS server, including those configured on a Celerra Data Mover that is a member of a Windows 2000 or later domain, use DNS to locate the Windows domain controllers that it needs to communicate with to operate. Disabling DNS in such an environment will prevent the CIFS server from locating Windows domain controllers, therefore, prevent the CIFS server from authenticating users attempting to connect to it. This means that users will be unable to access the data hosted by the CIFS server. Connecting an internal switch to an outside network The internal networks in the Celerra are designed to provide private communication paths between Data Movers and the Control Station. If that network is connected to an external network, the resulting routing, addressing issues, and/or volume of network traffic may disrupt the communications between the Data Movers and the Control Station, which will in turn disrupt the operation of the Celerra. Duplicate IP Duplicating an IP address on two Data Mover interfaces will have the same result as duplicating IP addresses on any two machines in a network. The network will either be unable to determine which location it should deliver packets to or block all traffic to both machines on the assumption that one of them is attempting to gain unauthorized access to information by impersonating the other.
54
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Problem prevention
FSN devices Create Fail-Safe Network (FSN) devices on multiple network interface cards (NICs) to ensure connectivity in the event of a hardware failure. Incorrectly adding network route Incorrectly adding a network route will cause network access to be limited to the same subnet/broadcast domain. The default gateway, or route, address tells any network client, including a Celerra Data Mover or Control Station, where to send IP packets that are outside its local subnet. If the default gateway is misconfigured, the network client will not be able to contact any host that is not within its local subnet. If the Celerra relies on hosts, such as domain controllers, outside its local subnet, it may not be able to offer service.
Parameters
Changing parameters Changing parameters could significantly affect the function of the Celerra Network Server. Take the following precautions before you make changes to Celerra parameters: Do not change system parameters without first consulting with EMC Customer Service. Change server parameters one at a time, in small increments, preferably in a non production environment. Be sure to thoroughly test each change before making additional changes. Keep a detailed log of which server parameters you changed and the reason for the change. This information will be useful if a problem arises because of the change.
Managing parameters EMC recommends using the server_param command or the Celerra Manager to manage parameters. If you manually edit the param file, note the following: Parameters are case-sensitive and must be entered exactly as described in the Celerra Network Server Parameters Guide. A typographical error, the omission of a dot, incorrect capitalization, or a value out of range could result in adverse effects.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
55
Technical notes
Storage
Multiple LUN 0s Assigning multiple LUN 0s will result in Data Movers being unable to boot successfully. Powering down all SAN Switches The Celerra does not store any user data itself. All data is stored on the Symmetrix or CLARiiON disk array(s) that the Celerra is attached to. With a Celerra gateway model the Celerra is attached to the disk array(s) it uses via a Fibre Channel SAN. If that SAN is unavailable then the Celerra will not be able to access the disk array(s) it relies on and hence will not be able to offer normal service. Reassigning volumes with SDDR or making fibre changes Like any host of a Fibre Channel SAN or array, the Celerra will remember the address of the LUNs that it uses in the attached disk arrays. Any change in the identity or location of those LUNs without a corresponding reconfiguration of the Celerra may result in the Celerra being unable to access the LUNs it needs to provide normal service. Reassigning Celerra LUN with Navi Same as Reassigning volumes with SDDR or making Fibre changes. Unbinding/rebinding LUNs Same as Reassigning volumes with SDDR or making Fibre changes above. Write disabling R1 and R2 In SRDF configurations, setting R2 disks to "read only" will cause R1 disks to become "read only" as well, and cause access issues.
Technical notes
Configuration guidelines
The values listed in this section represent the highest or lowest values tested by EMC. In some cases, actual capacities may exceed the capacities tested by EMC. Theoretical means in theory, not always practical. Refer to the E-Lab Interoperability Navigator for the most up-to-date information regarding these guidelines.
56
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Table 4 CIFS Guidelines Guideline/specification CIFS TCP connection Maximum tested value 64K (default and theoretical max.), 20K (max.tested) Comment Param tcp.maxStreams sets the maximum number of TCP connections a Data Mover can have. Maximum value is 64K, or 65535. TCP connections (streams) are shared by other components and should be changed in monitored small increments. Unicode: The maximum length for a share name with Unicode enabled is 80 characters. ASCII: In ASCII mode, the maximum share length is 12 characters. You may notice a performance impact when managing a large number of shares. Limited by the number of network interfaces available on the Data Mover. From a local group perspective, the number is limited to 509. NetBIOS and compnames must be associated with at least one unique network interface. NetBIOS names are limited to 15 characters (Microsoft limit) and cannot begin with an@ (at sign) or a - (dash) character. The name also cannot include white space, tab characters, or the following symbols: / \ : ; , = * + | [ ] ? < > “. If using compnames, the NetBIOS form of the name is assigned automatically and is derived from the first 15 characters of the <comp_name>.
Share name length
80 characters (Unicode). 12 chars (ASCII)
Number of CIFS shares
10,000 per Data Mover 40,000 per Celerra 509 (max.)
Number of NetBIOS names/compnames per Data Mover
NetBIOS name length
15
EMC Celerra Network Server Version 6.0.70.4 Release Notes
57
Technical notes
Comment (ASCII chars) for NetBIOS name for server
256
Limited to 256 ASCII characters. Restricted Characters: You cannot use double quotation ("), semi-colon (;), accent (`), and comma (,) characters within the body of a comment. Attempting to use these special characters results in an error message. You can only use an exclamation point (!) if it is preceded by a single quotation mark (’). Default Comments: If you do not explicitly add a comment, the system adds a default comment of the form EMC-SNAS:T<x.x.x.x> where <x.x.x.x> is the version of the NAS software. For integration with Windows environment releases greater than Windows 2000, the CIFS server computer name length can be up to 21 characters when UTF-8 (3 bytes char) is used. The maximum number of Windows domains a Data Mover can be a member of. To increase the max default value from 32, change param cifs.lsarpc.maxDomain. The Parameters Guide contains more detailed information about this parameter. Maximum buffer size that can be negotiated with Microsoft Windows clients. To increase the default value, change param cifs.W95BufSz, cifs.NTBufSz or cifs.W2KBufSz. The Parameters Guide contains more detailed information about these parameters. This value is fixed and defines the number of requests a client is able to send to the Data Mover at the same time, (for example, a change notification request). To increase this value, change the maxMpxCount parameter. The Parameters Guide contains more detailed information about this parameter.
Compname length
63 bytes
Number of domains
10 tested 509 (theoretical max.)
Block size negotiated
64 KB
Number of simultaneous requests per CIFS session (maxMpxCount)
127
58
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Total number files/directories opened per Data Mover Number of Home Directories supported Number of Windows/UNIX users
200,000 (NS/500/700, DM514, DM 514T,NSX, NS40, NS80) 20,000 unlimited
A large number of open files could require high memory usage on the Data Mover and potentially lead to out-ofmemory issues.
Earlier versions of the Celerra Network Server relied on a basic database, nameDB, to maintain Usermapper and secmap mapping information. DBMS now replaces the basic database. This solves the inode consumption issue and provides better consistency and recoverability with the support of database transactions. It also provides better atomicity, isolation, and durability in database management. To increase the default value, change param cifs.listBlocks (default 64, max 255). The value of this parameter times 256 = max number of users. Note: TID/FID/UID shares this param and cannot be changed individually for each ID. Use caution when increasing this value as it could lead to an out-ofmemory condition. Refer to the System Parameters Guide for parameter information Number of threads for CIFS users (if there is 1 GB of memory on the Data Mover, the default is 96 threads; however, if there is over 1 GB of memory, the default number of threads is 256). This is the default, you can increase the number using the server_setup command with cifs start=xx value.
Number of users per TCP connection
16K (default), 64K (max)
Number of CIFS threads
96 (1GB memory) 256 (>1 GB memory)
EMC Celerra Network Server Version 6.0.70.4 Release Notes
59
Technical notes
Number of CIFS connections (share accessed) per TCP connection
16K (default), 64K (max)
To increase the default value, change param cifs.listBlocks (default 64, max 255). The value of this parameter times 256 = max number of connection. Note: TID/FID/UID shares this param and cannot be changed individually for each ID. Use caution when increasing this value as it could lead to an out-ofmemory condition. Refer to the System Parameters Guide for parameter information. To increase the default value, change param cifs.listBlocks (default 64, max 255). The value of this parameter times 256 = max number of files/directories opened per CIFS connection. Note: TID/FID/UID shares this param and cannot be changed individually for each ID. Use caution when increasing this value as it could lead to an out-ofmemory condition. Be sure to follow the recommendation for total number of files/directories opened per Data Mover. Refer to the System Parameters Guide for parameter information. Note: The total number of VDMs, file systems, and checkpoints across a whole cabinet cannot exceed 2048.
Number of files/directories opened per CIFS connection
16K (default), 64K (max)
Number of VDMs per Data Mover number of ntxmap User Mapping entries Table 5 FileMover Guideline/specification Max connections to secondary storage per primary (Celerra) file system Number of HTTP threads for servicing FileMover API requests per Data Mover
29
1000
Maximum tested value 1024
Comment
64
Number of threads available for recalling data from secondary storage is half the number of whichever is the lower of CIFS or NFS threads. 16 (default), can be increased using server_http command, max (tested) is 64.
60
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Table 6 File systems guidelines Guideline/specification Maximum tested value Mount point name length 255 bytes (ASCII)
Comment The "/" is used when creating the mount point and is equal to one character. If exceeded, Error 4105: Server_x:path_name: invalid path specified is returned. For nas_fs list, the name of a file system will be truncated if it is more than 19 characters. To display the full file system name, use the –info option with a file system ID (nas_fs -i id=<fsid>). With Unicode enabled in an NFS environment, the number of characters that can be stored depends on the client encoding type such as latin-1. For example: With Unicode enabled, a Japanese UTF-8 character may require three bytes.1 With Unicode enabled in a CIFS environment, the maximum number of characters is 255. For filenames shared between NFS and CIFS, CIFS allows 255 characters. NFS truncates these names when they are more than 255 bytes in UTF-8, and manages the file successfully. Note: Make sure the final path length of restored files is less than 1024 bytes. For example, if a file is backed up which originally had path name of 900 bytes, and it is restoring to a path with 400 bytes, the final path length would be 1300 bytes and would not be restored.1 This is a hard limit and is rejected on creation if over 255 limit. The limit is bytes for UNIX names, Unicode characters for CIFS. 1 This is a hard limit, code will prevent you from creating more than 65,533 directories. This max number includes VDM and checkpoint file systems.
File system name length
240 bytes (ASCII) 19 chars display for list option
Filename length
255 bytes (NFS), 255 characters (CIFS)
Pathname length
1,024 bytes
Directory name length
255 bytes.
Subdirectories (per parent directory) Number of file systems per Celerra
65,533
4096 (NSX, NS40/NS80) 2048 (All other models)
EMC Celerra Network Server Version 6.0.70.4 Release Notes
61
Technical notes
Number of file systems per Data Mover
2048
Mount operation will fail when the number of file systems reaches 2048 with an error indicating maximum number of file systems reached. This max number includes VDM and checkpoint file systems. Unified platforms — Running setup_clariion on the NS-960 platform will provision the storage to use 1 LUN per RAID group which may result in LUNS > 2 TB, depending on drive capacity and number of drives in the RAID group. On all CX4 integrated platforms (NS-120, NS480, NS-960), the 2 TB LUN limitation has been lifted--this may or may not result in LUN size larger than 2 TB depending on RAID group size and drive capacity.For all other unified platforms, setup_clariion will continue to function as in the past, breaking large RAID groups into LUNS that are < 2 TB in size. Gateway systems — For gateway systems, users may configure LUNs greater than 2 TB up to 16 TB or max size of the RAID group, which ever is less. This is supported for NS-40G, NS80G, NSX, and NS-G8 NAS gateways when attached to CX3, CX4 and Symmetrix DMX-4 backends. Multi-Path File Systems (MPFS) — MPFS supports LUNs greater than 2 TB. Windows 2000 and 32-bit Windows XP, however, cannot support large LUNs due to a Microsoft Windows OS limitation. All other MPFS Windows clients support LUN sizes greater than 2 TB if the 5.0.90.900 patch is applied. Use of these 32-bit Windows XP clients on NS-960 systems require RPQ (Request for Price Quotation) approval.
Maximum Disk Volume Size
dependent on RAID group (see comments)
62
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Total storage for a Data Mover (Fibre Channel Only)
Celerra Version 6.0 16 TB = NX4 32 TB =NS20, NS-120 64 TB =NS40, NS480, NS40G, NS-G2 128 TB = NS80, NS80G, NSX 256 TB = NS960, NS-G8 Celerra Version 5.6 8 TB = CNS (514 DM) 10 TB = NX350, NS500 16 TB = NX4, NS700 32 TB = NS20, NS120 64 TB = NS40, NS480, NS40G, NS-G2, VG2 128 TB = NS80, NS960, NS80G, NSX, NS-G8 See right column for TB information per Data Mover
These total capacity values represent Fibre Channel disk maximum with no ATA drives. Fibre Channel capacity will change if ATA is used. Note: On a per-Data-Mover basis, the total size of all file systems, and the size of all SavVols used by SnapSure must be less than the total supported capacity. Exceeding these limit can cause out of memory panic. Refer to the Celerra Capacity Limits tables for more information, including mixed disk type configurations.
File system size
16 TB (hard limit, enforced) NS700, NSX, NX4, NS120, NS480, NS960, NS-G8 10 TB NS350, NS500 8 TB 514/514T Data Mover When planning a file system, with respect to maximum desired size, be sure to factor in the time required for FSCK, ACL checks, file system build & restore. This guideline applies to both Fibre Channel and ATA drives. This hard limit is enforced and cannot be exceeded. Same as number of inodes in file system. Each 8 KB of space = 1 inode.
File size Number of directories supported per file system Number of files per Directory
16 TB (4 TB with file size policy of quota)
500,000
Exceeding this number will cause performance problems.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
63
Technical notes
Maximum number of files and directories per Celerra file system
256 Million (default)
This is the total number of files and directories that can be in a single Celerra file system. This number can be increased to 4 billion at file system creation time but should only be done after considering recovery and restore time and the total storage utilized per file. The actual maximum number in a given file system is dependent on a number of factors including the size of the file system.
1 All other industry-standard caveats, restrictions, policies, and best practices prevail. This includes, but is not limited to, FSCK times (now made faster through multi-threading), backup and restore times, number of objects per file system, snapshots, file system replication, VDM replication, iSCSI LUN replication, performance, availability, extend times, and layout policies. Proper planning and preparation should occur prior to implementing these guidelines.
Table 7 ISCSI guidelines Guideline/specification Maximum tested value LUN size Minimum: 3 MB Maximum: 2 TB less 1 MB
Comment Because an iSCSI LUN requires overhead, you cannot create a LUN size equal to the total available space; the maximum size for a LUN is slightly less (typically 1 MB less) than the amount of available space in the file system. Each target can have as many as 256 LUNs (numbered 0 through 255). Windows, however, limits each target to 255 LUNs (numbered 0 through 254). To comply with the Windows limitation, Windows - compatible mode, which is enabled by default on the Data Mover, limits the number of LUNs to 255. If you use a non-Windows iSCSI initiator, you can disable Windows--compatible mode by modifying the iscsi.WinCompat parameter. Refer to the System Parameters Guide for parameter information. Due to a volume formatting limitation on Microsoft platforms, when extending a LUN, the total LUN size after extension must be greater than 129 MB.
LUNs per target
256
LUN extension size
Minimum: 8 MB Maximum: same as LUN size
64
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Target alias
255 bytes
This name is an alias for the formal iSCSI name. The alias is an alphanumeric string of as many as 255 characters. The alias can include Unicode characters in UTF-8 encoding. Because the alias is the key field in the Celerra Network Server’s iSCSI databases, each alias must be unique on a given Data Mover.1 EMC strongly recommends you leave this field empty and let the Celerra Network Server generate a legal iSCSI name. If you type a name, be sure it is globally unique and in either IQN or EUI format. An IQN can have as many as 223 characters. The Celerra Network Server converts all uppercase characters in an IQN or EUI to lowercase characters. Only letters, numbers, "-", ".", and ":" are allowed in iSCSI IQN names. CHAP secrets must be a minimum of 12 characters and a maximum of 255 characters. Microsoft iSCSI initiator supports CHAP secrets of 12 to 16 characters. Maximum includes temporary writable snapshots.
iSCSI qualified name (IQN)
223 characters
CHAP secret
255 characters
Snapshots per LUN Replication session alias Replication application label Replication attached comment text or binary data Configured replication sessions per Data Mover Concurrent active replication sessions per Data Mover Destinations per source file system
2000 64 characters 128 characters 4096 bytes
1024
256 (theoretical max)
Number of configured iSCSI replication sessions that can be actively transferring data at one time. Exceeding this number may cause performance implications on the source file system.
4
EMC Celerra Network Server Version 6.0.70.4 Release Notes
65
Technical notes
iSCSI timeout
See comments
EMC recommends client iSCSI initiator timeout be longer than failover time. Note: MS iSCSI Initiator has a default timeout of 1 minute.
Table 8 Naming Services guidelines Guideline/specification Maximum tested value Number of DNS domains 3 - WebUI unlimited - CLI
Comment Three DNS servers per Data Mover is the limit if using WebUI. There is no limit when using the CLI (command line interface). You can configure up to 10 NIS servers in a single NIS domain on a Data Mover. Note: A Data Mover supports only one NIS domain. Each time you configure a NIS domain and specify the servers, it overwrites the previous configuration. A Data Mover can read 1004 bytes of data from a NIS record.
Number of NIS Servers per Data Mover
10
NIS record capacity Number of DNS servers per DNS domain
1004 bytes 3
Table 9 NFS guidelines Guideline/specification Maximum tested value Number of NFS exports 2,048 per Data Mover Tested Unlimited theoretical max. 64K with TCP (theoretical) Unlimited with UDP (theoretical) 16383
Comment You may notice a performance impact when managing a large number of exports using Celerra Manager.
Number of concurrent NFS clients
Limited by TCP connections.
Netgroup line size
The maximum line length that the Data Mover will accept in the local netgroup file on the Data Mover or the netgroup map in the NIS domain that the Data Mover is bound to. 2 billion max value of any GID. The maximum number of GIDs is 64K, but an individual GID can have an ID in the range of 0- 2147483648
Number of UNIX groups supported
64K
66
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Table 10 Networking guidelines Guideline/specification Maximum tested value Link aggregation/ether channel 8 ports (ether channel) 12 ports (link aggregation (LACP)
Comment Ether channel: the number of ports used must be a power of 2 (2, 4, or 8). Link aggregation: any number of ports can be used. All ports must be the same speed. Mixing different NIC types (i.e., copper and fibre) is not recommended. IEEE standard. Theoretically 509. By default the value is (in theory) 0xFFFF, but it is also limited by the number of TCP streams that can be opened. To increase the default value, change param tcp.maxStreams (set to 0x00000800 by default). If you increase it to 64K before you start TCP, you will not be able to increase the number of FTP connections. Refer to the System Parameters Guide for parameter information.
Number of VLANs supported Number of interfaces per Data Mover Number of FTP connections
4094 45 tested Theoretical value 64K
Table 11 Quotas guidelines Guideline/specification Maximum tested value Number of Tree Quotas Max size of Tree Quotas Max number of unique groups Quota path length 8191 256 TB 64K 1024
Comment Per file system Per file system
EMC Celerra Network Server Version 6.0.70.4 Release Notes
67
Technical notes
Table 12 Replicator V2 guidelines Guideline/specification Maximum tested value Number of Replication sessions per Celerra Max number of Replication sessions per Data Mover Max number of local and remote file system and VDM replication sessions per Data Mover Max number of loopback file system and VDM replication sessions per Data Mover 1365 (NSX) (other platforms): 682 1024
Comment
This enforced limit includes all configured file system, VDM, and iSCSI replication sessions as well as copy sessions
682
341
The max number of iSCSI LUN replication sessions per Data Mover is the difference between the max number of replication sessions (1024) minus the number of configured file systems and VDM replication sessions per Data Mover. Table 13 SnapSure guidelines Guideline/specification Maximum tested value Number of checkpoints per file system 96 read-only 16 writable
Comment Up to 96 checkpoints per file system are supported, as well as 16 writable checkpoints.
Table 14 Virus Checking guidelines Guideline/specification Maximum tested value Virus checker file pending 90% of available VNODES in system
Comment The percentage of total cached open files available in the system that can be pending on Virus Checking. An event is sent to the Control Station when the maximum is reached. To change the default value, modify param viruschk.vnodeMax(default=2000) and/or param viruschk.vnodeHWM(default=0x5a(90% ). Refer to the System Parameters Guide for parameter information.
On systems with Unicode enabled, a character may require between 1 and 3 bytes, depending on encoding type or character used. For example,
68
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
a Japanese character typically uses 3 bytes in UTF-8. ASCII characters require 1 byte.
Capacity guidelines Celerra capacity limits - NS series
Following is a list of the versions 6.0 Celerra Capacity limits for the NS series. The E-Lab Interoperability Navigator is updated monthly and contains the most up-to-date capacity limits.
Table 15 Celerra capacity limits – NS series
NX4
NS20
NS120
NS40
NS480
NS80 (XB-60 or XB-65)
4
NS960
Usable IP Storage Capacity limit per blade (all disk 5 types and uses) Usable MPFS Storage Capacity limit per blade (all disk types and 5 uses) Max FS Size Max # FS per 7 DM/Blade Max # FS per 7 cabinet Max configured replication sessions per DM/Blade for 3 Replicator v2 Max # of checkpoints per 2 PFS Max # of NDMP sessions per DM/Blade
16TB
4
32TB
4,6
32TB
6
64TB
4
64TB
128TB
256TB
N/A
128TB
1
128TB
1
128TB
1
128TB
1
128TB
1
256TB
1
16TB 2048 4096
16TB 2048 4096
16TB 2048 4096
16TB 2048 4096
16TB 2048 4096
16TB 2048 4096
16TB 2048 4096
1024
1024
1024
1024
1024
1024
1024
96
96
96
96
96
96
96
4
4
4
4
4
4
4
EMC Celerra Network Server Version 6.0.70.4 Release Notes
69
Technical notes
Table 16 Celerra capacity limits – NS series (continued)
NS40 G
NS-G2
NS80 (XB-60 or XB-65)
4
NSX (XB-60 or XB-65)
4
NS-G8
V-G2
VG8
Usable IP Storage Capacity limit per blade (all disk 5 types and uses) Usable MPFS Storage Capacity limit per blade (all disk types and 5 uses) Max FS Size
64TB
4
64TB
4
32TB
128TB
128TB
64TB
256T B
128TB
1
128TB
1
128TB
1
128TB
1
128TB
128TB
256T B
16TB
16TB
16TB
16TB
16TB
16TB
16T B 2048 4096
Max # FS per 7 DM/Blade Max # FS per 7 cabinet Max configured replication sessions per DM/Blade for 3 Replicator v2 Max # of checkpoints per 2 PFS Max # of NDMP sessions per DM/Blade
2048 4096
2048 4096
2048 4096
2048 4096
2048 4096
2048 4096
1024
1024
1024
1024
1024
1024
1024
96
96
96
96
96
96
96
4
4
4
4
4
4
4
Notes: 1. 2. 3. No snaps or replication –If planning to use snaps or replication, the Data Mover capacity limit will be the same as non-MPFS. PFS is Production File System. A maximum of 256 concurrently transferring sessions are supported per DM.
70
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
4. 5.
When using Replicator –To achieve storage capacities listed, Replicator v2 is required. This is the usable IP storage capacity per blade. For overall platform capacity, consider the type and size of disk drive and the usable Fibre Channel host capacity requirements, RAID group options and the total number of disks supported. NS20, NS120 – All ATA configuration can go up to 48 TB per blade. This count includes production file systems, user defined file system checkpoints, and two checkpoints for each replication session.
6. 7.
Celerra virus checking solution
The following table lists the requirements for using CAVA with thirdparty antivirus engines.
AV Vendor Celerra Version CAVA Version
CA
eTrust r8.1 eTrust r8 TrendMicro ServerProtect for EMC NAS 5.3.1 ServerProtect for EMC NAS 5.5.8 Sophos AntiVirus 3.x All currently-supported DART versions1 1.8.9; 2.2.4; 3.4.0; 3.5.6; 3.5.7.1, 3.5.8, 3.6.2 All currently-supported DART versions1 1.8.9; 2.2.4; 3.4.0; 3.5.6; 3.5.7.1, 3.5.8, 3.6.2 All currently-supported DART versions1 1.8.9; 2.2.4; 3.4.0; 3.5.6; 3.5.7.1, 3.5.8, 3.6.2
AntiVirus 5.x
AntiVirus 6.x AntiVirus 7.x AntiVirus 9.x Kaspersky Kaspersky Anti-Virus for Windows Servers Enterprise Edition McAfee VirusScan 8.0i All currently-supported 1.8.9; 2.2.4; 3.4.0; 3.5.6; 3.5.7.1, 3.5.8, 3.6.2 5.6.46.4 and later Celerra Event Enabler (CEE) version 4.5.1
EMC Celerra Network Server Version 6.0.70.4 Release Notes
71
Technical notes
VirusScan 8.5i Patch 2 and above Symantec Symantec Endpoint Protection 11.04 SAV for NAS 5.1.x SAV for NAS 4.3.x
DART versions1
5.6.48.x and later All currently-supported DART versions1
4.5.2.2 and later 3.6.2 3.5.7.1, 3.5.8, 3.6.2
1Currently supported NAS versions include the entire 5.6.x and 6.0.x TM families. Refer to the EMC E-Lab Interoperability Navigator for the latest supported version information.
Celerra scripting guidelines
Following are guidelines to keep in mind when creating Celerra command scripts.
Issue Scheduling Action NAS Database Backups: The Celerra Network Server backs up the NAS database that stores specific configuration information required for each Data Mover every hour, at one minute after the hour. During part of the backup, the database is locked, and some commands that rely on the database might not have access. EMC recommends that command scripts avoid starting at one minute after the hour. Note that scripts with complex commands that run for an extended period may overlap the backup period. The duration of the backup may vary. Use the following Linux command to check the state of the backup process prior to executing scripts: ps -ef | grep nasdb_backup. If a lock condition occurs, wait a few minutes and retry. Some commands must lock the database in order to execute. If multiple user-entered commands or scripts are active at the same time, some of these commands may lock the database and prevent other commands from executing. To avoid this, you should sequence commands whenever possible. Some processes within a script can take time to complete. Use proper timing and adequate sleep statements to prevent timingrelated issues. Piping script outputs through grep is a helpful tool to check the status of the script. Use periodic checks to grep for file or database locked messages, timeouts, resource unavailable warnings, and other failure or success messages, and use this information to check status, pause the script, or halt it. All commands return a UNIX-style return code (for example: 0 for success or 1 for failure) or a text-based status code (for
Command sequencing
Sleep statements
Pipe and grep
Return code check
72
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Issue
Action example, done) that can be used to help determine if the command completed or if there was an error or a conflict with the NAS database backup, or other commands being run. If a lock condition occurs, wait a few minutes and retry. If you create and run scripts, be sure to incorporate return code checks and verify for proper return codes from individual operations.
Scripting examples: The RECOVERABLE variable contains the following errors to retry on: Unable to acquire lock Resource temporarily unavailable Unable to connect to host Socket: All ports in use Database resource vanished Connection timed out NAS_DB locked object is stale
RECOVERABLE="unable to acquire lock|Resource temporarily unavailable|unable to connect to host|socket: All ports in use|database resource vanished|Connection timed out|NAS_DB locked object is stale".
Following is an example of what the RECOVERABLE variable looks like:
The res variable contains the command output.
#!/bin/sh ######################################################## # File: nas_cmdrcvr.sh # Created by: NAS Engineering # Date: Thursday, May 25, 2006 # Version: 1.0 # Notes: # 1) script will retry commands for specified period of time #2) script will log messages to file only if there's available disk space ######################################################## NAS_DB=/nas export NAS_DB PATH=$PATH:$NAS_DB/bin:$NAS_DB/sbin:/usr/sbin:.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
73
Technical notes
export PATH RETRIES=60 SLEEPTIME=30 LOGDIR="/home/nasadmin" LOGFILE="$0.log" LOGLAST="$0.last" DISKFULL=98 RECOVERABLE="Resource temporarily unavailable|\ unable to acquire lock|\ unable to connect to host|\ socket: All ports in use|\ database resource vanished|\ Connection timed out|\ NAS_DB locked object is stale" # # function to log messages to a file # nas_log() { DISKCHK=`df -k $LOGDIR|awk 'NR>1{print $5}'|sed 's/\%//'` # if there's enough free disk space, append to log if [ $DISKCHK -lt $DISKFULL ]; then TDSTAMP=`date +%Y-%m-%d' '%T` echo $TDSTAMP: $LOGMSG >> $LOGDIR/$LOGFILE fi # regardless of available space, always write last error echo $TDSTAMP: $LOGMSG > $LOGDIR/$LOGLAST } # # function to execute (and potentially retry) commands # nas_cmd() { # initialize variable(s) retry_count=0
EMC Celerra Network Server Version 6.0.70.4 Release Notes
74
Technical notes
# loop until either successful of retry count exceeded while [ $retry_count -le $RETRIES ]; do # execute command and gather response RES=`$CMD 2>&1` # check if response means command is recoverable if [ `echo "$RES"|egrep -c "$RECOVERABLE"` -ne 0 ]; then # check retry count if [ $retry_count -ne $RETRIES ]; then # retry count has not been exceeded LOGMSG="Command ($CMD) failed with ($RES)...retrying in $SLEEPTIME s" nas_log sleep $SLEEPTIME else # retry count has been exceeded LOGMSG="Command ($CMD) failed with ($RES)...exiting (retry count of $RETRIES exceeded)" nas_log exit 1 fi else # command was either successful or failed for an unknown reason LOGMSG="Command ($CMD) successful with ($RES)" nas_log retry_count=$RETRIES exit 0 fi #increment counter for retries retry_count=`expr $retry_count + 1` done } # # main
EMC Celerra Network Server Version 6.0.70.4 Release Notes
75
Technical notes
# CMD="nas_volume -d mtv1" nas_cmd
Using the NAS database and query facility EMC has partially changed the layout or format of Celerra internal databases. This change can impact the use of awk or grep utilities when used in scripts that assume specific positions of fields in databases. To enable searching of the NAS database, Celerra has developed a new query subsystem that appears as a hidden option on some of the nas_commands. This query subsystem enables you to specify the information you are interested in, allows you to format the output, and is independent of the database format. Caution Do not use grep and awk to scan the database files. Database positions may change and substrings may return false matches for database objects. Following is an example of a query to view unused disks: nas_disk -query:inuse==n -format:'%s\n' -Fields:Id To filter root disks, refer to the List all non-root disks that are not in use in Table 10. Examples Use the following commands to view the tags (fields) that you can query: nas_disk –query:tags nas_fs –query:tags nas_volume –query:tags nas_slice -query:tags Table 17 contains a list of examples to help you get started. Note that these commands can be run on the Control Station CLI, so the hardcoded values can be replaced with shell script variables.
Table 17 Task and query examples Task Query the ID of a named file system Example
nas_fs -query:Name==RLL_fs10 format:'%s\n'-Fields:Id
76
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Technical notes
Query the ID of a named file system without the new line Query the name of a file system ID that corresponds to a particular ID List of all server ID List of all server names List of all the checkpoint file systems List type of file system with ID 20
nas_fs -query:Name==RLL_fs10 format:'%s'-Fields:Id nas_fs -query:id==20 -format:'%s\n' -Fields:Name nas_server -query:* -format:'%s\n' Fields:Id nas_server -query:* -format:'%s\n' Fields:Name nas_fs -query:type==ckpt fields:name -format:"%s\n" nas_fs -query:id==20 -format:'%s\n' -Fields:Type nas_fs -query:inuse==y format:'%s\n' -Fields:Name or nas_fs -query:inuse==y format:'%s\n' -Fields:Id nas_fs -query:id==28 -format:'%s' Fields:BackupOf nas_server -query:id==2 format:'%s\n' -fields:name`" nas_fs -query:Name==my_fs -format:%d -fields:VolumeID nas_volume -query:Name==my_meta3 format:%d -fields:Blocks nas_volume -query:Name==JAH_meta3 format:%d -fields:BlockSize nas_fs -query:name==fs123 format:'%s\n'-fields:ServersNumeric nas_disk query:inuse==n:IsRoot==False -format:"%s\n" -fields:name nas_volume query:inuse==n:IsRoot==False:name=dc -format:"%s\n" -fields:name nas_disk -query:inuse==n:SymmID==$ symm_id:IsRoot==False -format:"%s\n" -fields:name
List the file systems that are in use
Identify file system of which file system ID 28 is a backup List the name of the server with ID 2
View which volume file system is built on
View the block count of meta volume
View the block size of meta volume
Find which server IDs use fs123
List all non-root disks that are not in use
List unused volumes that contain “dc” in the volume name List all available disks on a particular storage device (symm_id is a script/env variable)
EMC Celerra Network Server Version 6.0.70.4 Release Notes
77
Environment and system requirements
Query operators Table 18 lists the operators to use when building your queries:
Table 18 Operator definitions
Operator = == ==+ =* =^ =^= =^=^+ =^*
Definitions
Having the pattern (sub-string match) Exact string match Integer Minimum (not less than) Integer Maximum (not more than) Any NOT having the pattern NOT an exact match NOT Integer Minimum (is less than) NOT Integer Maximum (is more than) NOT Any (None)
Environment and system requirements
Symmetrix microcode and CLARiiON FLARE levels
The E-Lab Interoperability Navigator contains the latest microcode and FLARE information. The E-Lab Interoperability Navigator is updated monthly and can be found on EMC Online Support at http://Support.EMC.com.
Centera FileArchiver not supported
The Centera® FileArchiver policy engine is not supported when using Celerra Network Server version 6.0.36.4 or later.
Java Runtime Environment (JRE)
Minimum JRE 1.6. After February 2013, Oracle will no longer post updates of Java SE 6 to its public download sites. Existing Java SE 6 downloads already posted as of February 2013 will remain accessible in the Java Archive on Oracle Technology Network. This Unisphere Client will run with
78
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Documentation
JRE 1.7 but it has not yet been fully qualified with JRE 1.7. Future releases of Unisphere will be qualified with JRE 1.7 and the future Unisphere Client can be used as a workaround if there are issues specific to JRE 1.7.
Documentation
Celerra Network Server Version 6.0 documentation
EMC provides the ability to create step-by-step planning, installation, and maintenance instructions tailored to your environment. To create VNX customized documentation, go to: https://mydocs.emc.com/VNX.
Documentation updates
CDMS The Celerra Data Migration Service documentation leads the user through a multi-step process to execute a migration. The latest version of this document highlights the success and failure criteria for various steps in this process. Celerra CDMS Version 2.0 for NFS and CIFS provides more information about CDMS. The following have also been added as warnings and cautions: A verify command should not be run if there are any migration threads running or active on other CDMS connections in the same file system. A verify command should not be run if there is another verify command still in progress on a different file system on the same Data Mover.
Multibyte support for Celerra data elements In the Celerra management interfaces, multibyte support allows to input, store, and display multibyte characters. Users can input and view Celerra object fields in their native language. The multibyte feature is supported since Celerra NAS 5.6 and later. The following data elements are enabled with multibyte support and also support the Unicode 3.0 standard: CIFS Share name
EMC Celerra Network Server Version 6.0.70.4 Release Notes
79
Documentation
CIFS share comment CIFS export path Tree quota pass Tree quota comment Quota user name CIFS server netbios name CIFS server computer name CIFS server alias CIFS server comment CIFS domain name CIFS workgroup Organizational Unit (OU) name DFS root name CDMS source share name CDMS local path CDMS source export path Mount point name File system name Checkpoint name NFS export name NFS export path DFS root name CDMS source share name CDMS local path CDMS source export path Mount point name File system name Checkpoint name NFS export name NFS export path
80
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Installation and upgrades
Installation and upgrades
Celerra installation process
This section lists some facts you should know about the Celerra Network Server and the installation and configuration process before proceeding further. How to prepare for Celerra installation To prepare for Celerra installation and configuration, complete the following tasks: 1. Perform the pre-site installation tasks as required by EMC to complete the installation and connectivity of the Celerra Network Server. Tasks include preparing electrical facilities, wiring the network, adding a telephone line, checking environmental conditions and related requirements, as stated in the Professional Services Statement of Work (PSSOW).
Note: For a copy of the PS-SOW document, contact your Professional Services Account Team. For examples of Celerra network topologies, refer to the Configuring and Managing Celerra Networking technical module.
2.
Configure your environment for UNIX clients, Microsoft Windows clients, or a combination of Windows and UNIX clients, as applicable. Complete network assessment to ensure proper performance and protocol configurations. (Optional EMC services can be ordered to assist with this process.)
3.
Celerra Network Server installation and configuration EMC Customer Service installs the Celerra Network Server hardware—a process that includes the physical setup, cabling, power application and diagnostic testing of the Celerra Network Server. Next, EMC Professional Services or an EMC partner might configure and integrate the Celerra Network Server. For example, for a Celerra Network Server with a Symmetrix system, this process typically includes the following tasks: Coordinating the activities to enable installation, cabling, and connectivity to system cabinets Loading and configuring the Celerra Network Server software
EMC Celerra Network Server Version 6.0.70.4 Release Notes 81
Installation and upgrades
Creating, mounting, and exporting file systems Configuring network protocols Integrating with a supported, existing user authentication environment Configuring standby Data Mover(s), if applicable Configuring and demonstrating system management and administration tools
EMC Professional Services can perform additional Celerra Network Server configuration tasks, such as host provisioning (with Celerra MPFS), backup protection, and network connectivity, depending on the individual EMC customer agreement defined in the PSSOW. These services are part of the EMC Network-Attached Storage Design and Implementation service offering, which is designed to provide continuous technical expertise as you manage your organization’s growing network storage needs. For more information about this service offering, contact your EMC Customer Support representative. After EMC installs, configures, and tests the Celerra Network Server, it is fully operational and ready to support networked storage operations.
Celerra upgrades
You can upgrade the Celerra NAS software two ways: Celerra Upgrade Tool Command Line Interface
Celerra Upgrade Tool You can upgrade the software for many Celerra integrated systems by using the Celerra Upgrade Tool (CUT), which is part of the Unisphere Service Manager (USM). CUT lets you upgrade Celerra systems without assistance from EMC or Customer Support by communicating over the Internet to EMC Online Support, and over a local network to a Celerra system. Three wizards guide you, step-by-step, through the process of upgrading the Celerra software, including updating the Control Station and the blades, but not the storage system: Go to EMC Online Support at http://Support.EMC.com for information about USM and CUT and for information about downloading a copy of USM to a Windows system. The Upgrading Unisphere for File online help on USM provides information and instructions about using CUT and its wizards. The Celerra Upgrade Tool can upgrade Celerra Network Servers running software versions:
82 EMC Celerra Network Server Version 6.0.70.4 Release Notes
Installation and upgrades
• •
5.6 to 6.0 6.0 to 6.0.x or later
The Celerra Upgrade Tool cannot upgrade systems with dual Control Stations, nor can it be used to upgrade Gateway systems. If your Celerra is not eligible for upgrade by means of CUT, you will need to use the CLI to upgrade your system. Running multiple upgrade wizards The USM Celerra Upgrade Tool wizard does not explicitly prevent CLARiiON upgrade wizards from running while it is in use. The Celerra Upgrade Tool presents a prominent warning on its 'Welcome' page in the GUI. Note that this scenario is the same as simultaneously running multiple instances of USM and/or CLI against the systems, which cannot currently be prevented. Also, the CLARiiON wizards correctly prohibit the Celerra Upgrade Tool from launching while they are in use. In general, upgrading the Celerra NAS and the CLARiiON FLARE software at the same time can cause serious system problems and result in system downtime. Other CUT restrictions: Selecting and rebooting blades at different times can cause errors. Select and reboot all blades at same time during the upgrade in the Reboot Blades dialog. The Celerra Upgrade Tool cannot upgrade systems with dual Control Stations.
NAS and FLARE upgrades EMC requires NAS upgrades to be performed before FLARE is upgraded on any attached CLARiiON. The reason for this is to maintain compatibility between the storage API (NAVI) on the Control Station and the FLARE on the CLARiiON. The API is built in as part of the NAS management software and is used to gather data about the backend such as statistics, health, and configuration. If the FLARE is upgraded first, the API may encounter issues while communicating with the new FLARE. Upgrading the NAS first allows the API to have limited backward compatibility with the FLARE code. Out-of-family upgrades The following issues have been reported during out-of-family upgrades.
EMC Celerra Network Server Version 6.0.70.4 Release Notes 83
Installation and upgrades
Error message during upgrade While upgrading from version 5.6 to 6.0, the following error message is displayed: Error - cannot open /usr/sbin/ext2online /dev/hda1. This error can be ignored, as it does not stop the upgrade and the task completes successfully. Pre-Upgrade Health Check (PUHC) fails When attempting to upgrade the software via CLI from version 5.5 to 6.0, PUHC fails abruptly with a run-time failure message which does not explain the cause of the failure. Note: A 5.5 to 6.0 upgrade is not supported. Upgrade Replication task While performing an out-of-family upgrade (5.6.x to 6.0.x) on a dual Control Station system, the Upgrade Replication task exits with the message: =======================Tasks======================= 23:06 [ 3/80 ] Upgrade replication 0 seconds Please shutdown control station in slot 1 and retry When this occurs, users need to power off the secondary Control Station, just running 'shutdown -h' is not enough. The secondary Control Station must be in 'power-off' mode during the upgrade. Reboot required at upgrade CNS software The CNS software is comprised of two installation components, a Linux package and a CNS software package. The software currently running on the Celerra and the CNS upgrade software versions determine the number of Control Station and Data Mover reboots required and whether Data Mover reboots can be scheduled to a later time after the upgrade is complete. The DART software runs on the Data Movers and is part of the CNS software package. The new version of DART is loaded and run on the Data Movers only after a Data Mover boot. Control Station rebooting Most software upgrades require a Control Station reboot. When upgrading CNS software, if either of the first two digits of the new version number change (for example, 5.6.x to 6.0.x), the Control Station requires rebooting.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
84
Installation and upgrades
At the beginning of upgrade, the system information section lists whether or not the Control Station reboot is mandatory. Note that this initial output is long and may scroll off the screen. You will be prompted whether to continue the upgrade after all the upgrade information is displayed. If you enter no, the upgrade stops, and you are returned to the command prompt. ===========System Information=========== Install Manager Version: 6.0.34-4 Install Manager Command: /celerra/upgrade/pkg_6.0.34.4/install_mgr Starts at: Thu Jun 3 10:36:56 2010 From version: 5.6.49-1 To: 6.0.34-4 Dual Control Stations: yes Slot: slot_0 Blade slot(s): 2, 3, 4, 5 Cabinet family: NS Cabinet type: NS-HAMMERHEAD Control Station reboot: multiple (mandatory) Blade reboot: 1 time(mandatory) Total number of attempts: 0 ========================================= . .< other information is posted here> . Do you wish to continue with the upgrade now [yes or no]? > yes Data Mover rebooting The Data Movers require a reboot in order to run the new version of DART. When upgrading CNS software, if either of the first two digits of the new version number change (i.e., 5.6.x to 6.0.x), the Data Mover requires rebooting.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
85
Installation and upgrades
The system information screen is the same as the information in the Control Station rebooting screen. Limited scheduling of the Data Mover reboot exists. An estimated time for when the Data Mover reboot will happen is shown at the beginning of upgrade: ==============Estimated Time for Services ===============
Current Time: Estimated Time when NAS service will be stopped: Estimated Time when Data Movers will be reset: Estimated Time when NAS service will be restarted: Estimated Time when upgrade will be complete: 10:36 10:38 11:28 11:39 11:50
================================================== The Upgrade Data Movers task prompts before it reboots the Standby Data Movers. It then provides a menu to reboot the Primary Data Movers similar to the following: ==================================================
11:58 [ 46/70 ] Upgrade Data Movers 7 minutes Start at: Thu Jun 3 11:58:38 2010 All servers must now be rebooted. Ready to reboot the Standby Data Mover (3). Press the ENTER key to begin... The following server(s) need to be rebooted to run the new release: 2) server_2 Enter slot number(s) or ALL to select server(s) for reboot:
================================================== When upgrading DART, if both of the first two digits of the new version of software remain the same (for example, 6.0..x to 6.0.y), you may delay the rebooting of the Data Movers. The system information screen is the same as the information in the Control Station rebooting example except for the following: Data Mover reboot: 1 time (advisory) The Upgrade Data Movers task asks if you wish to reboot the Data Movers. If you enter yes, you will be prompted to press ENTER to reboot the standby Data Movers, then prompted to reboot the primary Data
86
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Installation and upgrades
Movers. If you enter no, the Data Movers continue running the current (pre-upgrade) version of CNS software until they are rebooted. =================================================
12:29 [ 44/67 ] Upgrade Data Movers 7 minutes All servers need to be rebooted to run the new release. Do you wish to reboot them now? [yes or no]: Ready to reboot the Standby Data Mover (3). Press the ENTER key to begin... The following server(s) need to be rebooted to run the new release: 2 ) server_2 Enter slot number(s) or ALL to select server(s) for reboot: Waiting for ping response from server_2... :up (0 secs)
================================================= Schedules during upgrades As part of the upgrade procedure, schedules that run between midnight and 1:00 A.M. may lose those run times. This affects only schedules that run every X hours of the day, (X being from 1 through 12), and not schedules that run at specific hours of the day. This is a known problem and can be corrected by modifying the affected schedules after the upgrade. Upgrade requirements Existing read/write-able file system must have 128K minimum free space prior to upgrading to CNS version 6.0.
CLARiiON upgrades
Occasionally, CLARiiON storage connected to a Celerra Network Server requires a software upgrade. When upgrading CLARiiON software, do not run heavy loads on the Celerra, as operations may be extremely slow or interrupted. Notes CLARiiON array software cannot be upgraded unless the Celerra Control Station software is upgraded first. Ignoring this rule may make the CLARiiON unmanageable by the Celerra. Incremental CLARiiON arrays cannot be added to systems running
EMC Celerra Network Server Version 6.0.70.4 Release Notes 87
Fixed problems
older Celerra software unless they are running a compatible release of CLARiiON software. A warning will appear when you upgrade the Celerra software if the CLARiiON software needs to be updated. If you receive this warning, finish upgrading the Celerra software and then upgrade the CLARiiON software as soon as possible to avoid compatibility issues.
Consult the E-LabTM Interoperability Navigator at http://Support.EMC.com for all software interdependencies.
Fixed problems
Fixed problems categories
The Fixed problems section is grouped by the following categories:
Category CEE Celerra Management Applications CIFS Control Station CORE Deduplication FileMover FLR Hardware MPFS Install, Config, and Upgrades Internationalization NDMP/Backup Networking NFS Replication Security Meaning Celerra Event Enabler, CAVA Unisphere, Celerra Startup Assistant, MMC snap-ins, and so on. Common Internet File System (Quotas, Usermapper, ans so on) Celerra Control Station Celerra Data Movers Deduplication Celerra FileMover File-Level Retention Hardware Multi-Path File System Celerra installs, callhomes, configuration, upgrades Internationalization (I18N) Network Data Management Protocol or Backup Networking issues Network File System Celerra Replicator, SnapSure, Checkpoint LDAP
88
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.70.4
SRDF Storage Tools UFS
Symmetrix Remote Data Facility Storage Storage EMCOPY, EMCACL, LGDUP, SHAREDUP UNIX File System
Version 6.0.70.4
Backup Severity Symptom description
Data Mover panic Page Fault Interrupt caused by sending NDMP an empty software set file filter or directory filter string.
Fix summary
The code has been modified to fix the filter logic for the empty file filter string or the directory filter string.
Tracking number
1
496542 / 48722540
Celerra Management Applications Severity Symptom description Fix summary
The code has been modified to resolve the issue by optimizing the Tree Quota query processing on the XML API server. The code has been modified to avoid collecting the wasted space statistics. The former behavior is still available by running acl database printstats slow.
Tracking number
2
High CPU and memory usage by the XML API server process when running the VNX Monitoring and Reporting Tool. All CS operations were impacted.
455143 / 43142954
1
On file systems with a very large number of acls, running the acl database printstats command might take a very long time.
467244 / 45564250
EMC Celerra Network Server Version 6.0.70.4 Release Notes
89
Version 6.0.70.4
CIFS Severity Symptom description
Could not immediately connect to share on the system when many users were connecting and disconnecting simultaneously. Sometimes could not connect to share when the system was overloaded by share connections.
Fix summary
Tracking number
2
The code has been modified to improve the connection logic. The length of the encrypted ticket component is now checked before allocating memory, and an error will be returned if the length is zero. Also, the share mount request fails, and the error is logged in the server log by the calling code. The single sparse FT log file has been replaced by two regular files which will be written to one by one. When the first is full, it will be renamed with a bak extension and the second one will be created. The code has been changed to not trigger the scan if a user closes a file opened as read-only, after another user has previously opened and modified it. The code has been modified to move a joined compname tranparently accross VDMs of the same Data Mover.
495620 / 48114692
1
Data Mover panic caused by receiving an invalid Kerberos service ticket.
505036 / 49707766
1
When all Common Event Publishing Agent servers were offline, the CIFS access could be lost briefly if the faulttolerance log file became full.
512489 / 50330376
2
Poor CIFS performance leading to DU could be caused by unnecessary virus scans after a file was opened as readonly.
519560 / 50672554
1
Issues were encountered after moving a compname to another VDM due to the Active Directory replication problem, especially for sites with multiple Domain Controllers.
522085 / 51136100
90
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.70.4
CIFS Virus Check Severity Symptom description
When the AV Engine was not available, the following message was logged: Warning: RecvStream failed I/O error.
Fix summary
The code has been modifed to suppress alert messages that occur while killing a MSRPC client connection. The code has been modified to disable CEPP events for users who have the CEPP bypass right, and also to enable the CEPP event bypass right for users with this right who are not on a CAVA/CEPP server or connected on CHECK$. The code has been modified to recognize this state and handle it correctly.
Tracking number
499970 / 48272012
2
2
CEPP events were generated even if the user had the CEPP bypass right. CAVA/CEPA users with "bypass" rights can be suppressed.
504473
2
When a file system was unmounted with files waiting to be scanned by a virus checker, the files were not scanned after the remount of the file system on another path.
506486 / 49683544
Control Station Severity Symptom description Fix summary
The code has been modified to mount NDMP checkpoints of a VDM root file system to the VDM server. The code has been modifed to send the fsck request to the parent Data Mover.
Tracking number
2
The checkpoint creation on the VDM root file system could fail during the VBB backup.
476488 / 46629630
2
The Data Mover could fail to start fsck on a writeable checkpoint mounted on an unloaded VDM.
502351 / 49425182
EMC Celerra Network Server Version 6.0.70.4 Release Notes
91
Version 6.0.70.4
2
A CIFS server created from GUI could fail to join the domain due to a leading dash in the administrator's password. NDMP backup failed for the production file system mounted with the mount point that contained two root vdm prefixes, for example, /root_vdm_1/root_vdm_1/productin_fs. The SNMP trap did not get sent for a file system notification.
The code has been modified to correctly handle such a password. The code has been modifed to avoid this problem. The code has been modified to avoid this problem.
503070 / 47564604
2
525208 / 51439440
2
487350 / 47698828
DART Core Severity Symptom description Fix summary
The code has been modified to better handle the case when clock interrupts stop unexpectedly and to allow more time for the system to recover.
Tracking number
1
An idle Data Mover might panic with SYSTEM WATCHDOG or System clock was stalled when there was no explanation for why this happened.
500581 / 49117864
EE Escalations Severity Symptom description
If the Control Station couldn’t resolve the IP address of ftp.emc.com, the following error could appear in the system log: transfer_support_materials[241 18]: The transfer script has failed. It cannot transfer data to ftp.emc.com or the remote directory. Please contact your authorized service provider to make sure the remote server has sufficient free disk space (100MB). The code has been modified to use the new IP address for ftp.emc.com as the default IP address for transferring support materials.
Fix summary
Tracking number
2
520928 / 48169232
92
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.70.4
FLR Severity Symptom description
Data Mover panicked due to the use of a NULL pointer when a file system that is mounted read-only attempted to log something to the flrLogHandler.
Fix summary
The code has been modified to check for read-only file systems before they attempt to use the flrLogHandler.
Tracking number
505349 / 49743518
1
IP Replication v2 Severity Symptom description
The memory leak in the Replication V2 error path was caused by the improper manual cleanup of a session. This could result in an out of memory panic.
Fix summary
The code has been modified to remove the memory leak that occurred during some error recovery operations.
Tracking number
494281 / 48485034
1
NFSv4 Severity Symptom description
The NFSV4 access might take a long time if multiple clients work on the same files. The delegation recalling process led to poor performance, especially when some clients did not reply to delegation recalls immediately. Page Fault Interrupt. Virt ADDRESS: 0x62f870 Err code: 0 Target addr: 0x10
Fix summary
The code has been modified to enhance the locking mechanism, allowing multiple requests to run in parallel. The code has been modified to synchronize the release of a lock owner with other lock owner operations. The code has been modified to avoid a deadlock related to the NFSv4 delegation recall and conflict checking.
Tracking number
2
490956 / 48099274
1
492876 / 48323880
1
The NFS access was lost during backup.
496917 / 48792858
EMC Celerra Network Server Version 6.0.70.4 Release Notes
93
Version 6.0.70.4
1
When using NFSv4, Data Mover panicked with a dump header similar to: DART panic/fault message: *** GP exception. Virt ADDRESS: 0x00006af142. Err code: 0 *** The NFSv4 service was not responding; "Pool:NFSD_Exec BLOCKED" appeared in the datamover log. The NFSv4 response was slow with the LDAP resolver.
The code was modified so that the NFSv4_setAttr could write, exceeding the mblk limit, in this rare condition. The code has been modified to resolve the dead-lock issue. The code has been modified to no longer query group members.
502522 / 49446140
1
503276 525643 / 50177872
2
Quotas
Severity Symptom description When the user renamed a directory on which online quota check was running, a race condition could occur which caused threads to be blocked and file system to be inaccessible. The system would panic if users tried to enable quota on an MGFS filesystem. Fix summary Tracking number
1
The code has been modified to eliminate this race condition. The code has been modified to disallow the quota feature on MGFS to avoid the panic.
509083 / 50156424
1
519520 / 50889786
RDE - PE
Severity Symptom description The state of deduplication showed 0% saved data even though the files had been deduplicated. This occured when a deduplication scan was not completed due to a file system or a checkpoint resource error. The hash table mechanism used by the Policy Engine to track hard links could cause memory fragmentation issues after the table was deleted.This could lead to an out of memory panic with a header like: PANIC in file: ../bufpool.cxx at line: 1309 : couldn't get a free page or PANIC in file: Fix summary The code has been modified to allow the FRDE PE to continue scanning the file system after encountering a resource error. The code has been modified to create a private hash table using a memory mechanism that avoids the risk of fragmentation. Tracking number
2
372977 / 35129780
1
499671 / 49145068
94
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.70.4
../addrspac.cxx at line: 503 : Out of memory The code has been modified to handle dedupplication of empty files in a manner that does not lead to poor performance.
1
The Data Mover became inaccessible after the EM copy was done. Need to reboot the Data Mover to get access back.
512668 / 50460864
Transport Protocols
Severity Symptom description Under rare circumstances, Data Mover might panic with the following stack: gdb) bt #0 PANIC (string=0x1918be0 ">>PANIC in file: ../malloc.cxx at line: 662 : \n Memory corrupted") at ../panic.cxx:277 #1 0x0000000000243e60 in free (base=<value optimized out>) at ../malloc.cxx:662 #2 0x00000000002440e2 in operator delete (p=0x500) at ../_delete.cxx:23 #3 0x0000000000485def in CLIENT::operator delete (p=0x500) at ../client.cxx:522 Fix summary Tracking number
1
The code has been modified to remove the real-time race condition by adding synchronization in deleting paths.
473726 / 46291616
UFS
Severity Symptom description The Data Mover used the netcap command to capture the network traffic of a network device and store the network packets to an internal file. When the file was full, a staging file would be used. If the staging file was not successfully opened, the Data Mover would panic. Fix summary Tracking number
1
The code has been modified to check the file handle of staging buffer before it is used.
511398 / 50349810
EMC Celerra Network Server Version 6.0.70.4 Release Notes
95
Version 6.0.70.4
1
When the Data Mover was enumerating file system checkpoints from a Windows client, it panicked with the following DART panic/fault message: Page Fault Interrupt. Virt ADDRESS: 0x000191688f Err code: 0 Target addr: 0x0000000008 **. This happened when at least two checkpoints for the file system had different time stamp for Month field while having the same stamp of Seconds, Minutes, Hours and Days.
The code has been modified to correct an invalid boundary check. This resolves the issue of the incorrect return of a NULL pointer for a valid date range.
523621 / 51266466
UFS Log
Severity Symptom description Fix summary The code has been modified to increase the size of the array, to sense and handle the overflow, and to track current users of the array elements to enhance diagnosability. Tracking number
1
Overflow of an array in a filesystem inmemory data structure led to a deadlock and hang condition.
522045 / 50805196
Backend Features
Severity Symptom description "nas_rdf -restore" failed with storage profile mismatch error if the file system was created from a storage pool. Fix summary The issue is resolved by refreshing storage profile name reference for rdf restore. Tracking number
1
531824
System Management
Severity Symptom description Hitachi EFD was showed as STD disk in the gateway system. Fix summary This issue is fixed by adding the code to correctly recognize the drive type from Hitachi. Tracking number 49746724/ 508000
2
96
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.70.4
VBB backup of VDM root file system failed. 2
This issue has been fixed by mounting NDMP checkpoints of VDM root file system to VDM server instead of server_2.
46629630/ 476488
1
XML API server processing cost high CPU and memory usage when running VNX Monitoring and Reporting Tool. All CS operations were impacted .
The issue has been resolved by optimizing the Tree Quota query processing on XML API server.
43142954/ 455143 51087704/ 521751 50912918/ 522374
SNMP trap did not get sent for a filesystem notification. 2
The issue has been resolved by improving the code responsible for configuration of SNMP traps. The issue has been resolved by sending fsck requests to the parent Data Mover of the unloaded VDM.
47698828/ 487350
It failed to start fsck on a writeable checkpoint mounted on an unloaded VDM. 2
49425182/ 502351
3
CIFS server created from GUI failed to join the domain due to a leading dash in administrator's password.
The issue has been resolved by adding a transient internal password prefix. The issue has been resolved by removing the code that strips off the root VDM prefixes from the mountpoint.
47564604/ 503070
1
NDMP backup failed for production filesystem mounted with the mount point containing two root VDM prefixes, for example: /root_vdm_1/root_vdm_1/producti n_fs.
51439440/ 525208
EMC Celerra Network Server Version 6.0.70.4 Release Notes
97
Version 6.0.65.2
NDMP backup failed due to automatic ckpt creation failure when the file system was created with a physical DMbut mounted with a VDM path through GUI. 2
This issue has been fixed by adding a check that will prevent the GUI from creating a file system with a physical DM but mounted with a VDM path.
48710464/ 496443
Version 6.0.65.2
Backup Severity Symptom Description
The Data Mover panicked with a Page Fault Interrupt when VBB backup ran on a file system larger than 2TB.
Fix Summary
The code has been modified to prevent panic due to integer overflow.
Tracking Number
458269 / 44465210
1
Celerra Management Applications Severity Symptom Description Fix Summary Tracking Number
396490 / 38065374
1
Users could not log on to Unisphere if the passwords contained URL special characters like "&, ; ,+".
The issue has been fixed by encoding the password and user name to URL encoding. This issue was fixed by correcting the formula for calculation of the graph data.
2
"% CPU Idle" graph on "Resource Utilization History" page in Celerra Monitor showed value of 0%.
481652 / 46960472
CIFS Severity 2 Symptom Description
Files renamed by SMB2 clients could not be submitted to the virus checker.
Fix Summary
Code has been added to submit files renamed by SMB2
Tracking Number
472623 / 45655462
98
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
Severity
Symptom Description
Fix Summary
clients to the virus checker.
Tracking Number
2
The user was not able to delete a DFS link that existed beneath another DFS link.
The code disallows creation of a DFS link beneath another DFS link. This is consistent with Windows server behavior. The ACE now displays the owner name resolved through NIS and its uid number. It is no longer necessary to go to the owner tab. The code has been fixed to restore ability to enumerate and report user' s quota. The code has been updated to not scan a file if it has just been created during the open operation. The code has been fixed to correctly handle unreachable path even when it is longer than 220 characters. The code has been improved to rearrange some tests in the lock manager so that checking lock conflicts consumes less CPU. The code has been fixed to allow open a file for read attribute even when another process has the file open.
475820 / 46376654
1
When a user changed the UNIX rights of a file located on a file system mounted with MIXED access policy, the ACE displayed on Windows showed only "Owner Rights". To check who corresponded to this ACE, the user should go to the owner tab. The access checking was not affected. Quotas enumeration stopped responding indefinitely on Windows7 and W2K8 clients (clients using SMB2 protocol). A file could not be created on File Level Retention (FLR) if scan on read was enabled.
478365 / 46563750
2
481974 / 46640214
2
482464 / 47200498
1
server_cifssuport -cred -path <path> command panicked on Data Mover when the path specified was longer than 220 characters and was unreachable.
486226 / 47626966
1
Data Mover CPU consumption reached 100% when there were a lot of file range locks.
487750 / 47801326
2
Internet Explorer returned an error when downloading files on Windows 7 with 64bit CAVA.
490177 / 47973450
EMC Celerra Network Server Version 6.0.70.4 Release Notes
99
Version 6.0.65.2
Severity
Symptom Description
OS X 10.7 could not mount CIFS FS because the reply of opening CIFS might report invalid maximum access rights of the logged user if rights were set from NFS on a file system mounted with mixed access policy. The Data Mover panicked when virus checker was enabled with param waitTimeout set.
Fix Summary
The open reply now correctly reports maximum access rights of the logged user. The code has been fixed to avoid the panic condition.
Tracking Number
2
491591 / 45540912
1
DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x000110f214 Err code: 0 Target addr: 0x0000000800 ** Data Mover panicked with “out of memory” when CEPP was started with FT level=1 and no more CEPP servers were available. The code has been fixed to avoid the panic condition.
496627 / 48742734
1
The following message was displayed: DART panic/fault message: >>PANIC in file: ../bufpool.cxx at line: 425 : couldn't get a free page
499427 / 49063234
CIFS Virus Check Severity Symptom Description Fix Summary
The code has been fixed so that when a client disconnects, all CIFS threads working for the client are unblocked. The code has been fixed to generate the correct event.
Tracking Number
2
The disconnection of a client did not unblock CIFS threads.
476743 / 46658574
2
CEPP generated a “file open read” event incorrectly instead of “open dir” event.
500020
100
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
Control Station
Severity Symptom Description Fix Summary Tracking Number 467458 / 45558122
1
Some of the Unisphere GUI pages like "New File System Page" and "File System Properties Page" took over 20 minutes to load on a gateway system connected to Symmetrix array.
The performance issues have been fixed by revising the code. The Unisphere GUI pages take approximately 2 seconds to load with the fix. CCMD message definition has been modified to clarify the term "matching storage pools". This issue has been fixed by displaying "NoLimit" instead of "EXPIRED" when file system quota grace period is defined as "unlimited". This issue has been fixed by removing an old restriction which limited the new HLU number to be assigned to a newly discovered LUN. This issue has been fixed by limiting filesystem inconsistent density check to CLARiiON-based filesystems only.
2
Creation of a Virtual Data Mover on a mirrored storage pool failed with an unclear error message "No space left in matching storage pools". nas_quotas report displayed "EXPIRED" in "Timeleft" column even when filesystem quota grace period was defined as "unlimited".
375163 / 35536804
2
466085
2
nas_diskmark failed to mark a new LUN and displayed a warning similar to the following:- "Warning: 17716815787: server_2 c16t0l6 skipping reserved LUN id, APM00111803762 stor_dev=0x00ED (237 decimal), maps to a reserved host LUN id 0x6." If nas_diskmark executed on a gateway system tried to attach to a Symmetrix array, it failed with an error similar to the following:- "Error 5008: server_5 :13421849340: File system root_avm_fs_group_1 is made up of disk volumes with inconsistent density."
472963 / 46066538
2
474902 / 46033776
EMC Celerra Network Server Version 6.0.70.4 Release Notes
101
Version 6.0.65.2
2
For a production file system mounted on VDM, if a new checkpoint was created with a name that has one or more spaces and with automount=yes, the checkpoint was created but remained unmounted. The following error was returned: Error 5005: Invalid argument failed to complete command. Disk Provisioning Wizard failed on NX4 system with the following error: "Not able to process templates. n Command failed, error code: 1, output: errorMessage:string="Timeout (180 seconds) waiting for state SS_TEMPLATES_GENERATED". " CLI command "nas_pool -list" failed with the following error: "Error 2237: Execution failed: Segmentation fault: Operating system signal. [VIRTUAL_DISK_TABLE.is_profile_s upported] "
The issue has been fixed by replacing every space with "_" character in mount path created from the checkpoint name.
476767 / 4665800
2
This issue has been fixed by revising the code responsible for generation of hot spare templates.
483778 / 46920576
2
The issue has been fixed by initializing the storage_profile_
name to empty string when the name is not present in database record. This issue has been fixed quoting file system name in "rootnas_fs" command to allow spaces.
487182 / 47678224
3
File system auto extension failed when there was space in file system name.
478780 / 46910078
102
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
DART Core Severity Symptom Description
Log event interface showed "Parsing error. Invalid XML” messages. The server_log showed following valid error messages, but with garbage characters at the end of each message. 2012-03-28 09:13:37: 150596354069: TIMESYNC: 3: ReadStream failed in time synchronization poll, error: _ŠÁ 2
Fix Summary
The code has been corrected to display the accurate error messages. The log event no longer displays "Parsing error. Invalid XML” messages.
Tracking Number
479218 / 46947888
.
2012-03-28 09:13:37: 150596354081: TIMESYNC: 3: Connection to the timeserver lost 2012-03-28 09:13:38: 73286942722: TIMESYNC: 1: 2: The NTP servers are not accessible or responding to the time service from [Wed Mar 28 09:13:37 2012]. The Data Mover panicked with the following message: DART panic/fault message: The code has been fixed to avoid the panic condition. 483525 / 47383678
1
>>PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG
Deduplication Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
103
Version 6.0.65.2
1
Due to a race condition with DART's internal dedupe process, deduplication could be attempted on an inode that was deleted and recycled into another type of file. This could cause an internal file to be mistakenly deduped. The result could be loss of access due to blocked threads.
The race condition has been fixed. It is no longer possible for the DART deduplication process to accidentally compress the wrong inode.
487774 / 47818786
File System Severity Symptom Description
When a file system was nearly full or reached a quota limit, a large copy from an SMB2 client could fail. If you retry the copy, it may complete successfully. Fsck panicked with header "PANIC in file: ../DeltaIO.cxx at line: 725 : DeltaIO_Read_CB::failed" due to bad block in the opaque data mapping pointers. This caused an extended DU situation. When the amount of metadata changed, it caused the ufslog to become full. Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../staging.cxx at line: 439 : failed to write UFS log The result was that file systems should be checked before they could be mounted.
Fix Summary
The Code has been fixed to correct the method to reserve and consume blocks during a copy from SMB2. The code has been enhanced to check for and skip bad blocks.
Tracking Number
1
477292 / 46343472
1
483028 / 46874118
1
The code has been fixed to prevent the panic. When the Staging buffers are full, the current buffer will only be updated if the next buffer is unused.
490046 / 48019080
MPFS
Severity Symptom Description A discrepancy may build between the du and df commands for an MPFS file system. Fix Summary Tracking Number 463971 / 45169514
1
The issue has been corrected to free space
104
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
on MPFS mounted file system when files are deleted.
Network
Severity Symptom Description Fix Summary Tracking Number
DART panicked when the user issued a very large SNMP bulkcache request with the following message:
DART panic/fault message: 1 >>PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG The code has been fixed to avoid the panic condition. 478969 / 46919840
The following panic might occur while running Nesus security scanner: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x000020f310 Err code: 0 Target addr: 0x0000000028 ** Panicking Processor: 2 DART stack trace: 2860: 0xc2aef0 _ZN7DskDump9checkdumpEPKc+0x7a 28a0: 0xc2b0fe dump_return+0x4 28b0: 0x14d0b2 _ZN12Memory_Frame13setProtectio nEm19Memory_Prot 28d0: 0x14d73d fault_dump+0x89 28f0: 0x14165d handle_fault+0x1d 69910: 0xd89489 _ZN13cPipeSimulate11checkPolicy EP13smb_threadC 69950: 0xde498b _ZN21cMsRPC_LsarpcSimulate15lsa rCloseHandleEP1 699b0: 0xdf202e _ZN21msrpc_legacyOperationI21cM sRPC_LsarpcSimu
1
The code has been fixed to handle invalid MSRPC requests gracefully, returning an error.
484635 / 47497156
EMC Celerra Network Server Version 6.0.70.4 Release Notes
105
Version 6.0.65.2
NFS
Severity Symptom Description It was possible to receive an illegal RPC reply with a NULL body and a status of "YP_TRUE" when using NIS with an NIS Server AIX 5.2 and/or AIX 5.3. The system expected the RPC message to contain at least 4 bytes and was not prepared to handle a NULL body RPC request, so it processed the request normally. This would result in storing data more than one byte into one byte space. That overwrite led to the following panic: DART X86_64 panic message: >>PANIC in file: ../malloc.cxx at line: 662 : Memory corrupted Fix Summary Tracking Number
1
The code has been fixed to check for a NULL RPC body to properly process the NULL message and reject the bad NIS response.
469561 / 45818200
NFSv4
Severity Symptom Description The Data Mover panicked when a file system was unmounted while there was NFSV4 activity on it, with the following message: Fix Summary Tracking Number
1
Page Fault Interrupt. Virt ADDRESS: 0x00006a4ba3 Err code: 0 Target addr: 0x0000000010 ** The premature revocation of LOCK states by the server caused BAD_STATEID errors. The server might revoke the LOCK state as soon as there was no file lock left attached to it;
A spinlock has been introduced in the code to avoid reusing NFSv4 states which are being released.
476566
2
This did not comply with the RFC.
The delay between the last LOCKU operation and the revocation of the LOCK state could vary between 0 and 5 minutes (nfsv4.cleaningInterval).
The Data Mover no longer releases the idle LOCK states associated with files which are open and idle.
484690 / 47472438
106
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
Proactive Health Checks
Severity Symptom Description If a CIFS server was present with WINS servers (used wins= option with server_cifs while adding CIFS servers), the server_checkup failed with the following error: Warning 17451974760: server_2 : VDM1: The WINS server '<IP address>' does not contain the correct IP address for the name '<CIFS server name>'. This one should be associated with <IP address>. The CIFS clients of this NetBIOS server might not be able to access the Data Mover. The code has been fixed to enhance the IP address comparison function used during server_checkup. Fix Summary Tracking Number
2
479534 / 46324040
Quotas
Severity
Symptom Description
While you had named streams in files, after running online quota check, quota block and inode usages would be less than the previous number. Querying user quotas on a tree produced no result, no error message from Linux client over NFS.
Fix Summary
The code has been fixed to calculate named stream and container files in quota check correctly. The code has been fixed to check the quota tree on/off status and return correct quotas.
Tracking Number
2
466450 / 43457680
3
473741 / 45791204
EMC Celerra Network Server Version 6.0.70.4 Release Notes
107
Version 6.0.65.2
3
Any query for tree quota usage following the modify tree quota operation returned 0. The same query would return values 24 hours later.
The code has been fixed; the query to return tree quota usage information following the modify tree quota operation now returns accurate results. A new option has been added to the param quota useQuotasInFsStat to enable reporting only file system and tree quota accounting to df queries from Unix clients. The existing values will behave as before. 0: (default) only the file system capacities will be reported to Unix clients 1: the file system and all quotas (user, group and tree) are taken into account when capacities are reported to Unix clients 2: file system and the tree quota are taken into account when capacities are reported to Unix clients The fix ensures the tree quota accounting is accurate.
474357 / 46279178
There was no option to report capacities and usage based on file system and tree quotas that did not take user and/or group quotas into account.
1
488023
1
When a large file was deleted, this was handled in the background as a deferred delete. If this process was interrupted by the file system being unmounted, it could result in tree quota accounting being incorrect.
493473
108
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
Replication
Severity
Symptom Description
Data mover panicked with the following message:
Fix Summary
Tracking Number
1
DART panic/fault message: *** Invalid Opcode exception. Virt ADDRESS: 0x0000006bf6 ***
The code has been fixed to avoid the panic condition.
489235 / 47953422
Snapsure
Severity Symptom Description Data Mover bootup or failover/failback could be slow due to checkpoint recovery when there were mergeDone checkpoints.These were caused by deleting or refreshing checkpoints which were not the oldest for the file system. The chunks for the deleted/refreshed checkpoints were merged into existing checkpoints and these chunks caused extra processing to be done during recovery after a reboot or failover/failback. Fix Summary Tracking Number
2
The checkpoint recovery code has been improved to skip reading the mergeDone chunk headers as it is not necessary.
408286
1
When latest ckpt was very large (several TB), it would take a long time (longer than 5 minutes) for reboot or failover. Depending on the content in slot B, this would cause bad cg, fs corruption or btree corruption. The data mover panicked with the following message: DART panic/fault message: finishTake: bad magic in CG
The checkpoint recovery code has been improved, leading to improved reboot or failover time.
436839
1
The code has been fixed to avoid the panic.
472453 / 46122574
EMC Celerra Network Server Version 6.0.70.4 Release Notes
109
Version 6.0.65.2
1
Writing a PFS where a checkpoint was created caused a panic with the following message: "io not progressing". The recovery process was suspended after the panic. PFS on the replication destination side might be corrupted as a result of this.
The code has been fixed to correct the Snapsure recovery process.
483546 / 47403530
1
When replication started in reverse with HWM 0, it always returned an File_NoSpace error.
The code has been fixed so that only when the usage is greater than or equal to 75%, the replication reverse would fail, otherwise it succeeds. The code has been fixed to avoid panics by: • Avoiding the insertion of invalid block numbers • Adding checks to bypass invalid block numbers that may have been written previously.
491140 / 48120034
1
Repeated Data Mover reboots could be triggered during replication transfer or checkpoint merge on >2 TB file system. Any of the following messages could be displayed: Panic in file: ../DeltaIO.cxx at line: 802 : DeltaIO_Read_CB::failed Assertion failure: 'cv>unpackBtree32(curadd) >= startAddress/VOLBLK_SIZE' on line 3286 in file '../PagedBlockMap.cxx' >>PANIC in file: ../assert.c at line: 27 : Assertion failure: 'val != INVALIDNUM' on line 428 in file '../include/PagedBlockMap.hx x' >>PANIC in file: ../assert.c at line: 27 : assert failed
497062 / 48857254
1
Rolling data mover panic occurred when a little amount of data wrote to PFS after a new CKPT creation and the corresponding save volume was corrupted. The following message was displayed:
The code has been fixed to avoid the panic condition by correcting the Snapsure recovery process.
497107 / 48872748
110
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
DART panic/fault message:
>>PANIC in file: ../PagedBlockMap.cxx at line: 2423 : Duplicate Mapping
Storage
Severity Symptom Description Data mover panicked with "Volume has zero paths" . Fix Summary The code has been changed to not prune paths to lun on which scsi inquiry fails. This avoids the panic. Tracking Number 458436 / 44563854
1
The following message was displayed:PANIC in file:
../include/volume.hxx at line: 414 : Volume has zero paths.
1
A rare panic was encountered while processing a Tachyon fibre channel driver's Outbound Completion Message (OCM) with a set Invalid Entry (INV) error bit, although neither the OCM's SEST Programming Error (SPE) bit nor the VAL bit in the associated SEST were set. This condition violated the rules of the OCM's INV bit as specified in table 9-2 of the Tachyon FC Protocol Controller User Manual. The driver explicitly panicked the Data Mover (panic code = 0x01902009 == FC_EXM_INVALID_API_ERROR_STAT US with an additional error code = 0x00003002 == FC_CTLR_PROG_ERR) whenever an OCM's invalid entry bit was set without even looking at the OCM's SPE bit or the SEST entry's VAL bit.
The fix prevents this explicit panic within the Tachyon fibre channel driver (panic code = 0x01902009 == FC_EXM_INVALID_A PI_ERROR_STATUS ) in the specific case (error code = 0x00003002 == FC_CTLR_PROG_E RR) where it violates the rules for the OCM's INV bit as specified in the Tachyon FC Protocol Controller User Manual.
459091 / 44655534
EMC Celerra Network Server Version 6.0.70.4 Release Notes
111
Version 6.0.65.2
1
DART panicked with a page fault. The following message was displayed: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000a69602 Err code: 0 Target addr: 0x0000000010 **
The offending code leading to page fault panic has been corrected.
470802 / 45934948
2
Deletion of the disk failed with error:"Error 3501: Storage API code=24: SYMAPI_C_DEVICE_NOT_IN_DG. Could not select requested device(s) from the specified device group"
The issue has been fixed by changing the way the code determines the name of the device group the disk belongs to.
494402 / 48470488
SRDF
Severity Symptom Description Fix Summary Tracking Number 473453 / 45851944
1
RDF Set mode sync operation failed when Sync license was not enabled.
The fix will ignore the license error message to continue the restore process. The fix should stop such a nd-clnt hung issue.
1
nas_rdf -activate could be hung in nbs update operation. ps -ef shows "nd-clnt 6 7" like "/usr/local/bin/nd-clnt 6 7 server_2,server_3,server_4,server_5".
475565
System Management
Severity 1 Symptom Description When a security user was renamed, the home directory did not get renamed to reflect new name. Fix Summary This issue has been fixed by renaming the home directory to match the new user name. If a directory with the new name exists, operation fails and an error is displayed. Tracking Number 90188
112
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.65.2
2
No event notifications e-mails were received when an event notification was configured with Action 'mail' and Destination 'user' .
This issue has been fixed by correcting the software error that prevented email notification to be sent to the correct destination. This issue has been fixed by correcting the user authentication code that handles disabled accounts. This issue has been fixed by proactively deleting outdated temporary files used to determine statuses of hardware components. The issue has been resolved by adjusting configuration parameter of Celerra Common Cache. This issue has been fixed by preventing Jserver from trying to access Symmetrix to check the threshold of the virtually provisioned pool. This issue has been fixed by moving the ECOM core files to /nas/var/dump/ partition, compressing the files and limiting the number of retained core files. The issue has been fixed by stripping off leading zeroes from Symmetrix serial number.
399760 / 37640588
2
After disabling a local user account, login into Unisphere through an LDAP authenticated account failed.
459428
2
Incorrect alerts about hardware components appeared in Unisphere GUI after the actual faults were corrected.
466065 / 45410530
2
XML API failed to return all provisioned volumes.
480447/ 45516028
2
Control Station periodically ran out of memory and rebooted.
486239 / 47586416
2
UIM could not discover the VNX and displayed the following error message: "HTTP error 405 method not allowed".
487238 / 47682672
2
Jserver returned disk volumes with Symmetrix serial number that had leading zeroes.
488668 / 44892982
EMC Celerra Network Server Version 6.0.70.4 Release Notes
113
Version 6.0.65.2
2
The following error was seen in sys_log and the Unisphere event log: transfer_support_materials[264 56]: The remote FTP server/directory cannot be reached. Verify your configuration in /nas/site/automaticcollection. cfg. Also, nas_checkup reported the following warning: Control Station: Check auto transfer status Warning HC_CS_18800050419: Cannot establish connection to FTP server sftphost sftp.emc.com Action : Command "/nas/tools/automaticcollectio n -verifyftp" failed to connect to FTP connection. Check the configuration in /nas/site/automaticcollection. cfg file. For more information, search the Knowledgebase on EMC Online Support as follows: 1. Log in to https://Support.emc.com and go to Support > Knowledgebase Search> Support Solutions Search. 2. Use ID emc221733 to search.
The issue was caused by a recent change in FTP server security, which caused the verification to fail. The code has been updated to work with the new FTP server security policy. See Primus article emc296211 for the temporary workaround for fixing the problem with the script until the system is upgraded to this VNX OE for File version.
491605 / 48169232
2
Checkpoint creation via XML API failed with error:- "The specified storage system '0' is not valid."
This issue has been fixed by modifying the XML API to correctly handle the configuration where both Control Station and the storage array have the same serial number.
493653 / 48326724
VDM
Severity Symptom Description Fix Summary Tracking Number
114
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
2
The size of the security.evt log file increased beyond 512 KB when located on the root file system or on the root VDM when the GPO setting was set to a higher value, thereby filling up the root file system.
The code has been modified to limit the size of the security.evt file to 512 KB when located on the root file system or on the root file system of a VDM, regardless of the GPO settings.
473430 / 46236640
Version 6.0.60.2
Backup
Severity 1 Symptom Description VBB restore operation failed in restoring directories with UTF8 characters. The Data Mover panicked when the SavVol filled up during a greater than 60-day long backup and the oldest checkpoint was inactivated. Fix Summary The code was fixed to handle directories with UTF8 characters. The code was fixed to protect against inactivated checkpoints panicking the Data Mover. The number of active versus inactive checkpoints was verified to be correct. The code was fixed to avoid the panic. Tracking Number 459567/ 43756396
1
460751/ 44831934
1
The Data Mover panicked with the following message when DMA sent garbage data in env list during ndmp restore. PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG
465634 / 45403186
Celerra Management Applications Following are the fixed problems related to the Celerra Management Applications, that is, Celerra Monitor, Celerra Manager, WebUI, or Celerra MMC Snap-ins:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
115
Version 6.0.60.2
2
Unable to join the CS to the storage domain. The CS appears to join properly and no errors are thrown but it never appears in the System List. This happened because systems on the same subnet with hostnames undifferentiated in the first 12 characters ended up with identical, non-unique System IDs.
The code was updated to generate unique System IDs, resulting in CS successfully adding to the storage domain.
435338 / 43225714
CIFS
Severity 1 Symptom Description Windows clients issuing a DOS 'dir' command which included a complex mask resulted in all CIFS access to the file server being blocked. A forced panic of the Data Mover was needed to restore client access. When the CIFS server had many open files (likely more than 100000, depends on file names size), the open files enumeration over MSRPC (MMC, srvmgr) could not complete. Fix Summary The code was changed to re-format DOS masks for optimal execution before they were submitted to the file server. The code was fixed to prevent the cleanup task from running on inprogress active requests. This allowed the open files enumeration to complete. The code was fixed to return the correct size of qtree. Tracking Number 460364/ 44773644
2
462010/ 44848002
2
When connecting by CIFS to a qtree using the share of the root of the FS and the qtree path, the size of the qtree may be incorrect (the size of the FS is returned instead of the size of the qtree). Users could experience access issues to server's resource when domain was configured to use Resource Groups through Active Directory Federation Services (ADFS), as described in the Microsoft document: http://technet.microsoft.com/
464557 / 45104472
2
The code was modified to handle credential resource groups correctly.
464981/ 45324904
116
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
enus/library/cc737832(WS.10).a spx
2 When a non server's administrator user attempted to connect C$ system share (access not granted), the access checking was not performed immediately, but access denied error was later reported on first request, when using the share. The server_stats and ”printstats cifs” did not report correctly for SMB2 TCPconnections and SMB1/SMB2 threads. The code was fixed to check for access immediately now. 466678 / 45510170
2
The SMB2 TCPconnections and SMB1 and SMB2 threads counters reported by server_stats were corrected. The code was enhanced to allow a CIFS user to create a file and get the Execute Unix mode bits set according the ACL of the file. The code was enhanced to improve performance.
468729/ No Clarify ID
2
The cifs acl.umask parameter was documented but not useable.
469523 / 45695180
2
Bad performance when copying several files from a Windows 7 client to a Data Mover share. This is because only one file was copied at a time.
473384 / 46207560
CIFS Virus Check
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
117
Version 6.0.60.2
Severity 2
Symptom Description When a file was being opened for writing by one client, and another client was reading attributes (or extended attributes) of this file, an unnecessary virus checker scan was triggered for the file causing write performance degradation.
Fix Summary With the code change, the virus checker scan could not be triggered when the second client was reading attributes only on a file that was opened for writing from the first client. The code was fixed to avoid the panic condition.
Tracking Number 463462/ 43901404
1
When an NFS user renamed a file on a file system with CEPP enabled for NFS, the Data Mover panicked with the following message: Page Fault Interrupt. Virt ADDRESS: 0x00007ec8f3 Err code: 0 Target addr: 0x0000000188
463584/ 45067744
CORE
Severity 2 Symptom Description The Data Mover panicked with the following message due to a memory leak when F-RDE was scanning a file system with a large number of hard linked files: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000159274 Err code: 2 Target addr: 0x0000000000 ** Under load, the Data Mover crashed with "Out of memory". The following message was displayed: DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory This could happen randomly Fix Summary The code was fixed to avoid the panic condition. Tracking Number 459484/ 44220834
1
The code was fixed so that the memory used by the DBMS internal mechanism was pre-allocated at the initialization of the system to avoid the lack of memory situation.
460260/ 44774392
118
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
Severity
Symptom Description when the memory was fragmented.
Fix Summary
Tracking Number
Deduplication
Severity 1 Symptom Description Heavy CIFS access to compressed file on an F-RDE enabled filesystem can cause excessive CPU load. Fix Summary The code was fixed to avoid the situation causing excessive CPU load. Tracking Number 469525 / 45397734
DHSM
Severity 1 Symptom Description The Data Mover panicked with the following panic message when the user created a DHSM http connection and tried to modify this connection using an incorrect fs_dhsm command (-user option was used but no user name was provided). DART panic/fault message: *** GP exception. Virt ADDRESS: 0x0000aa783f. Err code: 0 *** The user cannot read archived files when the stub file is restored from backup. Fix Summary The DHSM command parser code was made more robust to avoid the panic. Tracking Number 422536/ 41631464
1
The code was enhanced in the DART kernel at the file system level to support the call that VBB restore operations was using.
440647 / 43018304
File System
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
119
Version 6.0.60.2
1
Data Mover server_2 panicked due to "Unnamed stream is wrong type". The release of DART internal resources for files containing (or once containing) alternate data streams is a multi-step release. DART incorrectly allowed an update to an inode that had already been freed. That inode had already been reused by another file and its contents were overwritten.
The code was fixed to avoid this condition during the release of files containing (or once having contained) alternate data streams.
440553 / 43857930
1
Using NFSv3 and lockd, could not unmount a file system on the Data Mover, resulting in a timeout.
Fixed a race condition which can occur when the lockd client uses several interfaces to set file locks on the Data Mover. The memory allocation needed to enable a tree quota was split up so one large contiguous chunk is no longer needed. This will prevent the Data Mover panic when it has been up long enough for its memory to become fragmented.
458487 / 44588032
1
1
When creating a new tree quota the Data Mover panicked with: PANIC in file: ../addrspac.cxx at line: 529 : Out of memory DART stack trace: e6e30: 0x1417d4 _ZN19Memory_AddressSpac e10allocPagesEl12Memory e6e70: 0x242509 malloc+0x10b e6ed0: 0x243dcb _Znwm+0x1b e6ef0: 0x243df6 _Znam+0x4 e6f00: 0x812804 _ZN14UFS_FileSystem20tr eeQuotaOn_InternalEPcS0 e71c0: 0x815a26 _ZN14UFS_FileSystem16qu otaOn_InternalEPcS0_+0x e7300: 0x786df6 _ZN14VFS_FileSystem7quo taOnE9quotaTypetPcS1_+0 e7350: 0x779f92 CPU usage intermittently spikes to 5% idle when there is high read activity on large read-only file systems.
459816 / 44708688
The code was fixed so that the process to free inode block objects was
464911 / 45307774
120
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
modified, so it consumes less CPU resources. 1 A corrupted NLM request caused Data Mover panic with the following message: Page Fault Interrupt. Virt ADDRESS: 0x000087d7fa Err code: 0 Target addr: 0 Datamover stack: #0 Lockd_Initial::start(th is=0x360cdc40) lockdthr.cxx:1401 #1 Sthread_startThread_int ernal(ip=0x2d4464a60) sthread.cxx:698 When the ownership is changed for a file or directory containing Alternate Data Streams (ADS), the inodes for the ADS are not changed. This will cause user or group quotas to be incorrect. The Data Mover hung in a rare situation while there were 3 tasks running at the same time: quota check or quota on task, IO task, and snapsure task. Threads become blocked eventually leading to a loss of access. This can happen when a file or directory containing a compressed named stream is deleted. In DART 5.6.47 and above, DART 6.0, and VNX OE for File 7.0, compression of named streams is not performed so this issue is only possible on a file system where dedupe was enabled on 5.6.46 or earlier. Fixed the use of uninitialized pointer. 466579/ 45508014
2
Corrected the logic error to make the ownership of ADS consistent with the file/directory.
467158 / No Clarify ID
1
The code was fixed to avoid the hang.
468422 / 45715556
1
The code was fixed to change the locking that is required when deleting a named stream to avoid the potential of a deadlock.
469851 / 45822506
EMC Celerra Network Server Version 6.0.70.4 Release Notes
121
Version 6.0.60.2
1
Data Mover panics with: PANIC in file: ../ufslog.cxx at line: 2728 : filesystem log transaction record overflow This happens due to a rare case when a file system operation results in a ufslog transaction larger than the maximum allowed.
The code was changed so that the number of blocks which can be included in a ufslog transaction have been reduced to ensure that the total transaction size will not exceed the maximum allowed.
476163 / 46597972
FLR
Severity Symptom Description Data Mover panicked with the following message: DART panic/fault message: >>PANIC in file: ../rawirp.cxx at line: 607 : Write Verify failed for FLR data Fix Summary The code was fixed to avoid the panic. Tracking Number 461016/ 44839716
1
FSCK
Severity 1 Symptom Description If a PBR check was required during FSCK or if quota recalculation was required when mounting the file system, the inode scan routine did not print any progress message making the user believe that the operation was hung. Fix Summary The code was fixed so that the inode scan could print the progress messages. Tracking Number 463902/ 45121204
Network
Severity
1
Symptom Description If IPv6 was configured in the network environment, but IPv6 interface was not configured on the Data Mover, the Data Mover could periodically panic
Fix Summary The panic was triggered by invalid sanity checking code, which did not expect a destination
Tracking Number
440268 / 43805094
122
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
Severity
Symptom Description with a panic dump header similar to: *** Page Fault Interrupt. Virt ADDRESS: 0x0000489c28 Err code: 2 Target addr: 0x0000000010 ** Data Mover panicked with the following: PANIC in file: ../bufpool.cxx at line: 425 : couldn't get a free page If a customer joins a CIFS Server to a Windows domain that is using a non-Windows DNS Server, the subsequent secure (Kerberos) DNS update from the Data Mover will fail because the DNS Server is not registered in Kerberos for the domain. If the Kerberos credentials for the CIFS Server are cached in a certain order, memory leaks will occur due to a credential being retrieved from the cache and selected as a 'match', then not used.
Fix Summary address to be IPv6, and source address to be IPv4. This sanity checking code has been fixed.
Tracking Number
1
Additional calls to free the credential contents were added to prevent the memory leaks.
460236 / 44786942
1
The Data Mover panicked due to experiencing IP fragmentation and any of the fragments are zero length. The following message was displayed: DART panic/fault message: Page Fault Interrupt. Virt ADDRESS: 0x00003d49a8 Err code: 2 Target addr: 0x0000000000
The code was enhanced to avoid UDP and optimize infrastructure to avoid IP fragmentation.
470446 / 45912724
EMC Celerra Network Server Version 6.0.70.4 Release Notes
123
Version 6.0.60.2
NFS
Severit y 1 Symptom Description The Data Mover panicked with the following message: Fix Summary The code was fixed to avoid the panic. Tracking Number 470457 / 45931264
Page Fault Interrupt. Virt ADDRESS: 6638df Err code: 2 Target addr: 4 There were server logs like "NFS: 4: xid XXXX in progress - reset after drop more than 8 times " NFSv4
Severity Symptom Description The Data Mover panicked with the following message:
Fix Summary The code was fixed to avoid the panic.
Tracking Number 460992/ 44859406
1
DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x00006a7fc0 Err code: 0 Target addr: 0x0000000000 **
2 The AIX client hung when using NFSv4. The Server could no longer grant a read delegation if the open access mode was read + write. If a write delegation could not be granted, no delegation would be granted at all. The code was changed, so that the conditions which could trigger this infinite loop were fixed. 464549/ 44422786
1
All NFS threads were blocked causing NFS access to be lost. This could have happened due to an edge condition with NFSv4 when verifying the attributes for an NFSv4 operation leads to an infinite loop. The NFSV4 server returned the error NFS4_RESOURCE when
465354/ 45380486
1
The code was fixed to release an open
465606/ 45378140
124
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.60.2
Severity
Symptom Description there were too many open owners. This could also happen if there were few opened files. The reason was that open owners were released only after a lease period to allow close replay detection.
Fix Summary owner when it had no more files opened. This avoided the issue.
Tracking Number
Platforms
Severity 3 Symptom Description HTTP 413 “Request Entity Too Large” page might return unsanitized client header. The connectemc daemon hung and stopped sending the callhome files. Fix Summary Changed webserver configuration file to use a static error message. A solution was implemented to detect and restart the connectemc main process if it was hung, and added a regular restart mechanism of the connectemc main process as the last recovery resort. Changed the web server configuration file to disallow weak ciphers and SSLV2 by default New code was added to read local user if no domain user found, and display the local user in AUDIT_Tool output. Tracking Number 355867 / 33486008
2
453530 / 44047694
2
By default, the web server on the Control Station allowed weak ciphers and SSLV2 to be used. syslog messages report blank for Username fields in case of local user when AUDIT_Tool is run.
456362 / No Clarify ID
2
470581 / 43023590
Quotas
Severity 2 Symptom Description ”User Name” column on quota information page in Unisphere showed the User ID and not the Domain\Username. Fix Summary The code was fixed to display the user information correctly. Tracking Number 464377/ No Clarify ID
EMC Celerra Network Server Version 6.0.70.4 Release Notes
125
Version 6.0.60.2
SnapSure
Severity Symptom Description The Data Mover panicked on ckpt fs with fs corruption; or on snapsure layer with Duplicate Mapping. The following message was displayed: Fix Summary Code was fixed to avoid the panic situation. Tracking Number 439471 / 43674774
1
DART panic/fault message: >>PANIC in file: ../DeltaVol.cxx at line: 3844 : doSavVolIOVec
Storage
Severity 1 Symptom Description "server_mt -f [device] status" fails when connected to DataDomain VTL devices. The NDMP Backup failed with the following medium write error message on direct attached LTO5 drive with append only feature : Fix Summary Code was fixed so that server_mt returns the correct status. The code was changed to add support for 16 byte SCSI commands for tape/robot devices. Tracking Number 392665 / 36965502
2
431941/ 41648218
---------- server log--CAM: 3: cdb: 0a 00 00 04 00 00 00 00 00 00 00 00 | SCSI ERR ON #3200 camstat 0x84 bstat InvalidOperation | scsi stat.skey.ascq 0x02.07.5a02 Check condition | skey [ ] Write protect |____________________
EMC Celerra Network Server Version 6.0.70.4 Release Notes
126
Version 6.0.60.2
Severity
Symptom Description
Fix Summary
Tracking Number
ascq Operator selected write protect System Management
Severity 2 Symptom Description Abort of "server_ping -send" command with Ctrl+C hangs serial connection to Control Station. Fix Summary Code was modified to protect output to console operation from being interrupted by external signal, and the interrupt signal is delivered after the operation completes. The code was corrected to ensure that server events get retrieved according to the user specified severity level. Tracking Number 406307 / 39377292
2
File Events filtering by severity does not work correctly on Unisphere Event Logs page: when user requests "Info" events only, both "Info" and "Warning" server events get displayed. And when the user selects "Warning" events only, no server events get displayed at all. Upgrade failed with error message "Error
439164 / 42989446
1
IM_14505279794 : Failed to setup the following local storage devices ..." due to ongoing RAID group defragmentation on storage array.
3 Server_export command failed with error message ”Invalid argument” when a host FQDN contained "sec" clause.
The code was changed to remove obsolete 'exdisks' check performed while running 'nas_storage -check -all' command.
458484/ 44421782
The code that processed export options was modified to ensure that "sec" clause in FQDN did not get process as "sec=" export option.
460366/ 44678140
EMC Celerra Network Server Version 6.0.70.4 Release Notes
127
Version 6.0.60.2
Severity
Symptom Description Errors like the following appeared in the server log of a standby server:
Fix Summary The code was modified to disallow sending VTLU queries from Control Station to standby server.
Tracking Number 465968/ 45378244
2
2012-01-09 02:35:26: 13160087552: SVTL: 3: Command failed: poolsInfo 2012-01-09 02:35:26: 26044989440: SVTL: 6: SvtlServer() --> svtl_rename(/svtl, /.etc/svtl) failed: Stale handle 2012-01-09 02:35:26: 13160087552: SVTL: 3: mkDir(/.etc/svtl/lock s/libraries) failed: IO_Error 2012-01-09 02:35:26: 13160087552: SVTL: 3: SvtlServer() --> pTlusMutex>Init(/.etc/svtl/lock s/libraries) SVTL_STATUS_ERROR 2012-01-09 02:35:26: 13160087552: SVTL: 3: SvtlServer::Alloc(/.e tc/svtl): SVTL_STATUS_ERROR
2 Control Station exhibits high CPU load.
The code was fixed to reduce Control Station CPU load by limiting periodic filesystems queries initiated from Jserver
470280 / 45558122
UFS
Severity 1 Symptom Description Failed to mount the file system with the following message: ”Mount not allowed: too many mounts on this server”. Fix Summary The code was modified to handle this situation. Tracking Number 165703/ 33316276
128
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
Severity 1
Symptom Description CPU usage spikes of 100 percent lasting for several seconds occurred when clients accessed tens of thousands of directories. The Data Mover panicked with panic header ”Assertion failure: 'aclHdr.getHdr()->revision == ACLRev” if ACL database was full (1 million different ACLs) and there were concurrent threads trying to allocate a new ACL.
Fix Summary The code was changed to reclaim memory used in directory processing more efficiently. The code was fixed so that the solution eliminated the condition where two threads attempt to allocate the last ACL entry.
Tracking Number 457456/ 44256920
1
467142/ 45564250
Version 6.0.55.3
Celerra Management Applications
Following are the fixed problems related to the Celerra Management Applications, that is, Celerra Monitor, Celerra Manager, WebUI, or Celerra MMC Snap-ins:
Severity 2 Symptom Description Celerra Monitor failed to start with error: "Component <serial_number> is not accessible to this user" Fix Summary Data types in Jserver have been modified to fit Symmetrix director configuration attributes without truncation The code has been modified to correctly handle filesystem names that included space characters. Jserver code was modified to stop producing problem reports when stale data was received from Common Cache. Jserver now records the record problem in the log file. Tracking Number 420801/ 41209492
2
Unisphere and Celerra Monitor did not display filesystem capacity and inodes utilization information for a filesystem name that included one or more space characters. A Data Mover failing over generated a very large number of problem reports which filled up the Jserver partition. This occurred due to a space issue under /nas/jserver partition.
432871/ 42832926
2
439763/ 11111111
EMC Celerra Network Server Version 6.0.70.4 Release Notes
129
Version 6.0.55.3
2
Numerous warning messages similar to the following in sys_log files: CS_PLATFORM:ADMIN:WARNING:18:::::d skMon[1245]: FS /dev/mapper/emc_vg_lun_5emc_lv_nas_var_emcsupport mounted on /nbsnas/var/emcsupport filling up (at 100%, max = 90%). Events were missing or the entire view was empty when viewing events from the Event Log Refreshing the view restored some or all of the events. More than one refresh caused events to be dropped or the view to empty. On very large (greater than 10 TB) MGFS filesystem after a large amount of data is migrated, the system started appearing unresponsive.
The code has been fixed to prevent the emcsupport partition from filling up. The issue that caused a background process to run infinitely has been corrected. The code has been fixed to display events correctly.
456533
1
462851
1
The code has been modified to initialize cg bitmap for MGFS filesystems to avoid unresponsive behavior.
455240 / 44242956
CIFS
The following entries describe fixed Common Internet File System problems:
Severity 2 Symptom Description In Windows Vista®, a new SID, OwnerRights, can be used in ACEs to control the rights on securable objects. The File OE did not support the new Vista OwnerRights SID. Data Mover panicked when Usermapper service received unexpected packet. Typically when security scanner sent invalid RPC to the usermapper service. An open file could not be forced closed on VNXe based storage systems. Fix Summary The code has been changed to support Vista SID. Tracking Number 368438 / 34831146
1
The code has been hardened to validate and reject incoming requests that are not formatted in the correct manner. The code has been modified to allow open files to be forced closed upon request. The code has been modified to add a parameter (cifs.admin.shareC_NotC reated)to hide C$. The default for the parameter leaves C$ visible.
404497/ 39107628
2
433118
1
All the storage system file systems were visible and accessible through the C$ share.
434264
130
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
Severity
Symptom Description
Fix Summary Disabling C$ visibility will not allow users to browse the different file systems from Microsoft MMC when a share is created.
Tracking Number
2
With Windows 7 and SMB2 client access, if a Word document had its archive bit un-set in DOS attributes, a modification in the Word document did not reset the archive bit. The Data Mover panicked when an application made a very large number of accesses to the same file from many different Windows Vista and Windows 7 SMB2 clients that used SMB2 file lock request operations. The message displayed is “Run out of free blocks in Buffer_CachedPool!" Users experienced a WERR_INVALID_PARAMETER error when attempting to configure shares from SQL clustered 2008 R2 server. This is because Celerra was rejecting the NetShareSetInfo request if the shi502_netname field was not populated. The command server_checkup generated an error in its report when executed on the FTPD test with a standard default configuration of FTPD. When the default persona of the FTP server is empty (no certificate and no private keys), then it generates the error “no available certificate for persona 'default'”. This error message is not correct. In quota reports, a discrepancy was observed in the reports shown in CLI and GUI quota screen for a approximately 234 users. This was in context with domain migration.
The code has been modified. Archive bit now works as expected with Word in Windows 7 and SMB2. The code has been modified to correct the memory leak in the SMB2 file locking mechanism.
434306/ 43096796
1
436698/ 43406710
2
If a netname is missing in a NetrShareSetInfo request, the code uses the share name from qualifier to prevent this error. The code has been modified. When the command is executed the validity of certificates in the persona is checked only if SSL is allowed, required, or enabled (that is,not disabled). The code has been fixed to avoid missing entries in the quotas report.
436846/ 42935986
2
440016/ 43715054
2
440448/ 43731696
EMC Celerra Network Server Version 6.0.70.4 Release Notes
131
Version 6.0.55.3
Severity 1
Symptom Description The Data Mover panicked when the Codenomicon/Defensics security scanner sent invalid SMB2 packets. The following message was displayed “>>PANIC in file: include/stream.hxx at line: 1130: datab:freeRef: db_ref <= 0” "The permissions are incorrectly ordered“ error message was displayed when the CIFS users opened the security tab for a file or folder. This was due to improper sorting of deny and access permissions by NFSv4 ACL. Windows 8 clients were unable to map EMC server shares. The net use command failed with system error 2148073478.
Fix Summary
Tracking Number
The code was fixed to protect against corrupted SMB2 commands. The code now rejects invalid commands and returns an error to the client. The code was changed so that NFSv4 ACL correctly sorts deny and access permissions.
453320/ 44042010
2
458223/ 44455658
2
The code has been fixed to support Windows 8 compatibility. This ensures that the Data Mover selectively checks an SMB2 request and signs the response even if the server was not configured to force signing. The code has been fixed. When access to share is checked, we return Read-Only if the file system is a snap and not a write-able snap. The code was modified to accept ACEs with SID 1-5-88 without returning an error. A new NT registry key HKEY_LOCAL_MACHIN E/Software/EMC/Widelin k/Global (Set per VDM) has been added as a choice if Widelinks is globally supported or per share using a new attribute. By default, HKEY_LOCAL_MACHIN E/Software/EMC/Widelin k/Global will be set to 1 which will result in keeping the existing
459810
2
A Windows user trying to create a new file (right click in Window Explorer) did not get a new option if the share they were working on was mounted on a network drive;file system is a writeable snap and client OS is Win7/W2K8 Copying files from MacOS X Lion to a Celerra CIFS share failed when the UNIX type SID had the form S-1-15-88. When using the Widelinks feature all shares for the CIFS server will have the DFS root attribute set, even those without any Widelinks under the share. This was found to cause an issue with Windows7 clients using offline caching mode because the DFS attribute affects which referral it selects.
461052
2
437759 / 43336414
2
438207 / 43020618
132
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
Severity
Symptom Description
Fix Summary behavior. When this key is set to 0 the Widelink attribute will need to be specially set on all shares containing Widelinks. The new Widelink attribute will be supported on share type (like for ABE and Offline cache). This attribute will have to be set using the type option on the server_export command: server_export -name myShare -option type=Widelink /myPath
Tracking Number
CORE
The following entries describe fixed problems related to Celerra Data Movers:
Severity 1 Symptom Description The Data Mover panicked when a file system mount point included special character "=". The root cause of the Dart panic is NULL fs volume ID (fsVolId). It returns from parsing Mount Command string with more than one pre-defined separator "=". The Data Mover panicked when a path larger than 256 characters is entered in a command. Fix Summary The code has been modified to handle special character “=”. 431762 456060/ 42742496 Tracking Number
1
The code was modified to handle paths greater than 256 characters in length.
436681/ 43287434
EMC Celerra Network Server Version 6.0.70.4 Release Notes
133
Version 6.0.55.3
Severity 2
Symptom Description Creating a checkpoint on the Control Station while the production file system was mounted on a Unix/Linux client did not allow that client to immediately cd into the .ckpt directory from the root of the file system, and/or all the checkpoints were not visible from the .ckpt directory at the root of the file system. Once the file system was unmounted and mounted, you could then cd to .ckpt, and all the checkpoints were visible from the root of the file system.
Fix Summary The code was fixed so that the checkpoints were assigned correct values when they were created, and all checkpoints and the .ckpt directory were visible to the client from the root of the PFS immediately after creation.
Tracking Number 455858/ No Clarify ID
Deduplication
The following entries describe fixed issues related to Deduplication:
Severity 1 Symptom Description The CIFS and NFS server became unresponsive and forced a reboot when using deduplication. The Data Mover panicked when Deduplication was scanning a file system containing a large number of hard linked files. The DART panic/fault message was “Page Fault Interrupt. Virt ADDRESS: 0x0000159274 Err code: 2 Target addr: 0x0000000000” Following a datamover panic and failover, users were unable to open deduplicated files. If a problem was found with the internal deduplication policy engine the state of deduplication would be set to Off. Under normal circumstances when deduplication is set off all files are recalled, however in this situation the state was set off but no recall was performed. This resulted in users being unable to access any deduplicated files. Fix Summary The deadlock was fixed by changing statistics logging. The logging caused a lock to be acquired in the wrong order. The memory leak was fixed. The code now releases allocated memory after the scan. Tracking Number 440939/ 43864948
1
455471/ 44220834
1
Deduplication will be suspended which means no new files will be deduplicated while still allowing access to any previously deduplicated files.
457093 / 43892970
134
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
File Systems
The following entries describe fixed File System problems:
Severity 1 Symptom Description The Data Mover panicked while creating a Read/Write check point filesystem. The message shown was: “PANIC:>PANIC in file:../addrspac.cxxat line: 529:” A Data Mover could deadlock if FS checkpoint or any other tasks that needed FS pause while iSCSI LUN/snap deletes were going on in the background. Fix Summary The code was modified to correct a memory leak in file system initialization. Tracking Number
436316
1
Locks were acquired in the proper order to avoid the deadlock situation.
459827/ 44694588
iSCSI
The following entries describe fixed issues related to iSCSI LUNs:
Severity 1 Symptom Description iSCSI LUNs that resided on the same filesystem were erroneously reported to the iSCSI initiator with the same WWN when the file system was unmounted on boot-up. If the file system issue was corrected and the file system was remounted, the iSCSI LUNs were still reported to the initiator with the same WWN until the Data Mover was rebooted. Fix Summary The code has been fixed so that the iSCSI service will not register a WWN for iSCSI LUNs whose file system is un-mounted and will not report those LUNs to the iSCSI initiator. Once the file system issue is corrected and the filesystem is remounted, the iSCSI service must be restarted by running the command "server_iscsi server_x -service -stop" and then the command "server_iscsi server_x -service -start" -or- by rebooting the Data Mover.After the iSCSI service is restarted, it will compute a proper WWN for the affected iSCSI LUNs and report them to the iSCSI initiator. Tracking Number
453767/ 44110140
Installation, Configuration, and Upgrades
The following entries describe fixed issues related to installations, configurations and upgrades:
EMC Celerra Network Server Version 6.0.70.4 Release Notes 135
Version 6.0.55.3
Severity 1
Symptom Description Earlier releases used ConnectEMC 2.6 software to deliver callhome messages to EMC. The outgoing mail functionality of this version issued duplicate HELO/EHLO commands. This was in accordance with RFC 2821, however, some mail servers do not support this (especially older mail servers). Biweekly health checks ran for a very long time and consumed a lot of CPU cycles if "server_checkup" reported more than 40 errors or warnings for one Data Mover.
Fix Summary
Tracking Number
This release uses ConnectEMC 3.0.1, which addresses the callhome issue.
377477/ 378198
1
The health check code has been modified to avoid this situation.
461161 / 44639742
NFS
The following issues describe fixed issues related to the Network File Server:
Severity 1 Symptom Description The Data Mover panicked when the NFSv4 client was revoked and requests to confirm its callback path had not been cancelled. The message displayed was “path is unavailable(frozen) or invalid” Fix Summary The code was modified to eliminate the race condition by making a check to verify the callback path has been cancelled before the NFSv4 revocation takes effect. The signature verification processing code was fixed to return the correct response. The code has been fixed to return the correct error status (OLD_STATE_ID). The code has been fixed. This flag will be correctly translated in NFSV4. Tracking Number
434416
1
The signature verification on a secure NFS request to delete the session security context was incorrectly processed, resulting in an auth_error response to the client. In the case of a Mac OS X client, the error caused the client to go into a rapid loop of creating and deleting additional sessions, which severely affected the Data Mover performance. When using NFSv4, an unexpected error status was returned by the server (BAD_STATE_ID), making the file unaccessible. After running chmod for files with NFSv4 ACL setting, inherited ACL settings got lost. INHERITED flag was dropped by NFSV4 ACL manipulations, because it was not returned to the client.
453465/ 43760246
1
456022/ 44309440 453335 / 43875390
1
136
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
1
Loss of NFS access after changing special mode bits (sticky, suid etc ...) from NFSv4. This issue is only possible when Windows security auditing is enabled and the file system is mounted using Mixed access policy. The server log will show messages like: KERNEL: 3: 3: ThreadsServicesSupervisor: Service:NFSD Pool:NFSD_Exec BLOCKED for 361 seconds: Server operations may be impacted.
The code has been modified to eliminate the potential deadlock when changing special mode bits (sticky, suid, and so on) from NFSv4.
455710 / 44297386
Replication
The following issues describe fixed issues in this release related to Celerra Replicator, SnapSure, and Checkpoint:
Severity 1 Symptom Description When the Data Mover panicked during a replication, the create BLO reported success to the user without writing the session signature to the sav volume. This later caused a problem for stop/start BLO, when a session was lost. The create remote replication command failed when a session on the source had the same FSID as a session on the destination. When a checkpoint was restored, it created a temporary checkpoint and snapped restored data. If there was not enough space, the restore failed leaving the filesystem inconsistent and replication inactive. The replication stop operation would attempt to restore a checkpoint to bring the destination to a consistent state. The restore operation failed as the destination savvol was full and could not be extended. Fix Summary The code has been modified to attempt to re-write the session signature after a panic interrupts a replication. If the re-write fails, the BLO will be rolled back and a CCMD message will be issued to the user. The code has been modified to reject the creation of a new session between a source and destination that is not unique. The issue has been fixed. When replication aborts and tries to restore the PFS, the special restore , which will not create temporary checkpoint and not allocate any other chunk, is launched. Tracking Number
426987
2
431722/ 42744966
1
398615 / 38346418
EMC Celerra Network Server Version 6.0.70.4 Release Notes
137
Version 6.0.55.3
1
A queued ReplicationV2 session checkpoint operation (create/delete/refresh/restore) on the destination experienced long delays when the SavVol needed to be extended in order for the transfer to complete. DART panicked with the following panic header: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 820db2 Err code: 0 Target addr: 0 ** DART panicked with the following panic header: DART panic/fault message
The code has been changed to allow extends to continue, And, to reject any queued checkpoint operation requests.
416453 / 40756596
1
The code has been fixed to avoid the DART panic. 421194 / 41485422
1
The code has been fixed to avoid the DART panic.
439878 / 43748484
UFS
The following entries describe fixed UNIX File System issues:
Severity 1 Symptom Description The Data Mover panicked due to an out of memory condition under a high work load and with many active file systems. The condition was caused by Cylinder Group (CG) buffers being held in-use and remaining indefinitely memory resident. The following message was displayed “>>PANIC in file: ../addrspac.cxx at line: 289 : Out of memory” The Data Mover panicked with the following message: DART panic/fault message: >>PANIC in file: ../ufs.hxx at line: 1377 : UFS hashRemove: not on hash chain Data Mover panicked with DART panic/fault message: Assertion failure: 'this->back == this' on line 551 in file ”../vfsnode.cxx” >>PANIC in file: ../assert.c at line: 27 : assert failed Fix Summary Tracking Number
The CG caching algorithm code was changed to more efficiently handle the aging of CG buffers. The code has been modified to avoid the panic
394853/ 37939478
1
421397 / 41500892
1
Fixed the race condition on the mount point when file system was being unmounted and another thread is still referring the mount point.
434499 / 43149914
138
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.55.3
1
Watchdog panicked while enumerating file-system stats using server_stats command.
The code has been modified to avoid the watchdog panic. The code has been fixed to release this lock in the error path. The code was changed to correct invalid free block recoveries at the time of Data Mover recovery, avoiding further Data Mover panics.
453141 / 43990760
1
When the system is short of memory, a rwlock may not be released in the error path if the I/O is in the uncached IO path. This causes NFS threads to hang. The Data Mover panicked and reported that a file system requires an File System Consistency Check (FSCK). The panic may report a corrupted directory entry or an invalid indirect block.
461028 / 44875620
1
466190 / 45446328
Control Station
The following entries describe fixed problems related to the Celerra Control Station:
Severity 1 Symptom Description Slow overall system performance was observed with nas_pool -query being issued from JServer. The poor performance could be seen in several different ways including RDF continuously failing over, slow GUI response times, certain task starvations, etc “dbchk –p” command failed when executing "symmir list", and gave out error “Error: File system with id /nas/symcli/bin/symmir list is found in .server_config mountdisplay's output but does not exits in /nas/server/{1}/mount file.” The error doesn't describe the actual problem. Fix Summary The code now provides a number of performance optimizations to improve performance. Tracking Number 422395 / 41595948
2
The error message was replaced with ”The exit status is neither “0” nor “1” while running command symmir list.”
454984 / 44202128
EMC Celerra Network Server Version 6.0.70.4 Release Notes
139
Version 6.0.55.3
Network
The following entries describe fixed Network-related problems:
Severity 1 Symptom Description BCM 5715 1GigE NIC ports intermittently showed link bounce. Fix Summary Changed the default behavior of the BCM NIC recovery mechanism to not reset the 5715. Tracking Number 432345 / 42831902
System Management
The following entries describe fixed issues related to system management:
Severity 2 Symptom Description When control station is in BRST timezone, if schedule fire-off time falls into the transition time of BRST to BRT, scheduler operations get hung. Fix Summary The code has been fixed. Scheduler no longer hangs. Tracking Number 436472 / 43375724
Quotas
The following entries describe fixed issues related to quotas:
Severity 1 Symptom Description When a CIFS client issues quota on while a online quota check task is running, the Data Mover panicked with: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x000083f4bc Err code: 2 Target addr: 0x0000000000 ** While logging a server log message indicating that an online check was already in progress, the code was dereferencing a null pointer Fix Summary The condition is now handled correctly. Tracking Number 453514 / 44008566
140
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Version 6.0.51.6
Celerra Management Applications
The following are fixed problems related to the Celerra Management Applications, that is, Celerra Monitor, Celerra Manager, WebUI, or Celerra MMC Snap-ins:
Severity 2 Symptom Description Following a Data Mover failover, VDM CIFS servers were not listed in Unisphere. They were visible through the CLI. Persistent failures of APL tasks over time, due to faulty logic validating UPS configuration (expecting 2 UPS configuration when only one exists) resulted in accumulation of large amount of tasks related files in /nas/tasks directory. Eventually 100% of file descriptors were exhausted. NBSnas becomes non writable, and nas services will fail to start. In Unisphere, when managing the Usermapper configuration and uploading "User Mapping File" or "Group Mapping File", or both with the "replace" option caused one of the files contents to be lost. As the result Windows users may lose access to their files. Automatic DHSM connection creation and connection verification for Rainfinity FMA using XML API failing due to case sensitivity of share queries. Users could not query the location if they didn't specify the share names in the actual case. Fix Summary CIFS servers created on a VDM are now visible in Unisphere after a Data Mover failover/failback. The logic has been corrected and these APL task failures will no longer occur. Tracking Number 406031 / 39311966
3
411571 / 40109354
2
Modified Mapping Files processing algorithm to ensure that using the "replace" option does not cause loss of the contents.
412850
2
The share name comparison have been changed to case insensitive.
422160
EMC Celerra Network Server Version 6.0.70.4 Release Notes
141
Version 6.0.51.6
Severity 1
Symptom Description When there was a large number of objects in user and group quota pages, user experienced delays in screen painting. The screen refreshes itself several times, and each refresh causes significant delay which appears like a hang to the user. User and group names were not displayed on the user and group quota page. Instead user and group IDs were displayed. Jserver reported incorrect port WWN for SP ports when a setup has more than 8 FC FE ports per backend. Creation of a new notification in Unisphere fails when e-mail address for notification contains '&' character. The error messages says: "Celerra Monitoring Agent could not be contacted". When using NTP QFS software with Celerra, Celerra Manager login page failed to load. That problem disappeared after a Control Station reboot, however returns back after several days. This was due to QFS software being disconnected abruptly and leaving HTTP socket open. Unisphere showed only one event starting 31st October 2011. The alert_log file was checked, it was normal and contained multiple entries. The Schedule Name column was not visible in the view even if the SnapSure license was enabled in Unisphere.
Fix Summary Fix involves some design changes in the way tables are built and updated.
Tracking Number 396840 / 37832254
2
Changed the APL to support the correct schema reporting of the SIDs value for user and group quotas. The fix is to hash ports info based upon the entire port WWN instead of just the "port number". The code now properly handles special characters, like '&', '<', '>', etc. when included in the e-mail address for notification. Improved code that handles abruptly disconnected XML API client. Introduced a limit on the number of outstanding XML API requests.
397447
2
420482 / 40714894
2
432113 / 42573498
2
431655 / 42590566
2
The APL filtering error that was causing this issue has been corrected. Changed the string concatenation operations to ensure that when "snapsure" license is enabled, user
454109 / 44118074
1
454360
142
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description
Fix Summary can see the Schedules column.
Tracking Number
CIFS
The following entries describe fixed Common Internet File System problems:
Severity 1 Symptom Description If a share path was being used by another share command while the path was being relocated, the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 359: Memory corrupted.. The stack trace showed:
0xe94ff740: 0x141278 waitForReboot+0x90 0xe94ff760: 0x141501 fault_dump+0x67 0xe94ff780: 0x1413e9 PANIC+0x29 0xe94ff790: 0x200731 free+0x6f 0xe94ff7c0: 0x200e34 _ZdlPv+0x10 0xe94ff7d0: 0x8aecce _ZN12volatilePathIcED0 Ev+0x2e 0xe94ff7e0: 0xc89ff2 _ZN14abstractStringD2E v+0x20 0xe94ff7f0: 0xc8a13d _ZN10pathStringD0Ev+0x 15
Fix Summary Code was modified to ensure that the share path is not being used or modified when the path is relocated.
Tracking Number 431284 /42686198
EMC Celerra Network Server Version 6.0.70.4 Release Notes
143
Version 6.0.51.6
Severity 1
Symptom Description When the virus checker service was restarted with new masks and exclusion rules, these changes were not applied to files that were open. This resulted in the virus checker scanning files with an excluded extension. When there were several open resources on a Domain Controller in trusted domain which no longer responded, each SMB thread entered a 20-second timeout, potentially creating a CIFS outage. During CIFS access outage, the server_log was logging "DOMAIN_CONTROLLER_NOT_F OUND" events. In addition to this, the Data Mover was sending MSRPC Close requests when the TCP connection was already dropped. When a Data Mover requested user’s group memberships, a Domain Controller, for some unknown reason, had not sent the primary group of a user. This resulted in access being denied to files that rely on group membership as the group membership was not stored in internal credential. File system outage occurred when enumerating CIFS quotas with slow mapping mechanisms. User lost access to his shares while the quota entries window was populating CIFS quotas window. When certain kinds of malformed CIFS/SMB or NFS packets were sent to a Data Mover’s CIFS/NFS server or that of a Celerra, the Data Mover panicked, resulting in a denial of service for legitimate users. One example of this was seen when Codenomicon's Defensics v3.12 (www.codenomicon.com)
Fix Summary Code now checks whether the virus checker configuration has changed before performing a scan of a file.
Tracking Number 435234 /43216602
1
The code has been fixed and Data Mover will no longer cause CIFS outage under such conditions. The Data Mover will no longer send unnecessary close requests when connection is already dropped.
372385 /35027576
1
The fix adds missing primary group to internal credential to prevent this condition.
407896 / 39421448
2
Code was modified to reduce the file system locking time when enumerating users' quotas from CIFS. Improved handling of invalid data being sent in fields within an NFS or an SMB2 request.
412335 /39230656
1
414090
144
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description vulnerability scanner was run against a Celerra Data Mover.
Fix Summary
Tracking Number
1
With widelinks enabled and many CIFS accesses to a share from one client, the Celerra CIFS server incorrectly returned STATUS_OBJECT_PATH_NOT_F OUND for a request for a valid path. This was due to a race condition where multiple threads were checking a variable to determine whether it was a widelinks connection.
The check for whether it is a widelinks connection is now done only once during tree connect or mount of a file system and not on each protocol request which was leading to the race condition.
414865 / 40421472
EMC Celerra Network Server Version 6.0.70.4 Release Notes
145
Version 6.0.51.6
Severity 1
Symptom Description The Registry.pol file is used to maintain certain GPO settings on a Windows domain controller. User had a (possibly corrupted) Registry.pol file where fields (in a single entry) were all zeroes, so the pointer to the data field was set to NULL. When the NULL data pointer was subsequently dereferenced, the Data Mover panicked with the following DART panic/fault message: ***
Page Fault Interrupt. Virt ADDRESS: 0x0000f76e6a Err code: 0 Target addr: 0x0000000001 **. The stack trace
Fix Summary The data length field value is now checked (to be greater than 0) before dereferencing the data pointer.
Tracking Number 422696 / 41650866
showed:
0760: 0xc1ad80 _ZN7DskDump9checkdumpEPKc+0 x7a 07a0: 0xc1af8e dump_return+0x4 07b0: 0x14c7d2 _ZN12Memory_Frame13setProte ctionEm19Memo ry_Prot 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x14162d handle_fault+0x1d 60d1b0: 0xf77f2d _ZN6SysVol19getRegistrySett ingsERK6Strin gRP8L 60d320: 0xf7beb0 _ZN6SysVol14getGpoSettingsE RK6StringRP12 gpo_s 60d770: 0xf52346 _ZN9DomainGpo18updateVdmGpo ServerERK6Str ingi+=
2
One of the Group Policy settings used by the Data Mover is the Maximum machine account password age. When this setting is defined or changed in the domain,
The code was changed so that the default setting (0) is applied to the applicable CIFS Servers, when the
418168
146
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description it is updated and used by each CIFS server. If the setting was set to not defined, the applicable CIFS servers were not updated with the new setting value until CIFS was restarted.
Fix Summary setting is changed to no longer defined in the domain.
Tracking Number
1
When an administrator tried to list open files the Data Mover panicked with the following DART panic/fault message: ***
Page Fault Interrupt. Virt ADDRESS: bc9a8e Err code: 0 Target addr: 24 **. The stack trace
Fixed the race condition that occurred when an enumeration of open files is being performed at the same time as an open file request is failing.
419382 / 41240528
showed:
0x2c3b91c: 0x14124c waitForReboot+0x90 0x2c3b93c: 0x1414d5 fault_dump+0x67 0x2c3b95c: 0x13a398 handle_fatal_fault+0xa 0x2c3b980: 0xb585cb _ZN8fid_info16getFileEnum3I nfoEP12_FILE_ INFO_3+ 0xee897970: 0xb58970 _ZN14smb_connection11srvsvc AuditERiP16cS rvsvcF 0xee897a10: 0xb5837d _ZN14smb_connection11srvsvc AuditERiP20cS rvsvcA 0xee897a30: 0xb581d2 _ZN13smb_streamCtx16srvsvcA uditParseEP20 cSrvsv 0xee897a70: 0xb57f8f _ZN10smb_server16srvsvcAudi tParseEP20cSr vsvcAu
2
The Data Mover allowed the user to zero out (truncate) a file even though he did not have any rights to do so.
If there is no write permission from the ACL, the open now fails and the file is not truncated.
422369 / 41403684
EMC Celerra Network Server Version 6.0.70.4 Release Notes
147
Version 6.0.51.6
Severity 1
Symptom Description If a file system became full, when user cleared the windows log file, the Data Mover panicked with the following DART panic/fault message: >>PANIC in file:
../sched.cxx at line: 154 SYSTEM WATCHDOG.
Fix Summary This issue occurred because the memory state of the log object was not correctly restored. This has been fixed, and the Data Mover will no longer panic under this condition.
Tracking Number 425818 / 42018528
The stack trace showed:
f3a0: 0x14dd29 system_watchdog+0x61 f3b0: 0x1407b3 t_nmi+0xc3 f468: 0xa653db _ZN11BasicVolume18completio nCallbackEP10 ccb_hea f4a8: 0xae10b1 _ZN14ScsiBus_Device12sim_fi nishIOEP14Scs i_IOReq f4b8: 0xae1348 _ZN14ScsiBus_Device10sim_fi nishEP14Scsi_ IOReque f4d8: 0xaede04 _Z14FCP_completeIOPv+0xf2 f528: 0xb4db26 cpd_osw_finish_srb+0xb0 f588: 0xb46b8d fc_exm_exchange_complete+0x fa
1
While using FileMover with many files (in this case, 873), an Arkivio server returned an invalid 0xFFFF value for File ID, which led to a Data Mover panic with the following DART panic/fault message:
>>PANIC in file: /disks/Archive_Area_nsg7/6_ 0_41_303/Dart/s erver/src/cifs/client/inclu de/cifsclient_n ode.hxx at line: 281 : getting size of an unopened file. The stack trace
This issue was fixed by changing the default value (for unitialized FID) from ushort 0xFFFF to ulong32 0xCC11FFFF.
426242 / 42082680
showed:
f21400: 0x141439e _ZN17DHSM_CIFS_RemoteP15rem
148
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description
oteBlockReadEjR 11 f21590: 0x13f374a _ZN10DHSM_Vnode15remoteBloc kReadEjR11File_ St f21610: 0x13c7c67 _ZN10MGFS_Vnode14migrateToB lockEjRiPP13Buf fe f21890: 0x13ed849 _ZN10DHSM_Vnode14migrateToB lockEjRiPP13Buf fe f218e0: 0x141169b _ZN10DHSM_Vnode10blocksRead EjjRiPP13Buffer _C f219d0: 0x13f1c94 _ZN10DHSM_Vnode9blockReadEj R11File_Status+ 0x f21a00: 0x7bc897 _ZN9File_Fork17readBlockInt ernalEjRP13Buff er_ f21a40: 0x7c70dc _ZN13File_OpenFile16readDat aInternalEyjR11 Fil
Fix Summary
Tracking Number
1
A page fault PANIC might occur when changing quota state to different values (ON/OFF) from both Windows and from Control Station side at the same time for the same file system. When an SMB2 client opened a file under a Quota Tree, then disconnected just before the file system unmount processing started, a file system unmount did not complete, which led to a loss of access.
A lock on quota objects was added to prevent this condition.
426890 / 42139428
1
the unmount processing was modified to properly clean up SMB2 durable handles on a Quota Tree.
427655 / 42232932
EMC Celerra Network Server Version 6.0.70.4 Release Notes
149
Version 6.0.51.6
Severity 2
Symptom Description CIFS CEPP events were not generated or incorrectly generated. This caused some third party applications to not work properly. When user applied a strong security policy to his W2K8R2 CIFS domain, FileMover, CAVA or CEPA on this Microsoft machine may fail with ACCESS_DENIED errors due to communication errors with the domain controller. When CIFS client application used file range lock operations intensively with conflicts between different clients, a race condition led to a Data Mover panic with the following DART panic/fault message: *** Page Fault
Interrupt. Virt ADDRESS: 0 Err code: 0 Target addr: 0 **.
Fix Summary Code was modified to better manage CIFS CEPP events. Code now uses the correct CIFS-based SPN (service principal name) to access the domain controller.
Tracking Number 386670
2
425859 / 41925924
1
This race condition between range lock and close operations has been eliminated.
429671 / 42298318
The stack trace showed:
0x2c3bbd0: 0x14124c waitForReboot+0x90 0x2c3bbf0: 0x1414d5 fault_dump+0x67 0x2c3bc10: 0x13a398 handle_fatal_fault+0xa 0x2c3bc50: 0x768586 _ZN14File_LocksData6umountE y+0xc0 0x27de1c30: 0x7681ef _Z12lockd_umounty+0x123 0x27de1c70: 0x72dafe _ZN15unmountCallBack7execut eEy+0x20 0x27de1c90: 0x72e8a3 _ZN14VFS_FileSystem10unmoun tNowEi+0x4d 0x27de1cb0: 0x72ebb5 _ZN14VFS_FileSystem9doUnmou ntEv+0x109
1
Concurrent command to start usermapper, particularly when the database was disabled, caused a Data Mover panic with the following
Modified code to protect Usermapper database initialization from concurrent start.
430768 / 42669876
150
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description DART panic/fault message: *** GP exception. Virt ADDRESS: 0x00012c5c2a. Err code: 0 *** . The stack trace showed:
6760: 0xbf96c0 _ZN7DskDump9checkdumpEPKc+0 x7a 67a0: 0xbf98ce dump_return+0x4 67b0: 0x14c7d2 _ZN12Memory_Frame13setProte ctionEm19Memory _Prot 67d0: 0x14ce5d fault_dump+0x89 67f0: 0x13378d handle_fault+0x1d a57b0: 0x12323c3 __db_c_close+0x35 a57f0: 0x123271d __db_c_close+0x38f a5840: 0x1233169 __db_c_get+0x7a3
Fix Summary
Tracking Number
2
Whenever a new user or group quota was created, it didn't show up in the appropriate Unisphere quota list until an internal refresh. Unable to access CIFS share when trying to rename a file that was waiting to be checked by virus checker and all VC servers were offline. New users could not be mapped via LDAP.
Code was fixed to ensure new user or group quotas are displayed when created. Code now checks if the client has disconnected before renaming a file using CIFS. If a current LDAP configuration is deleted and a new configuration is created, CIFS code now automatically switches to the new LDAP setup.
378714
1
414744 / 40668160
2
414760 / 40449194
EMC Celerra Network Server Version 6.0.70.4 Release Notes
151
Version 6.0.51.6
Severity 1
Symptom Description During a share enumeration, if the share name contained 118N characters that generated a NULL hash key, renaming the share could cause a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x00011ef9c4 Err code: 0 Target addr: 0x0000000000 **. The stack trace showed: 0760: 0xc1ac70 _ZN7DskDump9checkdumpEPKc+0x7a 07a0: 0xc1ae7e dump_return+0x4 07b0: 0x14c7d2 _ZN12Memory_Frame13setProtectionEm1 9Memory_Prot 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x14162d handle_fault+0x1d cbb7d0: 0xc3e590 _ZN11smb_sdShare8getDBKeyEPKt+0x1a cbb820: 0xc37564 _ZN11smb_sdShare8getDBKeyEv+0x54 cbb840: 0xc377f3 _ZN9shareList9getLockedERP9smb_share PKcP9xdrS Samba clients were unable to use Kerberos authentication. The following messages were found in server_log: SMB: 3:
SSXAK=RPC_NT_INVALID_BINDIN G Client=a.b.c.d origin=550 stat=0x40000,0
Fix Summary
Tracking Number
Shares having a NULL hash key are now handled correctly, and the Data Mover will no longer panic under this condition.
418870 / 41139864
2
Code has been fixed to enable support for CKSUMTYPE_HMAC_ MD5 checksum type.
427779 / 42247186
They indicate that Kerberos checksum type cannot be checked due to usage of CKSUMTYPE_HMAC_MD5 checksum type instead of regular Windows client's CKSUMTYPE_KG_CB=0x8003. As a result, Kerberos authentication cannot complete, but client falls back to NTLM authentication seamlessly.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
152
Version 6.0.51.6
Severity 1
Symptom Description Memory was not correctly released after a DIC sending error. This memory leak led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../bufpool.cxx at line: 1309 : couldn't get a free page. The stack trace showed: 08e90: 0x1ff58e _ZN18Buffer_Cached_Pool17makeCached BuffersEm+0 08ee0: 0x14497f _ZN19Memory_AddressSpace16makeCac hedBufferEv+0 08f10: 0x1ff631 _ZN18Buffer_Cached_Pool12findAfreeBuf Emiii+0x6 08f50: 0x1ff6b9 _ZN18Buffer_Cached_Pool15getAnyCach edBufEmiii+ 08fa0: 0xa8f082 _ZN17UnCachedIO_Object9readBlockEyj1 6BlockIO_P 09060: 0x8ba443 _ZN14UFS_FileSystem9readBlockEiiRP13 Buffer_Cac 090c0: 0x8b4550 _ZN9UFS_Vnode5breadEjR11File_Statusi +0x352 09130: 0x92a676 _ZN9UFS_Vnode9blkatoff2EyRPvRP13Bu ffer_Cachedi
Fix Summary Memory allocated for the issue is now released once it has been used.
Tracking Number
432293 / 42848676
EMC Celerra Network Server Version 6.0.70.4 Release Notes
153
Version 6.0.51.6
Control Station
The following entries describe fixed problems related to the Celerra Control Station:
Severity 2 Symptom Description Administrators with a space in their ID could not process Join/Unjoin requests. Celerra replaced spaces with an asterisk which rendered the admin ID invalid Following a NAS code upgrade, when trying to query the status of SNMP on the Control Station, the system was returning the following error:
nasadmin]# /sbin/service snmpd status snmpd dead but subsys locked /sbin/service snmpd stop Stopping snmpd: [FAILED]
Fix Summary Corrected code to ensure that spaces embedded inside 'admin' are handled correctly. Code has been fixed to save the old config files and properly update the new config files during upgrade.
Tracking Number 406140
2
409494 / 39688446
1
After a power outage the CIFS servers came back as NT4 type instead of Windows 2000 type.
Code has been fixed to check for an existing server entry with the same name as the new server entry. If found, the existing entry and the new entry get merged to create a new entry. If the new and existing entries differ in server type, or domain, or netbios (W2K only), code now replaces the new entry with the existing one. The Control Station no longer ignores these event dispatch
381939 / 36515374
1
The Control Station ignored the event dispatch policies of the CLARiiON, causing a
415637 / 39816782
154
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description very high amount of call homes which led to the flooding of the call home utilities and servers.
Fix Summary policies.
Tracking Number
CORE
The following entries describe fixed problems related to Celerra Data Movers:
Severity 2 Symptom Description When issuing the warm reboot, if the Data Mover panicked for any reason, Data Mover failover occurred. Fix Summary Code has been fixed to ignore any unexpected events during warm reboot PreShutdown operation and not fail over. Tracking Number 403054
Deduplication
The following entries describe fixed issues related to Deduplication.
Severity 1 Symptom Description The deletion of deduplicated files while open transactions were in progress led to a deadlock situation where the UFS transaction log was unable to make progress. DART panic/fault message: >>PANIC in file: ../ufsvfsop.cxx at
line: 278 : sync: no progress reducing dirty list. The stack trace showed: 0x2752bf40: 0x141278 waitForReboot+0x90 0x2752bf60: 0x141501 fault_dump+0x67 0x2752bf80: 0x1413e9 PANIC+0x29 0x2752bf90: 0x7dcb25 _ZN21UFS_SyncThreadInitial5startEv +0x17f 0x2752bfe0: 0x145c26
Fix Summary Changed the way file system deletes compressed data produced by deduplication to prevent this panic.
Tracking Number 422705 / 41604572
Sthread_startThread_internal+0xe
EMC Celerra Network Server Version 6.0.70.4 Release Notes
155
Version 6.0.51.6
Severity 1
Symptom Description When using alternate data streams on a Deduplication-enabled file system, removing a deduped unnamed stream file caused the Data Mover to panic with the following DART panic/fault message: isAValidIndBlkBuf: bad bn. The stack trace showed: 0x21e47b40: 0x141280 waitForReboot+0x90 0x21e47b60: 0x141509 fault_dump+0x67 0x21e47b80: 0x1413f1 PANIC+0x29 0x21e47b90: 0x784dd7 _ZN14UFS_FileSystem28markCorruptedFs ForU nmount 0x21e47d50: 0x79b64a _ZN14UFS_FileSystem17isAValidIndBlkBuf EP 13Buff 0x21e47dd0: 0x7915c3 _ZN9UFS_Vnode17readIndirectBlockElRP 13Bu ffer_C 0x21e47e00: 0x791d25 _ZN9UFS_Vnode4bmapElRlPiS1_+0x237 0x21e47e60: 0x10366f0 _ZN10DHSM_Vnode14itruncInternalERP20 UFS_ Update User reported poor performance and high CPU utilization during a deduplication scan on file systems which contained thousands of hard linked files.
Fix Summary Code has been fixed to ensure blocks are flushed to disk properly before files are removed or truncated.
Tracking Number 432149 / 42837196
2
Replaced linked list with hash table to significantly improve scan times on file systems containing thousands of hard linked files. This new code will allow the migration inode to be reingested even if the UID/
424654 / 41006630
1
User was unable to read deduplicated files. The following error was displayed in the server_log: corrupt mignode for offline inode x. This can happen if the stub file and migration inode uid / gid is out of sync.
421654 / 41472302
156
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description
Fix Summary GID is out of sync so that stub file read operations are possible. It will also try to resync the uid / gid on file system update operations such as write, setLength, or recall.
Tracking Number
1
Compressing files caused their ctime to be updated which triggered unnecessary activity for incremental backup even though neither the data nor the file meta-data had been changed from the users perspective.
Code was modified to ensure deduplication and reduplication do not impact the ctime for either the internal Dart PE scan or for CIFS client compression / decompression.
424721
FileMover
The following entries describe fixed FileMover related problems:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
157
Version 6.0.51.6
Severity 1
Symptom Description While doing read operations using FileMover, threads hung and prevented CIFS access when the secondary storage became unavailable. Group quota usage is incorrect after recalling file from secondary file server.
Fix Summary During read operations, code was modified to require a shared lock only under this condition which will prevent hung threads.
Tracking Number 417842 / 41062586
1
This issue occurred because of a typo in DHSM feature that typed a GID as UID. This quota usage issue has been corrected.
440598
File Systems
The following entries describe fixed File System problems:
Severity 1 Symptom Description Write performance degraded substantially on a file system with checkpoints compared to a file system without checkpoints. Performance degradation was caused by inefficiencies in checkpoint handling code path. Fix Summary The code has been optimized to improve write performance on file systems with checkpoints. Code has been fixed to properly use the lock and prevent checkpoint from being falsely marked corrupted under this condition. This issue occurred Tracking Number 429333
1
Following an upgrade to version 6.0.42.0, when reading a deduplicated file for the first time and creating/refreshing a checkpoint, the checkpoint was marked as corrupted and unmounted. This was due to improper lock handling and creating a checkpoint with a dirty flag while there was a superblock update on the production file system.
429662 / 42529536
1
158
Migration between CFA and FMA was failing
382368 /
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description for a file system with Conversion file is different than an existing one error observed in the server log.
Fix Summary because the file system was replicated from an older system while a conversion was in progress. This caused the new FS to be marked as "conversion active", preventing the new conversion from failing. Code has been fixed to properly handle this condition and finish the conversion on the new destination automatically. The code has been fixed so that server_df is now able to display the correct number of free inodes regardless of the nbpi setting.
Tracking Number 36303236
2
The server_df server_x -inode command reported a negative a number of free inodes if the file system was created with lower than the default setting for nbpi (number of bytes per inode). This was because server_df was not able to report more than 2G free inodes.
376435 / 35689866
EMC Celerra Network Server Version 6.0.70.4 Release Notes
159
Version 6.0.51.6
Severity 1
Symptom Description When accessing a large directory (in this case 1.25 GB) and storing hash of the directory into memory, code did not properly release the memory which led to a Data Mover panic with the following DART panic/fault message: >PANIC in file: ../bufpool.cxx at line: 421 : couldn't get a free page. The stack trace showed: 0x79d0f8c0: 0x140ce4 waitForReboot+0x90 0x79d0f8e0: 0x140f69 fault_dump+0x67 0x79d0f900: 0x140e51 PANIC+0x29 0x79d0f910: 0x1c6af9 _ZN18Buffer_Cached_Pool17makeCached BuffersEi+0 0x79d0f940: 0x13c9c7 _ZN19Memory_AddressSpace16makeCach edBufferEv+0 0x79d0f960: 0x1c669c _ZN18Buffer_Cached_Pool15getAnyCache dBufEjii+0 0x79d0f990: 0x77e8eb _ZN14UFS_FileSystem8getBlockEliRPv+0x 49 When user tried to rename a file without write access into target directory, the rename failed with "Access denied". When this occurred, some CEPP events were generated twice for some NFS V3 operations. NFS lock was not preserved upon a Data Mover failover.
Fix Summary Code has been fixed to properly reclaim memory under this condition and prevent the panic.
Tracking Number 383034 / 36705796
2
Code was modified and will only send the CEPP PostEvents once.
491155
2
Code now preserves the statd registry when file systems are unmounted by administrator. If the client is 4.0 we return only 4.0 attributes even if 4.1 attributes are requested.
407035 / 39509656
1
Some version of AIX clients request NFSv4 4.1 attributes with a 4.0 mount. This caused AIX 6.1 host to crash when trying to mount a file system using NFSv4.
409247 / 39816612
160
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity 1
Symptom Description After running FSCK on a deduplicated file system, some files were inaccessible.
Fix Summary Improved FSCK handling of deduplicated files containing Alternate Data Streams (ADS). Code has been fixed to properly handle NFSv4 owner sequence and prevent this condition. The code has been optimized to improve write performance on file systems with checkpoints.
Tracking Number 410661 / 39762892
2
When using NFSv4, if a client sends a bad StateId this could result in NFSv4 errors, forcing the user to umount and remount NFSv4 file systems to get access again.
419732 / 40998286
1
Write performance degraded substantially on a file system with checkpoints compared to a file system without checkpoints. Performance degradation was caused by inefficiencies in checkpoint handling code path.
429333
EMC Celerra Network Server Version 6.0.70.4 Release Notes
161
Version 6.0.51.6
Severity 1
Symptom Description When using NDMP, a file was truncated when backing up the R/W file system on which the deduplicated file resided. The last coming data size exceeded the bytes in the last block and caused a Data Mover to panic with the following DART panic/fault message: Memcpy size 0xfffff28e itoo large. The stack trace showed: 909a68: 0x25147e memcpyPanic+0x42 909ac8: 0x282578 memcpy+0x54 909b40: 0x140009a _ZN19File_OpenDedupeFile11writeStreamE PhjRj+ 909bc0: 0xbf32d9 _ZN15PaxDedupHandler11WriteStreamEyP cjRj+0xaf 909c10: 0xbf469d _ZN14pax_cmd_thread13dedup_restoreEP 5ARCHDP15 909d50: 0xbafcc0 _ZN14pax_cmd_thread7extractEP5Event+0 x31a0 909f40: 0xbb1004 _ZN15restore_Initial8appStartEv+0x24a 909fb0: 0xb46723 _ZN15Uthread_Initial5startEv+0xf When using NMFS and concurrent CIFS and NDMP access, a race condition occurred if two sharedLocks on the same thread had an exclusiveLock request. Loss of access when attempting to generate a tree quota report while the tree quota directory was being deleted.
Fix Summary Code now compares the size of data it is trying to write with the remaining space of the migVnode (before EOF), and writes the smaller. If there's not enough remaining space, code returns EOF.
Tracking Number 413977 / 40557024
1
The fix eliminates the second sharedLock attempt on the thread. Changed the locking associated with looking up tree quota paths to eliminate a potential deadlock. The Data Mover now handles domain SID in Restricted Group correctly.
415992 / 40821538
1
416425 / 40875472
1
When Restricted Groups contained a domain SID, the Data Mover panicked when applying GPO that contained Restricted Groups. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS:
417912 / 41067196
162
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description 2020c9 Err code: 0 Target addr: 0 **. The stack trace showed: 0x2c4d92c: 0x14124c waitForReboot+0x90 0x2c4d94c: 0x1414d5 fault_dump+0x67 0x2c4d96c: 0x13a398 handle_fatal_fault+0xa 0x2c4d990: 0xc7c6fb _ZN17abstractConverter12strTranslateERP KcS1_+0x 0xe61fd9d0: 0xcbb8cc _ZN16localgroupServer11name2MemberEP 13smb_thre 0xe61fdaf0: 0xcbbd3c _ZN16localgroupServer13names2Members EP13smb_th 0xe61fdb20: 0xcbc99b _ZN16localgroupServer12addRGMembersE P13smb_thr 0xe61fdbd0: 0xcbc428 _ZN16localgroupServer19addRestrictedGro upsEv+0
Fix Summary
Tracking Number
2
Renaming a file from an NFSv4 client took an unnecessary lock on the parent directory as well as locking the file itself. This lock on the parent directory was not released when the rename was finished, and had to wait for 10 seconds before it timed out. When user/group quota was enabled on a file system, the default file system limits were not inherited by the user/group quota records.
The code now correctly only locks the file being renamed, and not its parent directory. User/group quota limits are now set correctly.
418716 / 40017698
2
420491 / 41347324
EMC Celerra Network Server Version 6.0.70.4 Release Notes
163
Version 6.0.51.6
Severity 3
Symptom Description Following an upgrade from Version 5.6 to 6.0 in a Linux environment, NFSv4 ACLs were not inherited from the parent and files created could not be used because the permissions were not correct.
Fix Summary When mode bits are set with NFSV4 in MIXED access policy, inheritance is preserved by a special ACL setting. Access right control on this processing was inconsistent with setAttr right checking. This has been fixed. When this occurs, the following message is now logged in the server_log: nfs4_checkSpecial A ce : SID <SID> cannot be translated to NFSV4 : ACL can't be modified. The memory leak issue has been fixed.
Tracking Number 424387 / 40438280
2
If an ACL for a file or directory contained an entry for CIFS-only well-known groups, NFSv4 users were unable to modify the ACL of the file. This is blocked intentionally, to prevent NFSv4 clients from replacing the CIFS-only ACE with an invalid entry. However, there was no indication why the user couldn’t change the ACL.
425723 / 41929606
1
When user changed ownership for millions of files, a memory leak condition accumulated to the point where the Data Mover panicked with the following message: >>PANIC in file: ../addrspac.cxx at line: 529 : Out of memory. The stack trace showed: 0x56c7fb80: 0x146dfe _ZN19Memory_AddressSpace10allocPages El12Memory 0x56c7fbc0: 0x8e54ff _ZN9UFS_Vnode15blockChainWriteEP10Bl ockChainiR 0x56c7fce0: 0x8e6183 _ZN9UFS_Vnode15blockChainWriteEiP10Bl ockChain+ 0x56c7fd50: 0x150dd10 _ZN10DHSM_Vnode15blockChainWriteEiP 10BlockCha
428118 / 42269136
164
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description 0x56c7fda0: 0x86b34c _ZN9VFS_Vnode18writeChainOfBlocksEP1 0BlockChai 0x56c7fde0: 0x7ed2c0 _ZN9File_Fork5flushERyii+0x1ac 0x56c7fe20: 0x7ed7ee _ZN9File_Fork16commit_cachedataEjji+0x1 58 0x56c7fe60: 0x7fc343 _ZN13File_OpenFile5closeEiii+0x55
Fix Summary
Tracking Number
1
User reported a higher than expected CPU usage during deduplication operations.
Eliminated unneeded debugging options which were found to be contributing to the CPU load during deduplication operations.
453242 / 43990760
EMC Celerra Network Server Version 6.0.70.4 Release Notes
165
Version 6.0.51.6
Severity 1
Symptom Description When using NFS V4, revoking an open state from an unconfirmed owner could double release the state, because in some cases the pointer to the (unreferenced) state was not reset to NULL. Since states are recycled, the release in excess caused a missing reference on a valid state. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 674dbb Err code: 0 Target addr: 0 **. The stack trace showed: 0x2c46464: 0x14124c waitForReboot+0x90 0x2c46484: 0x1414d5 fault_dump+0x67 0x2c464a4: 0x13a398 handle_fatal_fault+0xa 0x2c46580: 0x66e244 _ZN11NFSv4_Owner4openEmmP8nfs4_ctx P8stateid4RP 1 0xdffb9590: 0x6704e7 _ZN16NFSv4_StateTable13OpenClaimNull EmmP11NFSv 0xdffb9a10: 0x662b4b _Z9nfs4_openRP4msgbS1_RPhR8nfs4_ctx +0x8a9 0xdffb9d90: 0x65c870 _Z23nfs4_convertNfsv3Statusm+0x626 0xdffb9ee0: 0x63ebf1 _ZN17nfs_serverInitial5startEv+0xad7 When running an online quota check for a file system containing a quota record of anonymous ID, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x00008c7dc6 Err code: 2 Target addr: 0x000000005c **. The stack trace showed: 0760: 0xc1ac70 _ZN7DskDump9checkdumpEPKc+0x7a 07a0: 0xc1ae7e dump_return+0x4 07b0: 0x14c7d2
Fix Summary This issue was fixed, and the state pointer in the owner is reset to NULL before releasing the state.
Tracking Number 407115 / 39374052
1
Code was fixed to ensure the quota record of anonymous UID no longer causes this panic during online quota checks.
421173 / 41453586
166
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description _ZN12Memory_Frame13setProtectionEm19 Memory_Pro t 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x14162d handle_fault+0x1d f7a70: 0x8d72c6 _ZN14UFS_FileSystem14readQuotaDescsE 9quotaType f7ee0: 0x8f0f9c _ZN14UFS_FileSystem9oqcEnableEP20onli neQuotaCh f7f30: 0x8f1221 _ZN20onlineQuotaCheckTask10initializeEv+ 0x149
Fix Summary
Tracking Number
2
When using NFSv4, if a file system was unmounted, some Linux clients could not return their delegations. Delegation are correctly recalled. But the client try an OPEN_DELEGATE_CUR to convert its delegation to an open and then flush its cache. This fails because the file-system is frozen.
During the umount process, code now allows OPEN CLAIM_DELEGAT E_CUR from clients holding delegations to allow flushing the client cache Code was fixed to ensure delegation states are global to a client and no longer conflict with other states Code has been fixed to allow the nas_version command to work properly under this condition.
428074
2
When using NFSv4, client were unable to write to a file opened with a share reservation. This issue was due to I/O having write delegations conflict with share reservation
428236
1
When the NFS file system mounted on the Control Station became inaccessible, issuing the nas_version hung.
424679
EMC Celerra Network Server Version 6.0.70.4 Release Notes
167
Version 6.0.51.6
Installation/Configuration/Upgrades
The following entries describe fixed issues related to installations, configurations and upgrades:
Severity 1 Symptom Description During an upgrade, the nas_version command hung if __db.00* lock files under /var/sadm/pkg/emcnas were corrupted. Fix Summary Code now removes the lock files that are no longer in use, preventing the command from hanging due to prior corruption. With the fix, the upgrade will make sure that pre-reboot task is complete before continuing to the next step in upgrade. Code now correctly ascertains slot number when non-standard DM names are in use. A mechanism to control whether or not inventory data should be included into email notifications has been introduced. Tracking Number 404601 / 3988124
1
During NAS upgrade, NAS service was required to be stopped on Control Station. In some cases when NAS service failed to stop, and Control Station reboot was initiated, the upgrade failed after reboot on the next step because pre-reboot task had not completed successfully. The upgrade Blade parameters task failed during out-of-family upgrades from 5.6 to 6.0 if the system had nondefault host names configured for the Data Mover.
417369 / 40921968
2
414362
2
User wanted to remove a portion from their emailuser alerts. There was a "System Configuration Snapshot" at the bottom of the e-mail that the user did not want to include. It included a "/nas/sbin/model" command and a "nas_inventory -list". The primary goal was to remove the "nas_inventory" command output from these e-mails. When booting a Control Station with ExpressInstall CD/DVD, if user selected Install option, the process did not verify existing backend data, and overwrote it. This was inconsistent with non-ExpressInstall CD/DVD.
423814 / 41763194
1
Fixed the code so ExpressInstall CD/DVD work the same way as nonExpressInstall
421316 / 41453962
168
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description
Fix Summary CD/DVD. Fixed command to properly handle this error condition and no longer report these errors.
Tracking Number
2
When using CUT to upgrade to version 6.0.50, nas_version command displayed the following error: $ nas_version /nas/bin/nas_version: line 21: lsof: command not found rm: cannot remove `/var/sadm/pkg/emcnas/__db.001': Permission denied rm: cannot remove `/var/sadm/pkg/emcnas/__db.002': Permission denied rm: cannot remove `/var/sadm/pkg/emcnas/__db.003': Permission denied Celerra was not picking up failed drive events
421334
1
The file that indicates events to the Celerra operating environment software was updated so that drive failed/removed messages are handled correctly.
432919
NDMP
The following entries describe fixed Network Data Management Protocol/Backup problems:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
169
Version 6.0.51.6
Severity 1
Symptom Description When running NDMP, if the getinq command failed, it accepted the current value of a different buffer. If the different buffer had random data, it resulted in a malloc failure, which led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 515 : malloc() incorrect size - not a positive value. The stack trace showed: 0x4edce14d0: 0x25d13e malloc+0x5e 0x4edce1520: 0x1723c25 _Z18GetScsiDevices NameRP16scsi_devic e_namei+ 0x4edce17e0: 0x1724283 _ZN24NdmpdConfig GetTapeInfoMH14me ssageHandle 0x4edce1880: 0x172c669 _ZN14NdmpConnec tion15processMessag esEi+0x509 0x4edce1910: 0x172c84f _ZN14NdmpConnecti on15processRequest sEv+0x9 0x4edce1920: 0x16f836d _ZN21ConnectionFil eHandler11handleEve ntEv+0x 0x4edce1940: 0x172d38f
Fix Summary The NDMP code now checks the status of the getinq function and exits properly, if the getinq command fails.
Tracking Number 430903 / 42684894
170
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description _ZN19NdmpEventDis patcher8dispatchEij+0 x729 0x4edce1a80: 0x16fb29a _ZN21NdmpConnec tionHandler17connect ionHandle
Fix Summary
Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
171
Version 6.0.51.6
Severity 1
Symptom Description While performing a VBB restore, during file creation, the lookupPath() failed. If the func call did not find the path, it returned NULL. The Data Mover panicked with the following DART panic/fault message: *** DART panic/fault message: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 124a3c0 Err code: 0 Target addr: 0 ** The stack trace showed: 0x2c62e14: 0x141278 waitForReboot+0x 90 0x2c62e34: 0x141501 fault_dump+0x67 0x2c62e54: 0x13a398 handle_fatal_fau lt+0xa 0x2c62e90: 0x1246211 _ZN14vbbCreateFi les21createDesti nationPathEPcm 0x58545f20: 0x1245c3d _ZN14vbbCreateFi les14createBaseT reeEv+0x463 0x58545fa0: 0x1244d84 _ZN14vbbCreateFi
Fix Summary The panic occurred when VBB used the returned NULL pointer. The fix checks the return code and returns failure if the function call fails, instead of panicking the Data Mover.
Tracking Number 416911 / 40908030
172
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description les3runEv+0xde 0x58545fc0: 0x125422b _ZN24vbbProjectM ember_Initial5st artEv+0x1f 0x58545fe0: 0x145c26 Sthread_startThr ead_internal+0xe If the residual data and actual read data of a file during an NDMP restore did not agree, the restore failed because the data read from tape was not what was expected. ERROR MESSAGE: Tape Silvering restore with error of extent length not multiple of 8K. While doing a VBB DAR full restore, the attributes of directories were not reset to the values saved in the archive. This issue did not exist in non-DAR restore.
Fix Summary
Tracking Number
1
This tape read error was fixed. Code now ignores residual data if the read returns OK.
409887 / 39959654
2
The attributes of directories are now reset to the values saved in the archive.
411250
EMC Celerra Network Server Version 6.0.70.4 Release Notes
173
Version 6.0.51.6
Severity 1
Symptom Description When backing up a file system to tape, a media error caused the server_archive command to fail and the Data Mover to panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000ca204c Err code: 0 Target addr: 0x0000000004 **. The stack trace showed: 08e0: 0xce0520 _ZN7DskDump9chec kdumpEPKc+0x7a 0920: 0xce072e dump_return+0x4 0930: 0x14d9f2 record_panic_tic ks+0x242 0950: 0x14e07d fault_dump+0x89 0970: 0x141da5 handle_fault+0x1 d 6ebd30: 0xca26ee _ZN14pax_cmd_t hread9wr_rdfileE P5ARCHDiP11MsgL 6ebdd0: 0xc99f78 _ZN14pax_cmd_t hread10wr_archiv eEv+0x1f42 6ebf40: 0xc9afbb _ZN14backup_In itial8appStartEv +0x243 When using NDMP, during reverse block level object (BLO) recovery, the cmd
Fix Summary NULL pointer protection was enhanced to prevent this panic.
Tracking Number 422966
2
The reverse BLO (block level object) recovery code was modified to work
424158 / 41813810
174
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description service could not find the task context and logged the following error to the server log every 5 minutes filling up the server log : 2011-06-26 04:07:23: CMD: 3: handleTaskObject: INDICATION action TASK Context not found for Task = APM0011100197720 07_2_53:0 2011-0701 12:23:18: CMD: 3: handleTaskObject: INDICATION action TASK Context not found for Task = APM0011100197720 07_2_53:0 After using NDMP/PAX to backup deduplicated files, and then attempting to restore them on a new file system (or a different file system that did not have dedupe enabled), user saw the following errors in the server_log... 011-1028 18:23:44: DEDUPE: 3: builder: RDE not initialized fsid 178 2011-10-28 18:23:44: DEDUPE: 3: writeStream: failed to get ssp and none of the deduplicated files will be restored on that file system.
Fix Summary correctly and no longer log these errors to the server_log.
Tracking Number
1
Fixed the deduplication initialization for this case.
435544
EMC Celerra Network Server Version 6.0.70.4 Release Notes
175
Version 6.0.51.6
Severity 1
Symptom Description While using NDMP, user was unable to run 8 simultaneous server_archive sessions, but could only get 4 to run in parallel, even on 8gb memory systems. NDMP failed to restore files with error message SHADOW: 6: setName: Cannot translate xxx in unicode (5=InvalidArgument) but did not return an error to the NDMP server When user tried to do a DUMP backup and set file or directory filters, if the filters matched a symbolic link and the symbolic link could not be stat()-ed, the backup job failed. Here are some cases when stat() will fail: 1. destination of the symbolic link doesn't exist. 2. the depth of the symbolic link is higher than 8(File_MaxSymLinksI nPath).
Fix Summary Code was fixed to enable user to run 8 server_archive on 8gb memory systems.
Tracking Number 429341
1
Code now sends this warning message to FMA/DMA.
424257
2
Code was changed to correctly account for symbolic links.
423990
176
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Network
The following entries describe fixed Network-related problems
Severity 2 Symptom Description FTP behavior was not working as expected, the -a option on the ftp ls command did not return . and .. This caused an infinite recursive list loop. Eventually, the list command was truncated, resulting in 'command not understood'. Fix Summary Updated the FTP server in order to allow .and.. to be shown in ftp ls -a command This race condition has been eliminated. Tracking Number 412323 / 40150774
1
If the peer server disconnected while the HTTP client completed the initialization of the TCP connection, a race condition led to a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000213187 Err code: 0 Target addr: 0x0000000000 **. The stack trace showed: 0760: 0xc1ac70 _ZN7DskDump9checkdumpEPKc+0x7a 07a0: 0xc1ae7e dump_return+0x4 07b0: 0x14c7d2 _ZN12Memory_Frame13setProtectionE m19Memory_Pro t 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x14162d handle_fault+0x1d 280d30: 0x41c2c0 readmsg+0xe0 280db0: 0x41c324 GetsocknameStream+0xe 280dd0: 0x10c064a _ZN14http_clientCnx11_tcpConnectE v+0xd84
421919 / 41579878
EMC Celerra Network Server Version 6.0.70.4 Release Notes
177
Version 6.0.51.6
Severity 1
Symptom Description When there were lots of fragments (typically UDP traffic) and empty network buffers at the head of the fragment chain, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x00003ce5e0 Err code: 0 Target addr: 0x0000000000 ** . The stack trace showed: 0760: 0xc1ad80 _ZN7DskDump9checkdumpEPKc+0x7a 07a0: 0xc1af8e dump_return+0x4 07b0: 0x14c7d2 _ZN12Memory_Frame13setProtectionE m19Memory_Pro t 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x14162d handle_fault+0x1d ed90: 0x3c3897 ip_mxrput+0x10d edd0: 0x37d8cd _Z7bcmreadPvP4msgbtjj+0x4df eee0: 0x38e1c2 _Z20LM_ServiceInterruptsP16_LM_DE VICE_BLOCK+0x 3 If a user made an SNMP bulk request with a repeat value of "0", the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 378 : malloc() - incorrect size - not a positive value. The stack trace showed: 825ce0: 0x24a290 malloc+0x5c 825d20: 0x10e49ef _Z28netsnmp_create_subtree_cacheP 23netsnmp_a 825d70: 0x10e4eb0 _Z10handle_pduP23netsnmp_agent_se ssion_s+0xc 825da0: 0x10e5d3e _Z22netsnmp_handle_requestP23nets nmp_agent_s 825dd0: 0x10e67c6 _Z18handle_snmp_packetiP12snmp_se ssioniP8snm 825df0: 0x1116a2d _Z12snmpv3_parseP8snmp_pduPhPmPS1 _P12snmp_se 825e40: 0x111868f _Z10_sess_readPvP8LinkListR12Sele ctStream+0x 825eb0: 0x11187de _Z14snmp_sess_readPvP8LinkListR12 SelectStrea When a BCM NIC was hung, it was immediately
Fix Summary Code no longer panics when empty blocks are removed via MSG_TRIM().
Tracking Number 422980 / 41687604
1
Code now checks for non-null 'r' repeat values before allocating memory using 'r' value.
426298 / 42091892
1
178
Changes
434044
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description reset.
Fix Summary were made to allow specifying that the BCM NIC device not be reset when it appears to be hung. This only impacts some troubleshootin g scenarios. Code has been fixed to properly allocate/relea se the required amount of memory used by TFTP stack, and prevent this Data Mover panic.
Tracking Number
1
Heavy usage of the TFTP protocol on a Data Mover caused memory overflow and led to a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x000089863e Err code: 3 Target addr: 0x0000b42f72 **. The stack trace showed: 1760: 0xbffe60 _ZN7DskDump9checkdumpEPKc+0x7a 17a0: 0xc0006e dump_return+0x4 17b0: 0x14c7d2 _ZN12Memory_Frame13setProtectionE m19Memory_Pro t 17d0: 0x14ce5d fault_dump+0x89 17f0: 0x13378d handle_fault+0x1d 2bdd00: 0x8b9261 _ZN14UFS_CgBlkEntry11postFreeBlkE PvS0_+0x5d1
434697 / 41576516
EMC Celerra Network Server Version 6.0.70.4 Release Notes
179
Version 6.0.51.6
Severity 1
Symptom Description When using a Domain name in an export list for NFS client, if the DNS server was unavailable, a Data Mover panicked while the NFS client was trying to mount the export. DART panic message: *** Page Fault Interrupt. Virt ADDRESS: 0x00004a4c77 Err code: 0 Target addr: 0x0000000048 **. The stack trace showed: 0x36d0760: 0xc1ad80 _ZN7DskDump9checkdumpEPKc+0x7a 0x36d07a0: 0xc1af8e dump_return+0x4 0x36d07b0: 0x14c7d2 _ZN12Memory_Frame13setProtectionEm 19Memory_Prot 0x36d07d0: 0x14ce5d fault_dump+0x89 0x36d07f0: 0x14162d handle_fault+0x1d 0x834ecb0: 0x40db4e udp_input+0x9fa 0x834ee00: 0x3ce997 ip_input+0x1425 0x834eeb0: 0x3c3897 ip_mxrput+0x10d
Fix Summary If the DNS server is unavailable, code will no longer panic under this condition, but will post the following message to the server_log: Portmap session is already removed because of timeout.
Tracking Number 437344 / 43462506
180
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Replication
The following replication-related bugs that were fixed in this release:
Severity 2 Symptom Description User could not create a checkpoint for a nested mounted file system for NDMP backup Unisphere did not display current transfer information of a replication session if the size was greater than 2TB. Fix Summary The fix will allow checkpoint creation from nested mounted file system during NDMP backup. Unisphere now displays the transfer information of a replication session regardless of the size. Tracking Number 407727
2
413467 / 40439600
EMC Celerra Network Server Version 6.0.70.4 Release Notes
181
Version 6.0.51.6
1
If an ongoing replication copy was disturbed (for example: unmounting the base line checkpoint during a nas_copy session), the Data Mover incorrectly marked the copy as complete although there was an error during the copy. The copied file system remained inconsistent which caused the Data Mover to panic when the file system was accessed. DART/panic fault message: mangled directory entry. The stack trace showed: 0x109412b00: 0x89c445 _ZN14UFS_FileS ystem28markCorru ptedFsForU nmoun 0x109412c70: 0x928411 _ZN9UFS_Vnode6 dirbadEjPKc+0xb1 0x109412d10: 0x90cb57 _ZN20UFS_DirHa shInfo_dir3C1EP1 1Local_Vno de+0x 0x109412db0: 0x93e582 _ZN15UFS_DirOp s_dir311initDirH ashEP9UFS_ Vnode 0x109412df0: 0x933dc9 _ZN15UFS_DirOp s_dir37scandirEP 9UFS_Vnode PK18F 0x109412ec0:
Code now properly checks the status codes for local replication using the loopback interface. This prevents marking a replication session to be completed when one of the base line checkpoints is unmounted, and eliminates the panic.
418323 / 41087088
182
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
0x91b353 _ZN9UFS_Vnode7 scandirEPK18File _NameCompo nentR 0x109412ed0: 0x93083c _ZN9UFS_Vnode1 0lookupCompEPK18 File_NameC ompon 0x109413000: 0x13bd49d _ZN10MGFS_Vnod e10lookupCompEPK 18File_Nam eCom
EMC Celerra Network Server Version 6.0.70.4 Release Notes
183
Version 6.0.51.6
1
When a checkpoint was taken, if the memory was fragmented and there was a lot of DBMS activity, the Data Mover panicked with the following .DART panic/fault message:>>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory. The stack trace showed: 0x5a9a1c30: 0x145966 _ZN19Memory_Ad dressSpace10allo cPagesEl12 Memory 0x5a9a1c70: 0x24a34e malloc+0x11a 0x5a9a1cb0: 0x24ad2f realloc+0x81 0x5a9a1ce0: 0x1285087 __os_realloc+0x3 c 0x5a9a1d10: 0x1283f76 __memp_sync_in t+0x280 0x5a9a1de0: 0x1284822 __memp_sync+0xe6 0x5a9a1e10: 0x12902db __txn_checkpoin t+0x1be 0x5a9a1eb0: 0x1290555 __txn_checkpoi nt_pp+0xf9 Too many duplicated final response for
The fix consists of allocating all the memory when the environment is initialized.
421132 / 41400444
1
Added a defensive mechanism to
398657 / 38323670
184
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
DicEvent were queued in the cmd queue, and each of them consumed about 24KBytes which exhausted the Data Mover memory. DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 289 : Out of memory. The stack trace showed: 0xaae2dda0: 0x14124c waitForReboot+0x 90 0xaae2ddc0: 0x1414d5 fault_dump+0x67 0xaae2dde0: 0x1413bd PANIC+0x29 0xaae2ddf0: 0x13bef6 _ZN19Memory_Addr essSpace10allocP agesEi12Me mory 0xaae2de10: 0x10a0283 _ZN10DpBuf_Pool8 getChunkEiiP16Dp Buf_PoolMe mbe 0xaae2de30: 0x10a0386 _ZN10DpBuf_Pool8 allocateEv+0x5c 0xaae2de50: 0x107fef8 _ZN9DpRequestC1E 15DpRequestMode_ e17DpReque stC 0xaae2de70: 0x10b1776 _ZN21DpService_R eplicaInfo19proc essInfoPar all
eliminate final response duplication.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
185
Version 6.0.51.6
1
In a case where an error occurred during a tape silvering restore process, and the restore was aborted, a RedirectReads(FALS E) call in the code caused the file system to enter a freeze state and be set inactive. Restore from the internal ckpt then failed due to the inactive file system state. The replication delete operation failed to delete the mount option record in the database. Leftover replication file system mount option record may silently change the file system readwrite attribute after a reboot, even without replication on it.
The code was fixed by moving RedirectReads(FALS E) to the end of the tape silver restore, preventing this issue and allowing the restore to complete.
411982 / 39959654
1
Code now checks whether there is replication on the file system, and only saves the mount option record when necessary. In addition, replication code removes duplicated entries during replication start.
421209 / 41475440
Security
The following entries describe fixed security-related issues
Severity 2 Symptom Description When the LDAP Service was reconfigured with no bind DN or password specified, the LDAP connection (and associated 'ldap' thread) were 'orphaned' or no longer used by the LDAP Service. While the connection and thread are usually Fix Summary If the connections are not used (idle for a period of time), they will be disconnected by the LDAP Server. Tracking Number 426901 / 42075838
186
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.51.6
Severity
Symptom Description cleaned up when (idle) timed out at the LDAP Server, if an LDAP disconnected, and multiple reconfigurations were attempted, the number of connections rapidly grew, consuming memory, threads, and other resources. The following known security vulnerabilities existed prior to this release: Apache HTTP server (CVE-2011-3192) Bzip2 (CVE-20100405), DHCP (CVE-20110997) GnuC libraries ( CVE2011-1071 ) GnuTLS (CVE-20081948, CVE-20081949, CVE-20081950 ) JRE (multiple vulnerabilities) Libpng (CVE-20086218) LibXML (CVE-20092414 and CVE-20092416) Logwatch (CVE-20111018) OpenSSL (CVE-20113207, CVE-20113210, CVE-20110014),
Fix Summary
Tracking Number
2
Several packages were updated to address these known security vulnerabilities. JRE update includes Oracle Java SE Critical Patch Update June 2011.
392895 , 413785, 428931, 428932, 428933, 431357, 433301, 433470, 433475, 433477, 432273 / 42301928 42301296 42301876 42302062
EMC Celerra Network Server Version 6.0.70.4 Release Notes
187
Version 6.0.51.6
Storage
The following entries describe fixed storage-related issues
Severity 1 Symptom Description An environmental issue occurred with HBA 1, during a self-test, which led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG. The stack trace showed: cee0: 0x14dd29 system_watchdog+0x61 cef0: 0x132913 t_nmi+0xc3 cfd0: 0x155263 Sthread_startThread_internal+0 x7 During a FLARE upgrade, when the LUN trespassed from the primary to the secondary processor, the Data Mover panicked. DART panic/fault message: >>PANIC in
file: ../BVolumeIrp.cxx at line: 405 : IO failure on Vol:89, blkNo:42139696, SKey:0x54. The stack trace showed: fed0: 0xa9fad5 _ZN8SgIO_IRP9irpFailedE14BlockIO_ Statusy +0x477 ff20: 0xb3baa1 _ZN14CamDisk_Device8completeEP10c cb_head er+0x14 ff90: 0xabf9c4 _ZN8Scsi_Xpt16processCallbacksEv+ 0xea ffc0: 0xabfa3d _ZN18Scsi_serverInitial5startEv+0 xd ffd0: 0x155263 Sthread_startThread_internal+0x7
Fix Summary Code has been fixed to prevent this environmental issue from panicking the Data Mover.
Tracking Number 408642 / 39774566
1
Fixed a timing window where two threads could each detect that a trespass is required and attempted to trespass to the wrong SP. Code now locks the volume and tests whether the current SP has changed before switching paths and issuing the trespass request.
423378 / 41629776
188
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.43.1
Version 6.0.43.1
Deduplication
The following entries describe fixed issues related to Deduplication.
Severity 1 Symptom Description Following an upgrade to version 6.0.42.0, when reading a deduplicated file for the first time and creating/refreshing a checkpoint, the checkpoint was marked as corrupted and unmounted. This was due to improper lock handling and creating a checkpoint with a dirty flag while there was a superblock update on the production file system. Fix Summary Code has been fixed to properly use the lock and prevent checkpoint from being falsely marked corrupted under this condition. Tracking Number 429662 / 42529536
Version 6.0.43.0
File Systems
The following entries describe fixed issues related to File Systems:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
189
Version 6.0.42.3
1
Celerra Network Server version 6.0.42 added a check for invalid Alternate Data Stream names. When a client tried to query an invalid streamname, the operation was rejected as expected but in the process a lock was left on the file or directory. This led to other threads getting blocked if they needed to gain exclusive access to the object.
The checking for invalid streamnames has been removed.
393688
Version 6.0.42.3
Celerra Management Applications
The following are fixed problems related to the Celerra Management Applications, i.e., Celerra Monitor, Celerra Manager, WebUI, or Celerra MMC Snap-ins:
Severity 2 Symptom Description File system names containing a space could not display correct space information on GUI. The usage bar was empty, indicating there was no space used, even if the file system had consumed some space. When a quota crossed the soft limit and the grace period was unlimited, that quota property was Fix Summary Modified code to correctly return the correct space information for file systems with names containing a space. Tracking Number 35613 / 33731506
2
Code now translates the unlimited value to string "No limit".
401502 / 38626372
190
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
not displayed in the GUI. Code represented an unlimited grace period as -1, and translated that to a string "NoLimit". The code path for the above scenario did not translate the value to a string and it caused an exception trace.
CIFS and NFS
This release addresses a potential vulnerability which could, under certain conditions, be exploited to cause a Denial of Service (DOS) attack.
CIFS
The following entries describe fixed Common Internet File System problems:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
191
Version 6.0.42.3
1
When using emcacl to set the security descriptor on a symbolic link on a Data Mover with the param cifs.acl.extacl=0x28 and CEPA enabled, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: c5f858 Err code: 0 Target addr: 1c ** . The stack trace showed: 0x2c39cb4: 0x14124c waitForReboot+0x90 0x2c39cd4: 0x1414d5 fault_dump+0x67 0x2c39cf4: 0x13a398 handle_fatal_fault+0xa 0x2c39d40: 0xc5228f _ZN8AppLibNT5closeEP8 fid_infoiR11File_St atusP17 0xe4307ec0: 0xae5301 _ZN14smb1_threadCtx10 replyCloseEP4msgb+0 x289 0xe4307f40: 0xbe4b0d _ZN14smb1_threadCtx7p rocessEP4msgbi+0xcd f 0xe4307fa0: 0xbd92ec _ZN13smb_threadCtx5st artEv+0x6d0 0xe4307fe0: 0x145bfa Sthread_startThread_i nternal+0xe A 0% CPU idle condition might be observed on CIFS servers with many existing connections.
The Data Mover panics when trying to get the size of the file when setting the security descriptor if the file is a symbolic link. The code now checks if the target is a symbolic link before attempting to get the file size.
416553 / 40899086
1
Code no longer browses existing connections during authentication. Code was incorrectly using the netBIOS name instead of computer name. This has been fixed.
404202 / 38611382
1
If the computer name of the CIFS server differed from its NetBIOS name, an attempt to join a CIFS Server to a Domain from GUI failed with the error "compname <comp_name> does not exists in CIFS
393863
192
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
configuration. 2 When using Virus Checker with a file system mounted on an NMFS path, the path sent to the virus checker was incomplete. This resulted in the file not being scanned When a CIFS user connects to a share attached to an NMFS, code now puts the mount path of all the file systems mounted on this NMFS in memory to ensure all files are scanned. Code now forces a retrytimeout internally to at least 300ms for the heartbeat. 410012 / 39861830
1
CEPA service was intermittently disconnected when retrytimeout was set to a value less than 200ms. This issue occurred because the TCP heartbeat request time out is always 300ms or greater. CIFS clients saw CIFS request timeouts in environments with high latency and high utilization. If the TCP queue for a CIFS was full, typically because a client makes several parallel requests, or there is high network latency (such as with a WAN), then clients may see the Celerra appear to ignore some CIFS requests. If the queue remained nearly full, eventually clients could see a CIFS timeout. This could also happen with CIFS clients that were not acknowledging TCP data quickly, but making CIFS requests aggressively.
397281 / 37879634
2
When the CIFS TCP queue was full, new responses were placed at the head of the queue instead of at the end. If the TCP queue never fully emptied (because there was always data outstanding on the network), then some CIFS responses might never be transmitted, leading to timeouts. The problem is resolved by appending new replies to the end of the TCP queue instead of the beginning. Code now drops incomplete requests, thus avoiding blocked threads.
391540 / 36829842
1
When running clients which don't fully respect protocol, our CIFS server became unavailable after consuming all internal threads resources due to pending requests.
404372 / 39134760
EMC Celerra Network Server Version 6.0.70.4 Release Notes
193
Version 6.0.42.3
1
The Celerra CIFS client was unable to connect to CIFS servers using NTLMv2. This would mean that if Kerberos authentication was unavailable and NTLMv1 was disallowed, the Celerra would fail to open CIFS connections. This could cause failures for DHSM recalls/file access, for CDMS, FTP authentication with Active Directory, or CIFS authentication of users. During a CDMS migration between 2 Celerras, the migration hung. On the target Celerra, user was getting various errors indicating, that the source wouldn't deliver the data. On the source, user was getting the following error at the exact same time at the fault occurred: 2010-12-11 00:50:42: SMB: 3:[codan] Too many open policies on pipe=cMsrpc_Policy for client=10.194.116.100 (257/256). During the migration, the LSA function opens policies, but was failing to close them properly, eventually leading to the process hanging. If a security descriptor (SD) was not defined, a memory leak condition occurred in shareEnum502 which led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../bufpool.cxx at line: 1309 : couldn't get a free page. The stack trace showed: 0x11757f080: 0x241633 _ZN18Buffer_Cached_Po ol17makeCachedBuffe rsEm+ 0x11757f0d0: 0x14e514 _ZN19Memory_AddressSp ace16makeCachedBuff
Fixed a DNS comparison issue preventing CIFS clients from using NTLMv2 in Active Directory environments.
399722 / 38523654
1
Added a use function to close LSA or SAMR open handles and prevent this error.
398486
1
Prior to the fix, code was building a SD which was automatically deleted. With this fix, the SD is no longer deleted.
393688
194
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
erEv+ 0x11757f110: 0x2416ef _ZN18Buffer_Cached_Po ol12findAfreeBufEmi ii+0x 0x11757f150: 0x2417d7 _ZN18Buffer_Cached_Po ol15getAnyCachedBuf Emiii 0x11757f1a0: 0xcf5d8b _ZN17UnCachedIO_Objec t9readBlockEyj16Blo ckIO_ 0x11757f290: 0xa6fc23 _ZN14UFS_FileSystem9r eadBlockEiiRP13Buff er_Ca 0x11757f300: 0xa647fe _ZN9UFS_Vnode5breadEj R11File_Statusi+0x4 de 0x11757f370: 0xb135f0 _ZN9UFS_Vnode9blkatof f2EyRPvRP13Buffer_C ached 1 When using a Windows backup tool to backup Unix files and data, a Data Mover became unresponsive if extended ACLs are validated. Removed a condition where two shared locks were taken in sequence which led to a deadlock. 403587 / 38982154
EMC Celerra Network Server Version 6.0.70.4 Release Notes
195
Version 6.0.42.3
1
When an NFS connection using Kerberos for authentication failed, a function was called using the wrong variable as a parameter, resulting in a Data Mover panic when it tried to obtain data that did not exist. DART panic/fault message: *** GP exception. Virt ADDRESS: 0x0000732c16. Err code: 0 ***. The stack trace showed: 1760: 0xbf6470 _ZN7DskDump9checkdump EPKc+0x7a 17a0: 0xbf667e dump_return+0x4 17b0: 0x14c7d2 _ZN12Memory_Frame13se tProtectionEm19Memo ry_Prot 17d0: 0x14ce5d fault_dump+0x89 17f0: 0x13378d handle_fault+0x1d 38be40: 0x733648 _ZN10NfsGssInit12proc essReplyERP4msgb+0x 42 38be70: 0x733105 _ZN13NfsGssRequest8di spatchEP4msgbPvP17A ccess 38bf00: 0x6f443a _ZN17nfs_serverInitia l5startEv+0xa5c
Code now uses the correct variable in the function call.
409927 / 39986788
196
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
1
Data Mover panicked when a Windows 7 client connected to a CIFS share through a Riverbed WAN accelerator with the SMB2 protocol enabled. The Riverbed WAN accelerator sent an SMB2_IOCTRL request compounded with SMB2 commands which the Data Mover did not decode correctly. DART panic/fault message: >>PANIC in file: /disks/NSG_Maint8/6.0_ build _nfs_env/Dart/server/sr c/kernel/include/st ream.hxx at line: 1119 : datab::freeRef: db_ref <= 0. The stack trace showed: 245ea0: 0xdcc4fd _ZN14smb2_threadCtx7pro cessEP4msgbi+0xf57 245f20: 0xdee8d6 _ZN13smb_threadCtx5star tEv+0xc4c 245fd0: 0x155263 Sthread_startThread_int ernal+0x7
The Data Mover now properly handles compound SMB2 requests received in this manner.
411721 / 40195472
EMC Celerra Network Server Version 6.0.70.4 Release Notes
197
Version 6.0.42.3
1
During a GPO update, if the connection to the domain controller is lost, an error case was used that deleted an object without setting the associated pointer to NULL. At the end of the update, the pointer was checked and deleted if not NULL. This resulted in a double-free, which caused a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 525 : Memory corrupted. The stack trace showed: 06d720: 0x24aaaf free+0x79 06d760: 0x24ae7d _ZdlPv+0x9 06d770: 0xf2b7fc _ZN9DomainGpo18update VdmGpoServerERK6Str ingi+ 06dd40: 0xf2be4b _ZN9DomainGpo15update GpoServerERK6String i+0x4 06dd80: 0xf250ae _ZN4gpod5startEv+0x117e 06dfd0: 0x155263 Sthread_startThread_i nternal+0x7 When using third party VPN software to connect to a CIFS server from a client that had the same machine name as the domain name, NTLM authentication failed with LOGON_FAILURE error and users were not able to connect to CIFS shares. A Data Mover panicked when trying to connect to a share on a file system which was corrupted. DART panic/fault message: ** Page Fault
Code has been fixed to set the pointer to NULL after the object is deleted to prevent this panic.
409288 / 39850746
2
The code has been fixed to ignore client names that match the domain name of the CIFS server, allowing authentication to succeed.
407582 / 39550232
1
Code has been fixed to check the state of the file system before processing the share request, thus
401537 / 38704210
198
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Interrupt. Virt ADDRESS: 72735b Err code: 0 Target addr: 1e8 ** . The stack trace showed: 0x2c4c914: 0x14124c waitForReboot+0x90 0x2c4c934: 0x1414d5 fault_dump+0x67 0x2c4c954: 0x13a398 handle_fatal_fault+0xa 0x2c4c9f0: 0x83e738 _ZN26passthroughShadowDir ectory16lookupN ameLock 0x28a2fa40: 0x83f662 _ZN26passthroughShadowDir ectory23lookupN ameLoc 0x28a2fa70: 0x836609 _ZN26passthroughShadowDir ectory23interna lLooku 0x28a2fb00: 0x8255a6 _ZN19ShadowDirectoryBase1 5lookupComponen tEPtmP 0x28a2fb30: 0xc6b217 _Z15getnodefromRootP15File _NamingNodePKc RS0_+0 1 When there were many CIFS users accessing a Data Mover, the Data Mover might panic due to 64-bit higher memory footprint. DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory.
preventing this panic.
Code has been fixed to reduce the size of various CIFS objects to improve memory usage.
404117 / 39062050
EMC Celerra Network Server Version 6.0.70.4 Release Notes
199
Version 6.0.42.3
1
Under a heavy CIFS load with many antivirus threads, there were 94 Write I/Os in progress which caused a memory fragmentation issue led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory. The stack trace showed: 06b0: 0x145966 _ZN19Memory_AddressSp ace10allocPagesEl12 Memory_ 06f0: 0x207f2e _ZN20Buffer_Uncached_ Pool10addBuffersEji j+ 0xa60750: 0x2098fd _ZN20Buffer_Uncached_ Pool14getUncachedBu fE15Str 0790: 0x20ae0d allocb+0x67 07c0: 0x20bc47 pullupmsg+0x79 0810: 0x231545 _Z10msgdsize_cPP4msgb+0 x1a7 0860: 0x231770 dlpiWPut+0xea 08d0: 0x3d0276 ip_output+0x432 With the dfsutil CLI from windows 2008R2, user could not reset the SD of a DFS link.
Implemented changes to the NIC driver to reduce the odds of panicking due to memory fragmentation.
400441 / 38611382
2
Code has been fixed to properly reset SD on the DFS link using dfsutil.
403696
Control Station
The following entries describe fixed problems related to the Celerra Control Station.
200
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity 1
Symptom Description The MirrorView activate command caused Control Station to reboot twice, 1st time after adding NBS service on RDF7 to DM2, and 2nd time after adding NBS service on RDF10 to DM2. The command finally succeeded on the 3rd attempt. While using Unisphere and trying to enable user and group quotas in one step, only the user quota was enabled. The group quota enable operation failed. LDAP connection in Unisphere under Settings->Manage LDAP Domain fails with the error 13693354039: CST (Common Security Toolkit) library internal error.
Fix Summary Both RecoverPoint and MirrorView DR features have been fixed and prevents Control Station rebppts during the DR failover at remote NBS volumes setup.
Tracking Number 388624
1
Code hass been fixed to detect the enabling of both user and group quota and use the "-both" command line operation.
404960
2
Code has been fixed to provide more information in the recommended action when this error occurs so that the user can correct it and proceed.
401078 / 38622682
Deduplication
The following entries describe fixed issues pertaining to Celerra Deduplication:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
201
Version 6.0.42.3
Severity 2
Symptom Description If the remote file system ID was a substring of another file system ID, user could not take a nearcopy/farcopy backup snap. This issue occurred if the remote file system size query substring match returned more than one item. If the file system had quota enabled with filesize policy, and there's sparse recalled files in the Deduplication feature, then after an online filesystem quota check finished, the quota block usage report will be incorrect.
Fix Summary Changed a remote query from sub-string match to exact match. The online quota check scanned both stub file and migration file, so the file size was calculated twice. This fix makes the online quota check skip the filesize of stub file if it has a sparse recalled migration file. There is a maximum of 4TB of deduped data. Code will now log a more specific error information in the server_log. Code has been fixed to properly declare all the functions and prevent this condition.
Tracking Number 402256
2
406699
2
If the group quota usage exceeded 4TB, and F-RDE was enabled, migrateFile failed. Only an internal error was logged, there was no specific error information in server_log to help issue diagnosis.
403974 / 38996546
1
When performing NDMP restore of a deduped file system on a highly utilized Data Mover, a buffer overflowed and a subsequential Data Mover panicked with the following DART panic/fault message: *** GP exception. Virt ADDRESS: 0x0000964d88. Err code: 0 ***. The stack trace showed: 0760: 0xbf6490 _ZN7DskDump9checkdumpEPKc+0x7a 07a0: 0xbf669e dump_return+0x4 07b0: 0x14c7d2 _ZN12Memory_Frame13setProtectio nEm19Memo ry_Prot
401349 / 38640140
202
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity
Symptom Description 07d0: 0x14ce5d fault_dump+0x89 07f0: 0x13378d handle_fault+0x1d b7a00: 0x13b584e _ZN10DHSM_Vnode30convertOffline BlockChai nToHo b7a80: 0x13faf8c _ZN19File_OpenDedupeFile10write HolesEjj+ 0x416 b7ad0: 0x13fc10e _ZN19File_OpenDedupeFile21handl eExtentListWri
Fix Summary
Tracking Number
2
Needless error messages similar to the following in the server log for Deduplication enabled filesystems being replicated. 201104-22 13:40:27: DEDUPE: 3: PE: PersistState() returned error: Filesystem is a read-only 2011-04-22 13:40:27: DEDUPE: 3: PE: WriteOperationBegin filesystem fsid 292 is read only. There were no dedupe operations initiated by the Celerra administrator at the DR site.
Code has been fixed to hide these informational messages
413576 / 39901452
FileMover
The following entries describe fixed FileMover related problems:
Severity 1 Symptom Description Incremental backup sometimes picked up unchanged FileMover (DHSM) files Fix Summary Corrected the way offline file size is calculated so it will not cause incremental backup to incorrectly backup unchanged data. Tracking Number 405361/ 37027684
File systems
The following entries describe fixed issues pertaining to File Systems.
Severity Symptom Description Fix Summary Tracking Number
203
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
1
The /nas/sbin/uc_config -verify command returned the incorrect translation settings for the client specified. This could lead to thinking the configuration is not configured correctly.
/nas/sbin/uc_con fig -verify now always gives the correct translation setting being used for the specified client. Code no longer logs a message when the retention period of a locked file in FLR-C file system matches the retention period that is passed in by requests.
387225 / 36947724
1
While using Networker 6.1 to backup the file system via a CIFS connection, the FLR log was flooded with a large amount of repeating log entries. The rapidly expanded FLR log contributed to extending backup times. Following is an example of the repeating FLR entries: <date> Extend Retention Period on worm committed file : Inode No = <Inode> : Generation No = <generation> : UID = <UID> : Gid = <GID> : File Mode = <mode> : Filesize = <size> : Old RP = <date> : New RP = <date> : Failed An NFSv4 state object was recycled prematurely (one reference missing) which caused a Data Mover to panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x000068cfbe Err code: 0 Target addr: 0x00000001c0 **.The stack trace showed: 6760: 0xbf96c0 _ZN7DskDump9checkdumpEPKc+0x7a 67a0: 0xbf98ce dump_return+0x4 67b0: 0x14c7d2 _ZN12Memory_Frame13setProtectio nEm19Memo ry_Prot 67d0: 0x14ce5d fault_dump+0x89 67f0: 0x13378d handle_fault+0x1d f7f9d0: 0x6a3257 _Z17NFSv4_lockExpiredP11NFSv4_S tate+0x15 f7f9f0: 0x6a32c2 _ZN17NFSv4_delegClient9sendBrea kEjiyy+0x 30 f7fa50: 0x874603 _ZN10Delegation9sendBreakEjyy+0 x5f3 A race condition occurred when some of the file system operations resulting in vnode closure occurred while the same file system was being unmounted on the Data Mover,
406428 / 39472638
1
The Data Mover now confirms the release of the object before allowing it to be used.
410524 / 40054076
1
Code has been fixed to properly release all the references on the
356875 / 33841178
204
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
resulting in a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../ufs.hxx at line: 1077 : UFS hashRemove: not on hash chain. The stack trace showed: 0xd1cedbb0: 0x13d890 waitForReboot+0x90 0xd1cedbd0: 0x13da8d fault_dump+0x67 0xd1cedbf0: 0x13d98d PANIC+0x29 0xd1cedc00: 0x61cd54 _ZN9UFS_Vnode4loseEv+0x8e 0xd1cedc20: 0x5f05f1 _ZN9VFS_Vnode7reclaimEv+0x81 0xd1cedc40: 0x5ee99e _ZN9VFS_Vnode11getNewVnodeEv+0x 40 0xd1cedc50: 0x5ee337 _ZN9VFS_VnodenwEmP14VFS_FileSys tem+0x9 0xd1cedc70: 0xb98871 _ZN15DHSM_FileSystem8newVnodeEm 8dirTypes +0x41 While unmounting a file system with a DFS root share on it, a race condition occurred which led to a Data Mover panic with the following DART panic/fault message: >>PANIC: I/O not progressing LastVol touched not Known. The stack trace showed: 0x802fe90: 0x14124c waitForReboot+0x90 0x802feb0: 0x1414d5 fault_dump+0x67 0x802fed0: 0x1413bd PANIC+0x29 0x802fee0: 0x8b3c56 _ZN9RawIO_IRP13checkProgressEv+ 0x11a 0x802ffd0: 0x8b3b32 _ZN17IrpProgressThread5startEv+ 0x6 0x802ffe0: 0x145bfa Sthread_startThread_internal+0x e
file system prior to the unmount, preventing this condition.
This race condition has been eliminated by locking the share only when we are sure it is unmounting file system.
383368 / 38343174
EMC Celerra Network Server Version 6.0.70.4 Release Notes
205
Version 6.0.42.3
2
Memory was consumed and not freed when a wildcard search (ex. dir *.txt) was run from an FTP session with no matches. This led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 289 : Out of memory. The stack trace showed: 0xe10d1c30: 0x14124c waitForReboot+0x90 0xe10d1c50: 0x1414d5 fault_dump+0x67 0xe10d1c70: 0x1413bd PANIC+0x29 0xe10d1c80: 0x13bef6 _ZN19Memory_AddressSpace10alloc PagesEi12 Memory 0xe10d1ca0: 0x1459f1 _ZN7SthreadC1EPcP10DQueueHeadIS _Eji+0x1e b 0xe10d1cc0: 0xfc4f42 _ZN21ckptBlockAllocatorThr6laun chEPKc+0x 6a 0xe10d1cf0: 0xf9f333 _ZN19PoolManagedDeltaVol14start Allocator Ev+0xa 0xe10d1d10: 0xf981a5 _ZN19PoolManagedDeltaVolC1EPKcS 1_P7LVolu meS3_m An NFS client issued two illegal cross file system rename requests to the Data Mover at the same time which led to a deadlock condition and loss of NFS access.
The memory was not correctly freed after each wildcard search within FTP. This has been fixed.
407209 / 39528390
1
Code was modified to reject the illegal request and return the NFS3ERR_XDEV error when it receives cross file system rename request. The fix is a protective measure which takes an exclusive mutex before revoking the locks held by a client. This will avoid the Data Mover panic, but users should upgrade the Linux clients’ kernel to
402586 / 38891086
1
Linux NFSv4 clients exposed to the issue described in
400183
https://bugzilla.redhat.com/show_bug. cgi?id=264721 present the same clientid
string to the Data Mover and therefore can be seen as a single client. This caused a race condition which led to a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 1ce857 Err code: 0 Target addr: c8 **. The stack trace showed:
EMC Celerra Network Server Version 6.0.70.4 Release Notes
206
Version 6.0.42.3
0x2c62c84: 0x141278 waitForReboot+0x90 0x2c62ca4: 0x141501 fault_dump+0x67 0x2c62cc4: 0x13a398 handle_fatal_fault+0xa 0x2c62d00: 0x67ca57 _ZN11NFSv4_State5closeEmP8state id4P8nfs4 _ctxRP4 0xe1059d20: 0x67b0d8 _ZN16NFSv4_StateTable5CloseEmP8 stateid4P 8nfs4_ 0xe1059d50: 0x669c7d _Z10nfs4_closeRP4msgbS1_RPhR8nf s4_ctx+0x 18b 0xe1059d90: 0x665e30 _Z23nfs4_convertNfsv3Statusm+0x 626 0xe1059ee0: 0x647bdd _ZN17nfs_serverInitial5startEv+ 0xad7 2 On write heavy workloads, a lot of the following messages "Run out of staging buffers" were observed indicating that the current static setting of 6 for agregateWork::numbuf was not sufficient during peak write loads.
the versions mentioned in the Bugzilla report to resolve the issue completely.
Introduced a new parameter that can be tuned to get rid of 'Run out of staging buffers' messages. This newly introduced parameter ufs.aggregateC ommitNbufs can be adjusted to handle peak write loads for iSCSI and uncached IO. The parameter values are as follows: 32 (default), 6 (minimum) and 256 (maximum). Refer to the Parameters Guide for more detailed information about this parameter.
409307
EMC Celerra Network Server Version 6.0.70.4 Release Notes
207
Version 6.0.42.3
1
After a file system corruption, FSCK discovered the corruption and fixed most of it by freeing the corrupted inodes. However, fsck could not properly handle some special file types (block type & char type) from the file system perspective. Further access to such file types caused the Data Mover to panic with the following DART panic/fault message: isAValidIndBlkBuf: bad bn. The stack trace showed: 0xdc2bb740: 0x14124c waitForReboot+0x90 0xdc2bb760: 0x1414d5 fault_dump+0x67 0xdc2bb780: 0x1413bd PANIC+0x29 0xdc2bb790: 0x778cb5 _ZN14UFS_FileSystem28markCorrup tedFsForU nmount 0xdc2bb950: 0x78f1aa _ZN14UFS_FileSystem17isAValidIn dBlkBufEP 13Buff 0xdc2bb9d0: 0x785233 _ZN9UFS_Vnode17readIndirectBloc kElRP13Bu ffer_C 0xdc2bba00: 0x7877ae _ZN9UFS_Vnode8bmapListEmmPmPi+0 x228 0xdc2bba70: 0x78a5e6 _ZN9UFS_Vnode15blocksReadAsyncE jiPP13Buf fer_Ca When file system had many corrupted directory entries and FSCK was running on this file system, the Data Mover panicked with the following DART panic/fault message: *** GP exception. Virt ADDRESS: 87db94. Err code: 0 ***. The stack trace showed: 0x2c3be94: 0x14124c waitForReboot+0x90 0x2c3beb4: 0x1414d5 fault_dump+0x67 0x2c3bed4: 0x13a398 handle_fatal_fault+0xa 0x2c3bff0: 0xe8155d31 _end+0xe495d509 0xe8151fc0: 0x87d471 _ZN12Notfoundlist6locateEmP6dir ValRbPPS_ +0x3f 0xe8152040: 0x87d471
FSCK code has been enhanced to properly handle block and char type of the files and prevent this panic.
404536 / 39140368
1
The FSCK code has been changed to handle large number of corrupted directory entries correctly.
391164 / 37582944
208
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
_ZN12Notfoundlist6locateEmP6dir ValRbPPS_ +0x3f 0xe81520c0: 0x87d471 _ZN12Notfoundlist6locateEmP6dir ValRbPPS_ +0x3f 0xe8152140: 0x87d471 _ZN12Notfoundlist6locateEmP6dir ValRbPPS_ +0x3f 1 An out-of-sequence redundant close operation for the same owner led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../include/stream.hxx at line: 989 : datab::freeRef: db_ref <= 0. The stack trace showed: 0xe00dbc30: 0x14124c waitForReboot+0x90 0xe00dbc50: 0x1414d5 fault_dump+0x67 0xe00dbc70: 0x1413bd PANIC+0x29 0xe00dbc80: 0x1be7cc _ZN5datab7freeRefEv+0x32 0xe00dbc90: 0x674ee3 _ZN11NFSv4_Owner9saveReplyERP4m sgb8nfsst at4P8n 0xe00dbcc0: 0x679376 _ZN11NFSv4_State5closeEmP8state id4P8nfs4 _ctxRP 0xe00dbd20: 0x6778fe _ZN16NFSv4_StateTable5CloseEmP8 stateid4P 8nfs4_ 0xe00dbd50: 0x6665e5 _Z10nfs4_closeRP4msgbS1_RPhR8nf s4_ctx+0x 18b Under heavy load, the quota usage value was incorrectly reported as being different between on-disk and in-memory. The error message in the server_log implied that the on-disk value was corrupted. 2011-01-24 18:11:47: 13157531648: UFS: 3:[SNXE-FS-02131] inode usage record is not consistent: on disk 2915057 in memory 2915058 type 1 id 32773 ouId 0 2011-01-24 19:28:15: 13157531648: UFS: 3:[SNXE-FS02131] Corrected the serialization of close operations to prevent this Data Mover panic. 401063 / 38621014
2
This error message will be printed only if there is an actual discrepancy between the ondisk and inmemory quota values.
402598 / 38854944
EMC Celerra Network Server Version 6.0.70.4 Release Notes
209
Version 6.0.42.3
1
During a FSCK of a file system, the Data Mover panicked if a block address contained the bad block flag rather than a real address. DART panic/fault message: >>PANIC in file: ../DeltaIO.cxx at line: 727 : DeltaIO_Read_CB::failed. The stack trace showed: 0xd939f430: 0x14124c waitForReboot+0x90 0xd939f450: 0x1414d5 fault_dump+0x67 0xd939f470: 0x1413bd PANIC+0x29 0xd939f480: 0xf9a361 _ZN15DeltaIO_Read_CB6failedE14B lockIO_St atus+0 0xd939f490: 0x8b3d33 _ZN9RawIO_IRP9irpFailedE14Block IO_Status y+0x21 0xd939f4b0: 0x8c8253 _ZN11HyperVolume17readBlockInte rnalEP10D iskIO_ 0xd939f510: 0xf99d69 _ZN15DeltaIO_Read_CB10doneCommo nEv+0x239 0xd939f570: 0xf99b1e _ZN15DeltaIO_Read_CB14getMappin gDoneEy+0 x82 A Data Mover panicked when a file system was mounted with the options: ntcredential and accesspolicy=NT or SECURE. DART panic/fault message: *** GP exception. Virt ADDRESS: 6d4599. Err code: 0 ***. The stack trace showed: 0x2c67d28: 0x141278 waitForReboot+0x90 0x2c67d48: 0x141501 fault_dump+0x67 0x2c67d68: 0x13a398 handle_fatal_fault+0xa 0x2c67e20: 0x6d440d _ZN15File_NamingNode8accessOKEm P17Access _Creden 0x34509e10: 0x64c1cd _ZN16worldAccess_List16printWor ldAccessE P9secF 0x34509ee0: 0x647afd _ZN17nfs_serverInitial5startEv+ 0xad7 0x34509fe0: 0x145c26 Sthread_startThread_internal+0x
FSCK will now skip the read when the block address has been flagged as bad.
393729 / 37862428
1
Code was performing an extra release on NT credential for UNIX in the case of file system mounted with ntcredential and accesspolicy=NT or SECURE. This has been fixed.
407971
210
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
e 1 When running FSCK on a corrupted file system, FSCK may cause memory corruption and a Data Mover may panic when FSCK tries to repair certain types of corruption. DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 351 : Memory corrupted. The stack trace showed: 0x23a81dc0: 0x14124c waitForReboot+0x90 0x23a81de0: 0x1414d5 fault_dump+0x67 0x23a81e00: 0x1413bd PANIC+0x29 0x23a81e10: 0x1fe8e7 free+0x69 0x23a81e40: 0x1fefb8 _ZdlPv+0x10 0x23a81e50: 0x1fefce _ZdaPv+0xc 0x23a81e60: 0x24880e _ZN13cMemoryBufferD1Ev+0x1e 0x23a81e70: 0x24a2fb _ZN11cMemBufListD1Ev+0x15 Server_2 hung and did not respond to commands, client did not have access to data. A failure to release all TCP streams as NFSv4 connections were being closed eventually resulted in there being no TCP streams available on the Data Mover to service commands or user I/O requests. For Redhat Linux, the nfs.v3xfersize transfer size was limited to 32K. The fix adds additional protection in FSCK code to avoid memory corruption. 399957 / 37582944
2
The fix ensures all TCP streams are released when NFS v4 clients disconnect.
412602 / 40332336
1
In order to have better NFS performance from Linux RedHat client running on VMware, the maximum nfs.v3xfersize is now tunable by using the param nfs v3xfersize=1310 72 command.
373951
EMC Celerra Network Server Version 6.0.70.4 Release Notes
211
Version 6.0.42.3
1
After a client rebooted, a Data Mover panicked with the following DART panic/fault message: Assertion failure: 'refLock.isExclusiveLocked()' on line 1691 in file '../nfsv4_state.cxx' >>PANIC in file: ../assert.c at line: 27 : assert failed. The stack trace showed: 0x11456ba70: 0x2a7d0b _assert+0x63 _ZN15NFSv4_ClientEnt7ConfirmEyP A8_KhR8nf s4_ct 0x11456bb10: 0x7dd53d _Z19nfs4_create_sessionRP4msgbS 1_RPhR8nf s4_ct 0x11456bc70: 0x792ec5 _Z9nfs4_linkRP4msgbS1_RPhR8nfs4 _ctx+0x16 87 0x11456be40: 0x81fea2 _ZN17nfs_serverInitial5startEv+ 0x1e40 0x11456bfd0: 0x16ad51 Sthread_startThread_internal+0x 7 An empty tree quota directory was removed by an unexport operation. This occurred if a name was specified when the export was created.
NFSv4.1create proper client mutex and prevents this condition.
403427
1
When processing an unexport, code now performs the remove operation only when the last component is a symbolic link and not a real directory. The fix allows early revocation of NFSv4 lock owners to free up resources, and avoid hang conditions. Corrected a deadlock that occurred when this param was set to zero and 1) server_export command was run, and 2)
388560 / 37329964
1
Under heavy NFSv4 load, NFSv4 clients may log "Remote I/O error" messages and responses from the Data Mover may be interrupted for 5-10 minutes. The network trace would show NFS4ERR_RESOURCE errors from the Data Mover. The error occurs because the code did not release NFSv4 lock owners correctly. When parameter mount.forceFullShowmount was set to 0, Data Mover MAC threads became blocked. GUI and CLI commands became unresponsive, and the following message was logged in the server_log periodically: ThreadsServicesSupervisor: Service:MAC Pool:MACExec BLOCKED for xxx seconds:
402884
1
403999 / 39088798
212
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Server operations may be impacted.
another process attempted to generate a list of FS mounts (such as the server_mount command) simultaneously. Added a threshold param cifs.acl.sdcompa ctthreshold. If a SD is larger than this parameter's value, code tries to compact it by removing redundant ACEs. The default is 16KB. This is to mitigate allocation problem in a fragmented memory. 404188 / 38935580
1
User set large security descriptors (SD) with redundant ACEs. When code tried to read those SDs after several days, it ran into a memory panic because the SD needed a large chunk of memory. Although there was enough memory, code could not grant memory allocation requests because of memory fragmentation. This caused the Data Mover to panic with the following DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory. The stack trace showed: 40bc10: 0x145966 _ZN19Memory_AddressSpace10alloc PagesEl12 Memory 40bc50: 0x24a34e malloc+0x11a 40bc90: 0x24ae1b _Znwm+0x1b 40bcb0: 0x24ae46 _Znam+0x4 40bcc0: 0xcf9e07 _Z11querySDInfoP13smb_threadCtx R11smb_Ou tMngt 40bdd0: 0xdd00e7 _ZN14smb2_threadCtx16replyQuery SDInfoEP4 msgb+ 40be70: 0xdd0bd0 _ZN14smb2_threadCtx14replyQuery InfoEP4ms gb+0x 40bea0: 0xdcbc33 _ZN14smb2_threadCtx7processEP4msg bi+0x68d Code was accepting non $DATA stream types, which are not compatible with NTFS. When trying to copy *.jpg files from a Celerra CIFS share to a local Window XP client, the following error was displayed: cannot copy xxx: The filename, directory name, or volume label syntax is incorrect.
2
Code no longer accepts non$DATA type ADS.
155357 / 31498524
EMC Celerra Network Server Version 6.0.70.4 Release Notes
213
Version 6.0.42.3
Install/Config/ Upgrades
The following entries describe fixed issues related to installations, configurations and upgrades.
Severity 2 Symptom Description Upgrade process might modify the value of some NDMP parameters as the maximum value of these parameters have been changed in newer releases. A new PUHC check is required to display a warning message regarding this modification. Fix Summary A new check has been included to warn user if any NDMP value would be deprecated during upgrade. With this fix, the /etc/nbs.conf file is updated at the same time as the connection setup. Any reboots after that, will include a successful nbs service restart. Code changes have been made to avoid the PAHC/PUHC failure when the hard drive in the Control Station is not in faulted condition. Tracking Number 394516
2
If a failure occurs during installation, or the system is rebooted at the beginning of the install process, the NBS service does not start properly at the next reboot.
401432 / 38569244
2
PAHCPUHC might fail with a hard drive failure error even when the hard drive is not in faulted condition.
------------Errors------------------Control Station: Check the hard drive in the control station Error HC_CS_14505083128: The 'smartctl -H' command failed. Action : Re-run "/usr/sbin/smartctl -H /dev/hda" to observe the output. Contact your system/network administrator to resolve the issue or escalate this issue through your support organization. -----------------------------------------
381917
The check was relying on the output of the smartctl command instead of the return value.
214
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity
Symptom Description
Fix Summary The fix is to stop audited for the duration of the upgrade, and restart the same with the pre-upgrade settings at the end of the upgrade.
Tracking Number 388554
1
Upgrade is seen to fail in the "Wait for NAS service to start" task. This occurred when a lot of access commands, to the file systems mounted on the Data Mover, were being issued from the Control Station while the upgrade was in progress.
NDMP
The following entries describe fixed Network Data Management Protocol/Backup problems:
Severity 1 Symptom Description VBB file level restore of offline/stub files did not work. When calling setAttr() for those files, VBB set file size which caused the file to be recalled and hence become online. Fix Summary Code was modified to set attributes without changing file size. The checkpoints created automatically by an NDMP backup are now removed Tracking Number 395232 / 37208224
2
Some NDMP backup automatic checkpoints were left behind and had to be removed manually. This occurred if the Celerra was very busy and the checkpoint creation completed after the timeout expired.
377919 / 35499712
Networking
The following entries describe fixed issues related to Networking.
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
215
Version 6.0.42.3
1
During a Data Mover reboot, it was possible that the SLIC did not power up and caused the Data Mover to fail bootup. This could lead to a data unavailability situation. Under rare circumstances, the Data Mover may not reply to all NULL requests from a client. This occurred if the transmit queue for the TCP session was full.
This was fixed by upgrading the firmware on the SLIC. The Data Mover will now reply to every NULL request, even if the transmit queue to that client is full. Code has been changed to remove the limitation of small listen queue. It is now unlimited and can be controlled with the param tcp.backlog. The fix adds support to remove stale IPv6 hosts from routing and neighbor caches and prevent this panic.
406453
1
404522 / 38052726
2
When having many simultaneous FTP connections to a Data Mover, some of the connections failed due to a small listen backlog queue. When this occurred, Data Mover logged the following error in the server log: Too many connections in port 21's listen queue.
405484 / 38895674
2
When using IPv6, the host cache table can monotonically grow as a result of expired host entries. These expired entries can make user route retrieval times exceed the watchdog timer and cause a Data Mover panic with the following DART panic/fault message:>>PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG. The stack trace showed: eee0: 0x14dd29 system_watchdog+0x61 eef0: 0x132913 t_nmi+0xc3 efd0: 0x155263 Sthread_startThread_internal+0x7
405168 / 39158304
NFS
The following entries describe fixed issues pertaining to Network File Systems.
216
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity 1
Symptom Description When a file system was close to 100% full, there was incorrect logic in the space allocation code which caused the Data Mover to allocate more space on the file system for a new writes than was available. This led to a Data Mover panic with DART panic/fault message: alloc failed: counters out of sync. The stack trace showed: 0xe1321840: 0x13fb94 waitForReboot+0x90 0xe1321860: 0x13fda9 fault_dump+0x67 0xe1321880: 0x13fc91 PANIC+0x29 0xe1321890: 0x7286c7 _ZN14UFS_FileSystem28markCorrup tedFsForU nmount 0xe1321a10: 0x737932 _ZN9UFS_Vnode5allocEP20UFS_Upda teDescrip torllj 0xe1321a50: 0x735cfc _ZN9UFS_Vnode10ballocListEP20UF S_UpdateD escrip 0xe1321af0: 0x73979a _ZN9UFS_Vnode15blockChainWriteE iP10Block Chaini 0xe1321bc0: 0x73910d _ZN9UFS_Vnode15blockChainWriteE iP10Block Chain+
Fix Summary The allocation logic code has been fixed to ensure this panic no longer occurs when a file system is close to full.
Tracking Number 159227
EMC Celerra Network Server Version 6.0.70.4 Release Notes
217
Version 6.0.42.3
Severity 1
Symptom Description While deleting a large file, an overflow in a file system data structure caused a memory corruption issue which resulted in a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000957f75 Err code: 2 Target addr: 0x1000000008 ** The stack trace showed: a760:0xbfb6b0 _ZN7DskDump9checkdumpEPKc+0x7a a7a0:0xbfb8be dump_return+0x4 a7b0:0x14c7d2 _ZN12Memory_Frame13setProtectio nEm19Memo ry_Prot a7d0:0x14ce5d fault_dump+0x89 a7f0:0x13378d handle_fault+0x1d 05610:0x886740 _ZN14UFS_FileSystem13getCgBlkEnt ryEj+0x10 05620:0x8a003a _ZN14UFS_FileSystem9hashallocEP 20UFS_Upd ateDes 05690:0x8a0335 _ZN14UFS_FileSystem8allocBlkEP2 0UFS_UpdateDesc When using NFS V4, a bug in the Linux clients made multiple clients look as a single client to the Data Mover. A race condition in this context caused the Data Mover to panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 1ce857 Err code: 0 Target addr: c8 **. The stack trace showed: 0x2c62c84: 0x141278 waitForReboot+0x90 0x2c62ca4: 0x141501 fault_dump+0x67 0x2c62cc4: 0x13a398 handle_fatal_fault+0xa 0x2c62d00: 0x67ca57 _ZN11NFSv4_State5closeEmP8state id4P8nfs4 _ctxRP4 0xe1059d20: 0x67b0d8 _ZN16NFSv4_StateTable5CloseEmP8
Fix Summary Code has been fixed to properly handle this condition and eliminate the panic.
Tracking Number 402066
1
This race condition has been eliminated.
411712
218
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity
Symptom Description stateid4P 8nfs4_ 0xe1059d50: 0x669c7d _Z10nfs4_closeRP4msgbS1_RPhR8nf s4_ctx+0x 18b 0xe1059d90: 0x665e30 _Z23nfs4_convertNfsv3Statusm+0x 626 0xe1059ee0: 0x647bdd _ZN17nfs_serverInitial5startEv+ 0xad7
Fix Summary
Tracking Number
1
NFS lock was not preserved upon a Data Mover failover.
Code now preserves the statd registry when file systems are unmounted by administrator.
407035
Replication
The following replication-related bugs that were fixed in this release:
Severity 2 Symptom Description If the Control Station timed out while waiting on a Replication Stop operation and retried, the operation hung and blocked further operations on the session. Fix Summary When the Data Mover receives a Stop request, it now checks whether the Replication Sessions is already in cleaning-up state and if so returns a TaskInProgress error. Tracking Number 404635
EMC Celerra Network Server Version 6.0.70.4 Release Notes
219
Version 6.0.42.3
Severity 1
Symptom Description When one checkpoint with the same snapVol ID as PFS Vol ID was inactivated, the PFS version was wrongly inactivated. User was unable to refresh or create any checkpoints on their main production file system. The following error was generated: Error: DpRequest_vs_InternalError.
Fix Summary The product version is now skipped when version for inactivated checkpoint is searched in inactivateVersi on. A flaw in index slot relocation process that can cause an inconsistency in checkpoint metadata has been fixed. The delete operation will not be accepted while the Replication session is in a clean-up state and the operation will fail with Abort in progress error to the user. A new mechanism was added to Data Mover code and Replication Manager code in order to handle slower response times under heavy load. Now Replication Manager can
Tracking Number 404437 / 39049562
1
Data Mover panicked with DART panic/fault message: PANIC in file:
../DeltaVol.cxx at line: 7585 : Value in TOC is different from the value in btree. After the failover, some
414584 / 40596048
checkpoints were inactive.
2
An attempt to delete a Replication V2 session hung if the command was issued while a failover operation was in progress.
402024
1
Under heavy load, in configurations with remote iSCSI Replication with many iSCSI LUNs and snaps, Replication Manager job may timeout due to the Data Mover not responding in time to iSCSI snap create request from Replication Manager.
396493
220
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.42.3
Severity
Symptom Description
Fix Summary query the status on Data Mover and if Data Mover cannot response fast enough, RM will retry the operation until it succeeds or until configurable time out value is reached. This fix is available in Replication Manager version 5.3.3 or later. Code now waits for the completion of the umount before taking a reference on the checkpoint file system. Code now returns TASK_IN_PRO GRESS directly to the Control Station to prevent the memory and reference leak.
Tracking Number
1
When a SnapSure SavVol becomes full, the oldest checkpoint needs to be unmounted and inactivated in order to free space. If a refresh of the oldest checkpoint occurs during this process it can result a deadlock, blocked threads, or a DART panic with the fault message PANIC: I/O not progressing.
399556
2
While using Replicator V2, if the ACK from the Data Mover failed/delayed to reach the Control Station, the Control Station retried sending the modify request. There was a problem in the code which caused a memory leak and cmd reference leak. Hence, there was an inflight task running on the Data Mover.
403651
Security
The following security-related bugs that were fixed in this release:
EMC Celerra Network Server Version 6.0.70.4 Release Notes
221
Version 6.0.41.4
Severity 1
Symptom Description Netgroups that are nested and begin with a special character were not being parsed.
Fix Summary Nested netgroups that begin with special characters are now fully supported. This issue occurred because the resource file contained a white space followed by the value. White spaces are now removed from the resource file for all account type values.
Tracking Number 402867
2
After creating the LDAP or storage domain user, the account type was displaying as "local" instead of LDAP in the GUI.
406555
Storage
The following entries describe fixed storage-related issues:
Severity 1 Symptom Description When a NAS was installed at location where SAN hosts had in-use FAST pools, Solutions Enabler (SE) code incorrectly parsed some float numbers as integer in the FAST pool data area. This caused SE discover failure and NAS install failure if there were existing SAN hosts with FAST pools. Error IM_14505279983 : Failed to synchronize with new backend storage Fix Summary This issue was fixed in SE code by changing from integer parsing to float parsing for several values. Tracking Number 400965 / 38569244
Version 6.0.41.4
Network File System
The following entries describe fixed issues related to Network File Systems:
222
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.4
Severity 1
Symptom Description When a checkpoint of the replication destination file system became corrupted and user was accessing the destination file system while replication data transfer was in progress, the read requests were redirected to the corrupted checkpoint. This caused a Data Mover panic with the following DART panic/fault message: D2320489_0: DeltaIO_Read_CB failed due to blk 16 corruption. The stack trace showed: 0x21a53f50: 0x14124c waitForReboot+0x90 0x21a53f70: 0x1414d5 fault_dump+0x67 0x21a53f90: 0x1413bd PANIC+0x29 0x21a53fa0: 0xfc2c9b _ZN22CkptCorruptMarkRequest13ha ndleReque stEv+0 0x21a53fc0: 0xfc2b26 _ZN23CkptCorruptMarkerThread5star tEv+0x80 0x21a53fe0: 0x145bfa Sthread_startThread_internal+0x e
Fix Summary Code no longer redirects read requests to a corrupted checkpoint.
Tracking Number 402304
Replication
The following entries describe fixed Replication issues:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
223
Version 6.0.41.4
Severity 1
Symptom Description If certain operations were performed on multiple Replication V2 sessions at the same time, the destination file system became out of sync with the source if a transfer was inprogress when the operation was initiated. The operations that can cause this issue are failover, stop, start -reverse and delete (if the session wasn't previously stopped). When one of these operations caused a transfer to be interrupted, the destination file system was restored to the state before the transfer is started. Only one of these restores could run at a time so while one was still in progress others failed and logged a message in the sys_log similar to: Apr 1 15:00:00 2011:DART:REP:ERROR:8:Slot 2:::1301670000:Destination=100_CK200064 400 000_0000_200_CKM00100200000_0000(ali as=fs1 ), restoring a checkpoint failed: DpRequest_vs_VersionExist. In some cases during bootup, the code might incorrectly decide that not enough space was reserved in order to recover a checkpoint. This can cause the Data Mover to panic and the checkpoints for the file systems to be inactive. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: fde91e Err code: 0 Target addr: 58 **. The stack trace showed: 0x2c62684: 0x141278 waitForReboot+0x90 0x2c626a4: 0x141501 fault_dump+0x67 0x2c626c4: 0x13a398 handle_fatal_fault+0xa 0x2c62700: 0xfbedcd _ZN19PoolManagedDeltaVol13start FlushLogE P10Delt 0x200816d0: 0xfbbabc _ZN19PoolManagedDeltaVol9indexS yncE18ind exSync 0x200817b0: 0xfd36ca _ZN10SnappedVol14notifyChildren EN7LVolum e8noti
Fix Summary The code has been fixed. Multiple file systems can now be restored to their state before the current transfer is started. This allows operations to be safely run in parallel for multiple Replication V2 sessions.
Tracking Number 406391 / 39554202
1
Code has been fixed to ensure checkpoints are recovered correctly.
399188
224
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description 0x200817e0: 0x8cee38 _ZN11HyperVolume14notifyChildre nEN7LVolu me8not 0x20081800: 0x8c7687
Fix Summary
Tracking Number
Version 6.0.41.3
CEE
The following fixed issues are related to Celerra AntiVirus Agent, Celerra Event Enabler, and Celerra Event Publishing Agent:
Severity Symptom Description Fix Summary Tracking Number 386093
1
When using Celerra Event Enabler, CIFS users were not able to access CIFS shares during multiple FS scans.
Created a new parameter, viruschk.fsscanburst, which defines the number of files sent in one burst by the scan process to the Virus Checker engines. The default is 20 files. The file system scanning will be suspended until the number of files reaches the value set. If the parameter is set to 0, the burst will be dynamically set to the number of AV thread times the number of online CAVA servers.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
225
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 388970 / 37306154
1
When stopping a virus checking service on a Data Mover while having a lot of unchecked files in virus checker queue, the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../sched.cxx at line: 846 SYSTEM WATCHDOG. The stack trace showed: 0xe6475ed4: 0x14124c waitForReboot+0x90 0xe6475ef4: 0x1414d5 fault_dump+0x67 0xe6475f14: 0x1413bd PANIC+0x29 0xe6475f24: 0x1432c9 system_watchdog+0x77 0xe6475f34: 0x139f6a _t_nmi+0x68 0xe6475f70: 0x71dcc6 _ZN15File_NamingNode14setChec kStatusEi+0 x34 0xe6475fb0: 0xd10d60 _ZN13check_initial5startEv+0x 19a 0xe6475fe0: 0x145bfa Sthread_startThread_internal+ 0xe When virus checking was enabled on a Data Mover, it caused failover to take additional time to unmount the Data Mover's root file system. This added approximately 1 minute to failover/ shutdown time.
Code has been fixed to properly handle the locks within the code and prevent this panic.
2
Virus checking no longer takes a reference on the root file system, so its unmount is not delayed during shutdown/failover. New CIFS client connections are no longer blocked by other unrelated connection attempts in progress. Code has been fixed and no longer tries to scan a file if it cannot
387885 / 37272266
1
The Data Mover was unable to send viruschecking requests to an AV server using MSRPC when a GPO update was trying to connect to a slow or nonresponsive domain controller.
390569 / 37333686
1
When using the nonotify mount option on a file system for which virus checker was configured, a deadlock condition occurred
395401 / 38038272
226
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
if a client tried to delete a file in the same directory where any other file was waiting to be scanned. 2 Following a manual check of a file using Symantec virus checking engine, the ctime of the file was changed. This caused a differential backup done by NDMP to be larger than expected.
get the path using the triggerset.
Code has been fixed to properly update the access time without changing the ctime, so it no longer affects differential NDMP backups. The Data Mover now removes files from the pending queue when this type of files is added to the exclusion list and virus checker is restarted.
400155 / 36478750
1
When a file type (.gz) was added to the exclusion list of the virus checker config file viruscheck.conf, the Data Mover still continued to scan these files. This issue occurred when .gz files were already in the pending queue to be scanned, and the filter was changed while the virus checker service was stopped. Once the service was restarted the Data Mover incorrectly sent the files in backlog queue to be scanned.
392790
Celerra Management Applications
The following are fixed problems related to the Celerra Management Applications, i.e., Celerra Monitor, Celerra Manager, WebUI, or Celerra MMC Snap-ins:
Severity Symptom Description Fix Summary Tracking Number 378575 / 36073712
2
User could not use Celerra Manager to change checkpoint schedule from a monthly to weekly recurrence. The following error was displayed: Modify schedule <schedule name>. The task scheduler is not responding. Please try again later.
Code was fixed to allow the modification of a monthly schedule to weekly schedule.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
227
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 404632
1
If a Data Mover was renamed (ex. server_2 was renamed to datamove2), the nas_alerterd could not monitor its CPU utilization. A Celerra system can only be managed by a version of Unisphere that has a Celerra plug-in of the same version or later. If Unisphere was downloaded from one of the Storage Processors in the array, an older version of the Celerra plug-in was installed and the 6.0.40-5 Celerra was not be manageable. A message of Unsupported or Newer Version was displayed. Rapid rebooting or restarting a Data Mover caused a process leak in the Control Station.
Fixed monitoring of Data Movers that are renamed.
1
Code has been fixed to allow all in-family management through Unisphere.
395531
2
Implemented a fix to properly close child processes to prevent process handle leaks, This issue occurred because VPFS file system did not have storageMB value and a null check was not performed. If the file system is of type VPFS, no details of the file system will be displayed. Added code to remove the trailing spaces from the serial value.
398424
2
When user queries for the capacity info or the capability info of the file systems, if there were checkpoints, XML API displayed the following error message and not the correct value: Unexpected exception of unknown nature.
384126
2
During an enumerate instance for an association, the SMI-S response received contained trailing spaces. There was no code to trim the APL returned value for serial. The device type 10 gigabit and the speed 10000FD were not added in the code. When APL returned this device type and speed, XML API displayed the value as other and incorrectly sent the files in backlog queue to be scanned.
385415
2
Added the new device type and the new value of speed to the code.
386449
228
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
3
The last boot time on the Data Mover properties page was not localized whereas the current time was localized.
Dates are no longer localized on the Data Mover properties page. The APL query attribute for fetching NTP server info contained an error which has been fixed. This issue was caused by a typo in an XML response name value which resulted in the setting not being correctly interpreted by the Celerra Management MMC. This has been fixed. Spaces in domain mapped users / groups are now supported in Unisphere.
381447
2
User could not see and confirm that an NTP server was configured. When CSA queried about ntpServers attached to a Data Mover, it did not list any, even though one was attached. User received the following error message when using the Celerra CIFS management snapin on the Data Mover: Internal Error: Unable to retrieve GPO setting from server. Please insure that the cifs server is still connected to the domain. Possible solution would be to rejoin the domain.
387693
2
394296
2
User was unable to add Domain Mapped Groups that contained spaces to Unisphere. When trying to setup LDAP with Unisphere to map domain groups, if user tried to add a group with spaces, an error was displayed which states only _ . are valid special characters. When a CIFS share spanned across multiple CIFS servers, XML API identified only one server and sent the request to APL. APL expected all the servers as the key, and returned the following error: Archiver::createDhsmConnections() Failed to query the file system of '<name>' for server '10.10.17.153'. Error Reason = 'XmlApi::recvXmlResponse Error.
396496 / 37607682
2
The code has been modified to check for share name and CIFS server name and return the information correctly.
137720
EMC Celerra Network Server Version 6.0.70.4 Release Notes
229
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 369551
2
On the Checkpoint Extend page, after selecting the Extend from Meta Volume option, the following message was displayed: No volumes are available' even though there were free meta volumes available. This occurred because meta volumes with the same disk type as the file system were listed in the dropdown on the Checkpoint Extend page rather than same disk type as checkpoint. Warning message was not generated and visible on the dashboard of the Unisphere manager or sys_log if FTP configuration was incorrect in /nas/site/automaticcollection.cfg file.
The GUI code has been modified so that filtered meta volumes have same disk type as the checkpoint and not the file system.
2
Added additional logging in the code path to generate the warning correctly if it cannot contact the remote FTP server. The MGFS state is now reported correctly.
403294
2
When the migration file system related migration encountered an error, for example, the migration source NFS server was down, its state was displayed as 'Idle' instead of 'Failed'. User created an nfs export with options using the CLI, for example, server_export server_3 –o root=root_hosts,rw=root_hosts:frhosts,access=root_hosts:fr-host s. When viewing the export through Unisphere, it was displayed as root:hosts instead of root_hosts. When the number of Celerras being managed was more than 5, the Celerra Capacity View Panel in the Unisphere dashboard showed incorrect information in the title and incorrect display order of the Celerras in the chart. Celerra/CUT plugin reported a different set of domain members upon startup than the CLARiiON plugin.
376681
2
Fixed the APL code that caused underscores to be replaced with colons.
399488 / 38452644
2
Changed the order in which the Celerras capacity details objects passed before updating the view.
372223
2
CimomAccess.pm treated the IPs in reverse order as a new entry and added it to domain_list file. The
374045
230
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
CimomAccess.pm file was modified to make sure it identifies IPs in reverse order as well and does not populate the domain_list file. 2 When using Unisphere and clicking on a list page button (e.g. Delete, Properties, etc.) user waited for the expected Confirm Delete Dialog page to pop up. Occasionally, the Confirm dialog popped up for a split second then disappeared behind the Unisphere window. On a slow or busy client, this happened more often. The creation of user quotas from the root of a file system could not be performed from the GUI, but worked fine when using the CLI. The option "none" was missing from the user and group creation pages in Unisphere. The Run CLI command in Unisphere was not localized. There were no leading/trailing character strings for the text and none of the buttons were converted. The checkpoint delete popup in Unisphere came up blank when running with a language pack enabled. Unisphere code was fixed so that the action pop-up window remains in the foreground above the main window. 378218
2
Code was modified to include the "none" option.
379661
2
This has been fixed, and the Run CLI command is now localized.
381117
1
Code has been modified to check for the localized version of the string. This issue occurred because APL did not send the unit as Megabytes (MB). This has been fixed.
394269
2
User could not modify the Maximum Storage Capacity through Unisphere, but could modify it using the CLI. The following error was returned: invalid size specified.
382336
EMC Celerra Network Server Version 6.0.70.4 Release Notes
231
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 379910 / 36289838
1
In Celerra Manager Group Quota page users were unable to search quotas by GID or group name.
Search by GID and group names capability was added to Group Quota page of Celerra Manager.
CIFS
The following entries describe fixed issues related to Common Internet File Systems:
Severity Symptom Description Fix Summary Tracking Number 406316/ 39423800
1
The Data Mover panicked and with a Page Fault Interrupt when a pre-released version (BETA) of MACOS X 10.7 (Lion) client accessed a CIFS share. The stack trace showed:
0x2ab3a08: 0x13fb44 waitForReboot+0x90 0x2ab3a28: 0x13fd59 fault_dump+0x67 0x2ab3a48: 0x138d68 handle_fatal_fault+0xa 0x2ab3a70: 0xb22430 _ZN21cMsRPC_LsarpcSimulate13lsa rProcess2 DEP13sm 0xc3745b60: 0xb29164 _ZN21msrpc_legacyOperationI21cM sRPC_Lsar pcSimu 0xc3745b90: 0xc9ffd1 _ZN22msrpc_interfaceHandler13ca llOperati onEtR1 0xc3745bf0: 0xc9edb5 _ZN15msrpc_serverCnx15_processR equestER2 3msrpc 0xc3745cc0: 0xc9df72 _ZN15msrpc_serverCnx10processMs gEP13smb_ thread
The code was modified to correctly handle the LSA Lookup call 0x2D and avoid this Page Fault condition
1
When accessing NAS files from Windows, the count of hard links was incorrect. When using the perl stat command on a Windows file, the link count on the file was always 1.
Code now returns the correct link count for the file (note: for directories, the link count is always 1).
392782
232
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
2
If a new file filter file was created with Windows XP or Windows 2003, it worked as intended for users with the same systems, but not on Windows 7 or Windows 2008 . For these newer Windows versions, files could be created but could not be deleted nor edited. During a domain SID migration, an invalid SID in a Security Descriptor (SD) caused a NULL referencing condition. This led to a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: bf1d26 Err code: 0 Target addr: 2 **. The stack trace showed: 0x2acfc64: 0x13fb94 waitForReboot+0x90 0x2acfc84: 0x13fda9 fault_dump+0x67 0x2acfca4: 0x138d94 handle_fatal_fault+0xa 0x2acfdb0: 0x7a9656 _ZN11AclDatabase5parseEP23Par seAclDataba seConte 0x56b7ff60: 0x7a8eed _ZN11AclDatabase5parseEP23Par seAclDataba seCont 0x56b7ffb0: 0x7a84c1 _ZN23ParseAclDatabaseContext5s tartEv+0x6 9 0x56b7ffe0: 0x144344 Sthread_startThread_internal+ 0xe
Code has been fixed to deny creation of files based on file filtering rules for Windows7/ Windows 2008 clients.
395700 / 37640396
1
Code now ignores the invalid SDs during a domain SID migration and the migration continues.
391200
EMC Celerra Network Server Version 6.0.70.4 Release Notes
233
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 384678
1
When using DHSM CIFS connection, recall of some files failed with the following: MGFS: 3: error direct read block 7 of entry 4248, status 20 – EndOfFile. The issue occurred because a secondary CIFS server returned a nonzero size for getAttr request, but a zero size for Open request. Copying a file over a DFS link failed from an SMB2 client, reporting there was not enough space, even though there was plenty of space on the target of the DFS link. This is because the size of the DFS root file system was being checked rather than the target of the DFS link.
Code now handles cases where Open and getAttr request sizes do not match.
2
The Data Mover now sets the REPARSE_POINT file attribute in the DFS link node. This attribute directs the SMB2 client to query for file system size of the DFS link target. Use correct CIFSbased SPN to access Domain Controller.
401201 / 38573658
1
Unable to access server when Microsoft security policy in raised Windows 2008 R2 domain. See Microsoft article: 2345886, Description of the update that implements Extended Protection for Authentication in the Server service,
393548 / 37839090
http://support.microsoft.com/defaul t.aspx?scid=kb;en-US;234588 6
2 When migrating SIDs from one domain to another using server_cifs –Migrate command, unknown SIDs were not migrated. The sever_cifs Migrate command has been fixed, and is now will include unknown SIDs. Permission inheritance is now handled correctly for "Files Only". Added support for share information through SRVSVC pipe, granted access to unprotected functions, 387781 / 37493702
1
When setting permissions from a Windows client, the special permission "Apply onto" as "Files only" did not get propagated to child directories. User experienced slow performance with Windows7 client using MSOffice applications when Celerra CIFS server was configured with UNIX security. Access to shares enumeration returned
383935 / 36713302
1
389804 / 36737358
234
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
an internal error status, causing delays on the client side. 2 When using CIFS auditing and only enabling events for Read operations, Data Mover generated logs for ALL events, regardless of configured events. The issue occurred because of an error in the mechanism that checks if an audit event log must be generated. Under certain stress circumstances, the Domain Controller replied to LSARPC requests with an error: STATUS_INSUFFICIENT_RESOURCES. The Data Mover kept attempting to use this DC, never getting any information. Server log contained errors like: LsarOpenPolicy2::Request failed, status=INSUFFICIENT_RESOURCES. From the client standpoint, SID was displayed instead of clear text names when attempting to get ACL content. User experienced experience slow performance with Windows7 client using MS Office applications when Celerra CIFS server was configured with UNIX security.
like shares enumeration. The error has been corrected, and the Data Mover will now generate logs only for specified events. 384108 / 36692600
1
The connection with the overloaded Domain Controller will be dropped. The Data Mover will now reconnect to another Domain Controller.
382103 / 36576304
1
Added support for share information through SRVSVC pipe to accommodate pipe paths with ASCII names. The code has been fixed to allow widelinks when SMB2 is enabled. The code now allows the single SMB thread timeout without holding a lock on all other SMB threads.
383876 / 36737358
2
When SMB2 was enabled on Data Mover, widelinks stopped working, and all links appeared as files in Windows Explorer.
381269
1
When using CEPP/CAVA over MSRPC and a CEPP or CAVA server did not respond to an SMB thread, other SMB threads were locked awaiting the timeout of the CEPP/CAVA client. This appeared as hangs or slow access for users.
379264 / 36219350
EMC Celerra Network Server Version 6.0.70.4 Release Notes
235
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 380397 / 36177168
2
Attempting to copy with EA (NTFS extended attribute) from a Window machine to Celerra using SMB2 (Win7/W2K8-R2 client and SMB2 enabled on Celerra) fails with popup message: The extended attributes are inconsistent or The file or directory is corrupt or unreadable.
Celerra does not support EA but the status being returned on SMB2 was not appropriate (STATUS_NO_EAS_O N_FILE). Celerra now reports STATUS_EAS_NOT_ SUPPORTED. Also, the parameter used to make Celerra ignore EAs and not return an error (cifs set_eas_ok) now applies to SMB2. The timeout value for delete CIFS server request to the Data Mover has been correctly set to 180 seconds.
1
When deleting a CIFS server ,the timeout of the Control Station command must be set to at least 120 seconds. Instead, it was set to 50 seconds. In some cases 50 seconds was insufficient for DART to perform the deletion. As the result, the Control Station timed out and failed to delete a server from netd and boot.cfg. Upon Data Mover reboot, the deleted CIFS server was unexpectedly recreated on the Data Mover. If the server_thread command was used with -all, the Data Mover panicked with the following DART panic/fault message: *** GP exception. Virt ADDRESS: 0x00006f3ed1. Err code: 0 ***. The stack trace showed: 0x3685760: 0xbf93d0 _ZN7DskDump9checkdumpEPKc+0x7 a 0x36857a0: 0xbf95de dump_return+0x4 0x36857b0: 0x14c7d2 _ZN12Memory_Frame13setProtect ionEm19Memo ry_Prot 0x36857d0: 0x14ce5d fault_dump+0x89 0x36857f0: 0x13378d handle_fault+0x1d
374845 / 36219350
2
NFS threads now correctly display the last operation executed.
390668
236
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
0x106585940: 0x15a5bf _ZNK15cThreadActivity33displa yIdleThread Info_ 0x1065859f0: 0x6f3e8e _ZNK17nfs_serverInitial21disp layIdleThre adInf 0x106585a00: 0x15af4a _ZN15cThreadActivity11parseTh readER12cSv cPars 1 User could not access a server from an NT4 machine (or any client like Samba using old NTLM SessionSetup request). This occurred when NTLMV2 was enforced with: LAN Manager Auth Level: Send NTLMv2 response only, refuse LM and NTLM NTLM SSP Minimum Client Security: Require NTLMv2 session security NTLM SSP Minimum Server Security: Require NTLMv2 session security NT4 client setup: LAN Manager Auth Level on NT 4 client configure to “Level 2 – Send NTLM authentication only”. The following entries were observed in server_log: 2010-07-30 10:22:55: 13157007360: SMB: 3: Client=customerClient NTLMV2 authn required 2010-07-30 10:22:55: 26041909248: SMB: 6: Client=customerClient killed by NTLMV2 required 2010-07-30 10:22:55: 26041909248: SMB: 6: SessSetupX failed with RPC_NT_UNSUPPORTED_AUTHN_LE VEL Code now allows authentication requests to be sent to the domain controller even if it is suspected to be NTLMV1, based on size. 386340 / 3620668
EMC Celerra Network Server Version 6.0.70.4 Release Notes
237
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 388629
1
A Data Mover panicked or the viruschecker stop operation hung when viruschecker was restarted multiple times. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000155833 Err code: 2 Target addr: 0x0000000008 **. The stack trace showed: 0x3686760: 0xbf90b0 _ZN7DskDump9checkdumpEPKc+0x7 a 0x36867a0: 0xbf92be dump_return+0x4 0x36867b0: 0x14c7d2 _ZN12Memory_Frame13setProtect ionEm19Memo ry_Prot 0x36867d0: 0x14ce5d fault_dump+0x89 0x36867f0: 0x13378d handle_fault+0x1d 0x1af6be20: 0x2069bd _ZN8RWlockRT10sharedLockEj+0x 65 0x1af6be60: 0xf72de4 _ZN12virusChecker5mountEP14VF S_FileSyste m+0x20 0x1af6bf00: 0xf6bb96 _ZN12check_survey5startEv+0x4 16 With Access Based Enumeration (ABE) enabled, some DFS links may not appear in directories even though the user has the right to see them.
Code now waits until the end of the viruschecker survey thread before restarting viruschecker. Also removed an extra shared lock that led to this deadlock condition.
2
The problem is now fixed and a user can see all DFS links if he has the right to see them. The offline flags of the CIFS share are now properly returned to the SMB2 clients.
392030 / 37302982
2
The offline function on a CIFS share did not work as expected for Vista and Windows 7 clients. When the offline setting was disabled on a CIFS share, a Vista or Windows 7 (SMB2) client still could use offline function on this folder. An out-of-memory panic occurred when using large credentials (containing more
386604 / 36952502
1
Code now allocates less contiguous
388145 /. 37290766
238
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
than 1500 groups from the forest) for NTLM authentication.DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 289 : Out of memory. The stack trace showed: 0xf84bd310: 0x13fb94 waitForReboot+0x90 0xf84bd330: 0x13fda9 fault_dump+0x67 0xf84bd350: 0x13fc91 PANIC+0x29 0xf84bd360: 0x13a8f2 _ZN19Memory_AddressSpace10all ocPagesEi12 Memory 0xf84bd380: 0x1f6c3a _Z11more_memoryjP9free_list+0 x4a 0xf84bd3a0: 0x1f6cf8 malloc+0x98 0xf84bd3d0: 0x1f7485 _Znwm+0x15 0xf84bd3e0: 0x1f74b2 _Znam+0xc
memory to complete user's NTLM authentication.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
239
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 383732
1
A memory corruption in a Kerberos context was used by a function that then returned a NULL pointer. The pointer was not checked before being passed into another function, which then tried to dereference the pointer, resulting in a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 1fa53e Err code: 0 Target addr: 0 **. The stack trace showed: 0x2acb158: 0x13fb94 waitForReboot+0x90 0x2acb178: 0x13fda9 fault_dump+0x67 0x2acb198: 0x138d94 handle_fatal_fault+0xa 0x2acb1c0: 0x56e267 krb5_cc_resolve+0x17 0x1eca92b0: 0x56e609 krb5_cc_default_suffix+0xa3 0x1eca92e0: 0x53e1db _Z26krb5_gss_release_principa lPjPPv+0x46 b 0x1eca93a0: 0x53ed90 _Z25krb5_gss_acquire_cred_ext PjPv22gss_b uffer_ 0x1eca9400: 0x5456ff gss_acquire_cred_ext+0x4b When the .etc of a share's file system was missing, each time the share was mounted for the first time, a message similar to the following was logged to the server_log: Cannot get .etc dir (28=InternalError). When trying to set ACL to a file which had an empty DACL (Discretionary Access Control List) using emcsetsd, the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: /disks/NSG_Root/Maint_4/5.6_bui ld_nfs_env/ Dart/
The fix is to check the return value of the pointer, and return an error if the value is NULL.
3
This message was informational only, as the code was correctly handing the issue. This message is now hidden from the user. Before the fix, code was overwriting the header of the ACL. Code now checks if the file has an empty DACL, and write the new ACL after the ACL header.
393943 / 37873442
1
397384 / 38188396
240
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
server/src/cfs/include/acl.hxx at line: 1522 bad SD storage size. The stack trace showed: 0xde7617e0: 0x141254 waitForReboot+0x90 0xde761800: 0x1414dd fault_dump+0x67 0xde761820: 0x1413c5 PANIC+0x29 0xde761830: 0xe824cc _ZN14set_celerra_sd7processEv +0x336 0xde761910: 0xe826e9 _Z21cifs_set_celerra_sd_1P27c ifs_set_cel erra_s 0xde761990: 0xe85ab9 _ZN29xattrp_IterSvc_Parent_In itialC1EP16 Rpc_Sv 0xde761a40: 0x4a9092 _ZN21Rpc_Svc_Child_Initial13s vc_getreqse tEP7rp 0xde761c30: 0x4ac16f _ZN25Rpc_IterSvc_Child_Initia l7svc_runEv +0x10b
EMC Celerra Network Server Version 6.0.70.4 Release Notes
241
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 374292
2
User could not install CelerraCifsMgmt on a Windows 64-bit OS.
Removed the restriction in the CelerraCifsMgmt install package that prevented user from installing it on 64-bit system. The following message is now displayed when installing CelerraCifsMgmt on a 64-bit system: You are now installing 32-bit CelerraCifsMgmt on a 64-bit Operation System. Celerra UNIX property page extensions are only available in Active Directory Users and Computers on 64-bit Operation System through the following instructions: 1. From the Start menu, select "Run ..." 2. In Open field, input the command: mmc /32 dsa.msc 3. Click the OK button to launch the Active Directory Users and Computers The fix respects the precedence of the DELETE_PENDING on the SHARING_VIOLATIO N status. Added /secforce option that force emcopy to overwrite destination security and prevent this condition.
2
A user application might not work properly when using temporary files on a CIFS share. The Celerra returned a SHARING_VIOLATION instead of DELETE_PENDING.
391399 / 37403078
2
During data migration from Windows 2003 file servers to CIFS servers on Celerra using EMCOPY, the following error appeared: EMCOPY: permission changes either copies or does not copy files from
393429 / 37414066
242
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
source to dest. 2 Using the nonotify mount option did not prevent notifications of changes to a directory from being sent to CIFS clients. Also, the option could interfere with directory renames. The nonotify mount option now disables CIFS notifications and does not interfere with directory renames. 395058 / 37393392
Control Station
The following entries describe fixed problems related to the Celerra Control Station.
Severity Symptom Description Fix Summary Tracking Number 144684 / 29744606
1
When an NBS session (which included failovers) was started on the Control Station, a race condition occurred in the NBS driver that led to a Control Station kernel panic and reboot with the following message: Kernel panic - not syncing: Fatal exception. The server_ldap -lookup command did not query groups containing spaces in the name.
To fix this issue, a locking mechanism was added to prevent this race condition.
2
The Control Station now correctly processes LDAP queries containing spaces in the group name. With the fix the command now rejects incorrect parameters and displays command usage.
374688 / 35603200
3
When a nas_quotas report command was entered with incorrect parameters, the command returned the following error message instead of the command usage: nas_quotas: Warning 17716809847: server_2: NIS Error: UID= <file system>.
376711 / 35626866
EMC Celerra Network Server Version 6.0.70.4 Release Notes
243
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 380335 / 36355420
2
If the interface name of the server_cifs command ended with CRLF which is "carriage return" on Windows , the Eiffel trace and vdm.cfg or netd table was corrupt. Management daemon crashed while attempting to allocate memory in order to service an XML message, causing the core dump to fill up the /nbsnas partition.
Code now truncates the CRLF character before doing transaction.
2
In a rare case where the memory is fragmented to the point that the service of an XML message was unable to complete, a retry mechanism was added to re-attempt to allocate the memory. Memory is now recovered in the background. Code has been fixed to verify the username and password correctly. Code now correctly detects the components present in the enclosure.
386399 / 36932966
2
User was denied permission error if the username exceeded 8 characters while executing the new CLI which needed special admin role privilege. When a VG8 system was configured with 3 Data Movers, a false alert that the power supply was removed from nonexistent Data Mover was reported. 0x330070051 Error: Local emc70pri common DME 1 Power Supply B0: Removed. If user specified a storage system for a file system extension and the storage system was different than what the file system already exists on, they might get storage from the wrong storage system.
377427 / 35919230
1
397608 /; 38148792
2
If a storage system is specified in file system extension, code now only takes storage from that storage system for the extension. When creating/refreshing checkpoint task from APL scheduler fails, only one error will be
378214
3
When creating or refreshing a checkpoint task from APL, if the scheduler failed, both APL task manager and APL scheduler generated the same error: Create checkpoint ckpt_alert_sched_001
378439
244
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
of schedule alert_sched has failed. 3 When running a dbchk -f -q command, the following error was logged in the sys_log: A Celerra database error was detected. An inconsistency in the Celerra database files was detected. Run the $NAS_DB/tools/dbchk tool to identify the exact error. However, when user ran the dbchk tool, the output was clear: [nasadmin@apoh0010cs03 ~]$ /nas/tools/dbchk -pvwx dbchk: done. The Data Mover reported too many errors in server log for software correctable errors and generated SMI (system management interrupt). The system clock got out of sync while the SMI was being processed, which resulted in a Data Mover panic with the following DART/panic fault message: >>PANIC in file: ../sched.cxx at line: 1541 : SYSTEM WATCHDOG. The stack trace showed: 0x3635a40: 0x14c81d _ZN12Memory_Frame13setProtect ionEm19Memo ry_Protection_t+0x23d 0x3635a60: 0x14ce29 fault_dump+0x55 0x3635a80: 0x14cf02 PANIC+0x5a 0x3635aa0: 0x14dd29 system_watchdog+0x61 0x3635ab0: 0x132913 t_nmi+0xc3 0x3635b68: 0x378674 _ZN12bcmnicDevice12bcmInterrupt EPv+0x5a 0x3635b98: 0x1d338f _ZN3ISR6invokeEv+0x2d 0x3635bb8: 0x1d3e91 shared_isr+0x207
displayed. If the nas command finishes within 10 seconds, dbchk will no longer report an error (unless nasdb finds some inconsistency problems). 395926 / 38096334
1
Fixes the PCI correctable error handler for devices on the motherboard so it correctly prints warning messages instead of panicking the system and falsely indicting the CPU module.
374413
EMC Celerra Network Server Version 6.0.70.4 Release Notes
245
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 389936
1
Under heavy system load, a large amount of data frames were dumped, causing dumpsize to exceed 24G and dump_slot not showing the right size. This could result in the dump being corrupted, or important data could be missed from the dump. When each checkpoint schedule was fired off, APL scheduler checked its next execution time. If the next_execution_time + 60 seconds < current time, this schedule was marked as miss, and the checkpoint did not create/refresh. If too many schedules fired off at the same time, some of them did not create/refresh the corresponding checkpoint. The server_ifconfig server_X -a command did not display netname for IPV6 interfaces. Netnames were displayed for IPv4 interfaces only.
Code has been fixed to skip the buffer data frames when not required and calculate the correct data dump size.
2
Increased the timeout value from 60 seconds to 5 minutes to allow more schedules to run at once.
384138 / 35719928
2
Corrected the query sent to the Data Mover ensure it displays both IPv6 and IPv4 netnames. Code now rejects an attempt to change the physical Data Mover's name to the same name as the VDM's. This memory leak condition has been eliminated.
384477
2
User changed the physical Data Mover's name to the same name as the VDM using the server_name command. This caused several server_xxx commands to not function properly. Using the nas_dbtable with a filter on a field which was not the primary key, or intensive use of the nas_dbtable caused the Data Mover to run out of memory
385998 / 37053750
2
394952
CORE
The following entries describe fixed problems related to Celerra Data Movers.
246
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 16063 / 32384702
1
Commands against the Data Mover failed with Error 5: server_4 : Input/output errorTimed out or the Data Mover did not respond to the command at all because the MAC threads were hung in an unmount deadlock. [root@A00DRFL003 log]# server_date ALL server_2 : Tue Dec 1 10:04:11 GMT+1 2009 server_3 : Tue Dec 1 10:03:43 GMT+1 2009 server_4 : Error 5: server_4 : Input/output errorTimed out server_5 : Tue Dec 1 10:04:41 GMT+1 2009
ReclaimFreeNode and giveUp are now called directly without checking for a useCount of zero.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
247
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
A memory leak in the rpcbind_getaddr() function of the code caused a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../bufpool.cxx at line: 389 : couldn't get a free page. The stack trace showed: 0xcd2ab8f0: 0x13f244 waitForReboot+0x90 0xcd2ab910: 0x13f441 fault_dump+0x67 0xcd2ab930: 0x13f341 PANIC+0x29 0xcd2ab940: 0x17d7ef _ZN18Buffer_Cached_Pool17make CachedBuffe rsEi+0 0xcd2ab970: 0x13b4ed _ZN19Memory_AddressSpace16mak eCachedBuff erEv+0 0xcd2ab990: 0x17d37d _ZN18Buffer_Cached_Pool15getA nyCachedBuf Eji+0x 0xcd2ab9c0: 0x6b1ec3 _ZN14UFS_FileSystem8getBlockEliRPv+ 0x41 0xcd2aba00: 0x6b1629 _ZN14UFS_FileSystem15readBloc ksAsyncElii PP13Bu
The code has been changed to avoid the memory leak.
380233
CSA
The following entries describe fixed issues pertaining to the Celerra Start-up Assistant.
Severity Symptom Description Fix Summary Tracking Number 379078 / 36174238
2
While configuring the Control Station through CSA, if user tried to set the CS and blade time to (GMT +5:30) Asia/Calcutta (as shown in the drop-down menu). The "Apply Changes" step in CSA failed with NTP server, time zone and unicode properties set up on primary
Time zone information has been changed in the CSA to reflect the updated spelling of Kolkata.
248
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
blade... - Invalid time zoneccmdEscColon Asia/Calcutta. 1 The site ID that was entered by CSA incorrectly used double quotes and was not visible in Celerra Manager or Unisphere. Code was fixed to ensure users can see the information in Celerra Manager or Unisphere. Code was altered to send these CLARiiON events to the sys_log. 361995
2
CLARiiON events for NX Systems with AX4-5F8 backends were passed to /var/log/messages but were not passed to sys_log as expected. When the user tried to upgrade a Fibre Channel (FC) enabled systems from version 5.6 to 6.0 or tried to install integrated systems (running NAS and FLARE), the Celerra Startup Assistant failed to perform the Celerra and CLARiiON system domain join operation. In the Celerra Startup Assistant (CSA), during Control Station install, the Gateway field was not indicated as a required field.
393090
1
The Celerra Startup Assistant was modified to perform a domain join for FC-enabled systems.
382973
2
The Gateway field is now marked as a required field.
361899
Deduplication
The following entries describe fixed issues pertaining to Celerra Deduplication:
Severity
Symptom Description
Fix Summary
Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
249
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 378549
2
Error messages similar to the following were incorrectly logged as ERROR in the server_log when mounting the FLR file system for the first time: Error 2010STORAGENumpaths option u, E EDT 2010FSTOOLSFLR data popula, Error T 2010FSTOOLSflr_clock = -1, flr_max_time. During a read operation of a file that has recently failed deduplication through the CIFS compression or DHSM API methods (via VMWare plugin), the Data Mover panicked with the following DART panic/fault message: isAValidIndBlkBuf: bad bn.The stack trace showed: 0x271b1450: 0x141254 waitForReboot+0x90 0x271b1470: 0x1414dd fault_dump+0x67 0x271b1490: 0x1413c5 PANIC+0x29 0x271b14a0: 0x77663b _ZN14UFS_FileSystem28markCorr uptedFsForU nmount 0x271b1660: 0x78ccbe _ZN14UFS_FileSystem17isAValid IndBlkBufEP 13Buff 0x271b16e0: 0x782d47 _ZN9UFS_Vnode17readIndirectBloc kElRP13Bu ffer_C 0x271b1710: 0x7852c2 _ZN9UFS_Vnode8bmapListEmmPmPi+0 x228
These errors were informational only, and are no longer displayed.
1
Code was updated to ensure data is properly flushed to disk before deduplication is started via the CIFS compression method or the DHSM API method (via VMWare plugin) to prevent this panic.
378009 / 36040202
1
There is a deadlock situation in the FRDE re-ingest code that may be hit if a Data Mover panics and reboots or a file system is re-mounted while having entries in the re-ingest queue. The deadlock will prevent files in the re-ingest queue from being processed. This may cause the reingest queue to become filled and reject new entries.
Code has been fixed to prevent this deadlock condition.
379193
250
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
1
When using NDMP backup on file systems that have deduplication enabled, NDMP backups could take more space than necessary, and NDMP jobs could fail with the following error messages in the log DEDUPE: 3: EDRS: RP CSX COMPRESSOR: internal error decompression failed: 2 DEDUPE: 3: EDRS: TRANSFORMING FILEDATA UNIXFSIS: readBlob: transformOut failed at offset 3244032: RDE_STATUS_DECOMPRESSION_ERROR. Because of a limitation in the compressor module of code a failure could occur if this code receives more than 2 requests for decompression at a time. When using Deduplication, if many CIFS clients were performing directory scanning on F-RDE enabled file systems, CPU utilization on Data Mover reached 100% and impacted performance. The issue occurred because code in deduplication policy engine (PE) was not optimized. There was a race condition between removal of deduplicated files and file system unmount operations. Under certain conditions, deduplicated data space was not reclaimed from the file system when files were removed. A message was displayed in the server log similar to: 2010-05-09 09:40:57: 13163036672: DEDUPE: 3: Cannot find file system with id 15, cannot cleanup object <dart://rde ...
Code now ensures that compressor module code does not receive more than 2 requests for decompression at a time.
381922 / 36502216
1
PE code has been optimized and will not cause 100% CPU utilization under these conditions.
385163 / 36929424
2
If you attempt an unmount during removal of deduplicated files, the code now waits for threads to complete before performing the unmount.
367398
EMC Celerra Network Server Version 6.0.70.4 Release Notes
251
Version 6.0.41.3
File system
The following entries describe fixed file system issues:
Severity
Symptom Description
Fix Summary
Tracking Number 387997
1
When a file system was mounted on a VDM, if the tree root’s inode number was the same as the inode number of the mount point to which that VDM was mounted, renaming such tree quota from the client resulted in “/..” path reported. The cylinder group bitmap initialization process is started asynchronously for each file system as they are mounted on the Data Mover. If a file system was unmounted or the Data Mover was rebooted, failed over, or failed back before this operation completed, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 7831ec Err code: 0 Target addr: 2c **. The stack trace showed: 0x2ba5e34: 0x140cec waitForReboot+0x90 0x2ba5e54: 0x140f71 fault_dump+0x67 0x2ba5e74: 0x139eec handle_fatal_fault+0xa 0x2ba5ee0: 0x1c35ab _ZN17SvcThread_InitialIS_I15St hread_Init ialEE5s 0xe6ca1fe0: 0x1455d4 Sthread_startThread_internal+0xe
Code has been fixed to properly handle this condition so the rename of quota tree will be completed as expected. The cgbitmap initialization is now aborted if the file system state is unmounting, unmounted, frozen, or corrupt. It will be reinitialized when the file system is remounted.
1
357994
1
Data Mover denied write I/O with NFS4ERR_ACCESS instead of NFS4ERR_LOCKED in when a file is locked. When a file was being truncated through setAttr or setLength calls and any panic or unexpected reboot happened during the
NFS threads now return the correct error code.
392703 / 37513228
1
Code has been fixed to properly handle SetAttr/setLenght logic
382208
252
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
same time, the Data Mover panicked when attempting to access that file. DART panic/fault message: isAValidIndBlkBuf: bad bn Panicking Processor: 3. The stack trace showed: 0x79b313f0: 0xa6bb4c _ZN14UFS_FileSystem28markCorru ptedFsForU nmount 0x79b31560: 0xa927dc _ZN14UFS_FileSystem17isAValidI ndBlkBufEP 13Buff 0x79b315e0: 0xa83009 _ZN9UFS_Vnode17readIndirectBlo ckEiRP13Bu ffer_C 0x79b31630: 0xa8407d _ZN9UFS_Vnode4bmapEiRiPiS1_+0x1c 1 0x79b316a0: 0x18c38a1 _ZN10DHSM_Vnode14itruncInterna lEP20UFS_U pdate 0x79b31710: 0xa7fbd1 _ZN9UFS_Vnode6itruncEP20UFS_Up dateDescri ptoryi 0x79b317a0: 0xa80208 _ZN9UFS_Vnode8inactiveEPvi11Fi le_Statusi i+0x47 0x79b31820: 0x1893362 _ZN10MGFS_Vnode8inactiveEPvi11 File_Statu sii+0 1 A Secure NFS mount from a client could caused a Data Mover panic with the following panic header: Virt ADDRESS: 0x00006dd730 Err code: 0 Target addr: 0x000000000c.
to prevent the panic under this condition.
Code was updated to prevent the panic and ensure Secure NFS mounts succeed.
398127
EMC Celerra Network Server Version 6.0.70.4 Release Notes
253
Version 6.0.41.3
1
During file system operations, code improperly passed a handle to a different function. Later the handle has changed, but the function that this handle was previously passed to did not realize this, and used the old handle which led to memory corruption. This resulted in the panic with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 525 : Memory corrupted. The stack trace showed: 4056c0: 0x24aa1f free+0x79 405700: 0x24aded _ZdlPv+0x9 405710: 0x9682b9 _ZN20UFS_UpdateDescriptor8comp leteE11Fil e_Sta 405820: 0x8a1c9f _ZN21aggregateCommitExtent6doWor kEP9VFS_Vn ode 405870: 0x930b47 _ZN13aggregateWork9writeListEP10 gatherList P9V 4058f0: 0x92f6d4 _ZN17VersionFile_Vnode15commitAl locListEP1 1ex 405940: 0x8aafa5 _ZN9UFS_Vnode18uncachedWriteChun kEyjRP4msg bRi 405b30: 0x8ab627 _ZN9UFS_Vnode13uncachedWriteEyjP 4msgbiRj+0 x1a
Code now properly passes the handle by reference to other functions.
397129
1
After a Data Mover failover/failback, certain client requests (allocating writes and file creates) were blocked until all cylinder group (CG) bitmaps were initialized. Depending on the configuration, background initialization took a few minutes or panicked after 6 minutes with the following DART panic message: >>PANIC in file: ../sched.cxx at line: 1543 : SYSTEM WATCHDOG. The stack trace showed: Panicking Processor: 0 0x730cd070: 0x14ee4b system_watchdog+0x61 0x730cd080: 0x140efb t_nmi+0xc3 0x730cd1b0: 0x8d929b _ZN14UFS_FileSystem8allocInoEP 20UFS_Upda teDesc 0x730cd300: 0x9436e3 _ZN9UFS_Vnode12getChildNodeEP2
The block allocate function was modified to only wait until any one cylinder group is found with free blocks prior to continuing the client requests.
399405
254
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
0UFS_Updat eDescr 0x730cd3a0: 0x8e615b _ZN9UFS_Vnode9setLengthEyiiiP1 3Buffer_Ca ched+0 0x730cd410: 0x150ceb6 _ZN10DHSM_Vnode9setLengthEyiii P13Buffer_ Cache 0x730cd460: 0x7ee121 2 In the xlt.cfg file, when the subnet address was defined in the format “x:y:z:w, a:b:c:d”, it failed to lookup the host address and ignored the subnet entry. When FSCK was run on a file system that had bad sectors, a memory leak occurred which caused the Data Mover to reboot. Code was fixed by adding delimiter support for "," along with the existing "/". The allocated bad inode objects and bad sector objects are now deleted. Full clone operations are now suspended during auto extend. If file system Auto Extension is enabled and the HWM is exceeded, code pauses the block copy loop until the extension completes (usually around 30 seconds). In the event we really do run out of pool space and can't complete the extend we only allow 5 delay tries during the full clone operation. The file system reference is no longer stored until the resume event completes. If there is a request to unmount the file system, it can now be processed immediately. 375781
2
114569
1
User got Out-of-Space errors on virtually provisioned file systems that started off at a small size (1 GB) and had Auto Extension enabled during a full clone operation.
383532
1
A resume event was not processed when all asynchronous input-output threads were busy deleting large sparse files. The file system became inaccessible during mount cycles and the user saw the following error: No such file or directory>>File system inaccessible.
378345
EMC Celerra Network Server Version 6.0.70.4 Release Notes
255
Version 6.0.41.3
1
When creating new tree quotas and copying large amounts of data, if there were not enough sequential pages for quotaTreeVnodes array, the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 486 : Out of memory. The stack trace showed: 562ee0: 0x145966 _ZN19Memory_AddressSpace10allocP agesEl12Me mor 562f20: 0x24a3de malloc+0x11a 562f60: 0x24aeab _Znwm+0x1b 562f80: 0x24aed6 _Znam+0x4 562f90: 0x8caddb _ZN14UFS_FileSystem20treeQuota On_Interna lEPcS 563210: 0x8cf8b7 _ZN14UFS_FileSystem16quotaOn_Int ernalEPcS0 _+0 563300: 0x833066 _ZN14VFS_FileSystem7quotaOnE9quo taTypetPcS 1_+ 563350: 0x8293d0 _Z17File_QuotaControlR6OptionP18 VFS_FileSy stem
Code was optimized so that sequential pages are no longer required for this operation.
399030 / 38339946
1
During a Data Mover failover in NAS 5.6, the internal mount command had become case sensitive. File systems with accesspolicy=mixed (lower case upgraded from older NAS code) failed to mount and printed the following message in the cmd_log.err 2009-10-31 15:26:38.777 db:0:31381:E: /nas/sbin/slot_recover 2 -panic 2009-10-31 15:30:54.196 db:0:31381:E: /nas/sbin/slot_recover 2 -panic: replace_storage: replace_mount: eng0:mount: unknown accesspolicy: mixed. home0:mount: unknown accesspolicy: mixed. corp0:mount: unknown accesspolicy: mixed. derived0:mount: unknown accesspolicy: mixed. This failure caused the system to abort the
Reinstated case insensitive behavior for the server mount accesspolicy options during failover.
158137 / 32026672
256
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
1
failover and restore all services on the original Data Mover. The Data Mover panicked because of the UFS Log recovery mechanism attempted to invoke FSCK resulting in NULL pointer dereference. DART panic/fault message: *** Invalid Opcode exception. Virt ADDRESS: c ***. The stack trace showed: 0x2c297e8: 0x141254 waitForReboot+0x90 0x2c29808: 0x1414dd fault_dump+0x67 0x2c29828: 0x13a3a0 handle_fatal_fault+0xa 0x2c29880: 0x8504b1 _ZN7Fsck_SB4readEP11NewFsckArgp+ 0x51 0x507b9e0: 0x84f8f7 _Z8check_sbP11NewFsckArgpP7Fsck_ SB+0x23 0x507ba70: 0x85c139 _Z9doNewFsckP9IO_ObjectiiiiRPci+ 0xed 0x507bac0: 0x85d8b8 _Z20kickOffFsckInRecoverP15fsRe coverStru ct+0x84 0x507baf0: 0x81a850 _Z20handleRecoverCommandR6Option +0x60a
The code now checks for a valid file system object for FSCK kickoff during ufs log recovery.
376227 / 35807576
2
User was unable to delete rep v2 iSCSI replication session when its production file system was full. The task hung and the following error was displayed in the server_log: VCS: 3: Failed to write attribute file, fsid=31, path=fs31_T1_LUN2_CK200082600835_0 000 name=.opaque status=NoSpace The Data Mover incorrectly returned NFS3ERR_NOENT (No such file or directory) when the checkpoint (.ckpt) directory was being looked up.
Code was modified to ensure the session delete completes under this condition.
359609 / 34143216
1
The fix eliminates the race by no longer caching checkpoint mountpoint names.
162103 / 32681254
EMC Celerra Network Server Version 6.0.70.4 Release Notes
257
Version 6.0.41.3
1
A Data Mover panicked when moving a directory to new parent which was being deleted. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 6d6db8 Err code: 0 Target addr: 44 **. The stack trace showed: 0x2c2a524: 0x141254 waitForReboot+0x90 0x2c2a544: 0x1414dd fault_dump+0x67 0x2c2a564: 0x13a3a0 handle_fatal_fault+0xa 0x2c2ab30: 0x81cfe9 _ZN19ShadowDirectoryBase6renam eEPtmPS_S0 _P17Acc 0xdd52bcf0: 0xc515d0 _ZN8AppLibNT6renameEPtS0_lR11F ile_Status R10ext 0xdd52bed0: 0xc427c2 _ZN14smb1_threadCtx11replyRena meEP4msgb+ 0x11e 0xdd52bf40: 0xbe1033 _ZN14smb1_threadCtx7processEP4ms gbi+0xcdf 0xdd52bfa0: 0xbd5820 _ZN13smb_threadCtx5startEv+0x6d0
If a parent directory to which the directory is being moved has been deleted, code now returns "no such file" and no longer panics.
371168 / 35185054
1
User reported poor write response time with VMware ESX using Celerra through NFS with uncached file systems.
In this case, the Data Mover serialized the writes and sent the replies after the last write storage. Code was changed to no longer serialize incoming NFS write requests. The cylinder group bitmap which is used to quickly find free blocks is now initialized at bootup.
404209
1
When writes were performed on a large (>2TB) file system which was nearly full (>80%), threads become blocked. This was observed by slow user access and messages like the following in the server log: ThreadsServicesSupervisor: Service:NFSD Pool:NFSD_Exec BLOCKED for 419 seconds. A panic is also possible with the message: PANIC in file: ../ufslog.cxx at line: 6895 : sync: no
386406 / 37135080
258
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
progress reducing dirty list. 1 Using IP addresses with an invalid format (separated by a dot instead of a comma, for example, 102.30.121.131.102.30.109.8) as part of the export command caused memory corruption. The Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 7cd814 Err code: 0 Target addr: 1d3 **. The stack trace showed: 0x2c29dd4: 0x141254 waitForReboot+0x90 0x2c29df4: 0x1414dd fault_dump+0x67 0x2c29e14: 0x13a3a0 handle_fatal_fault+0xa 0x2c29e70: 0x7cd63f _ZN14UFS_FileSystem17startMetada taSyncEv+0 xf 0xb278ff40: 0x7cd4dc _ZN14UFS_FileSystem4syncEi+0x260 0xb278ff90: 0x7cc395 _ZN21UFS_SyncThreadInitial5start Ev+0x2c7 0xb278ffe0: 0x145bd6 Sthread_startThread_internal+0xe 1 A user file system reported corruption of the gidmap file which prevented it from being mounted on the Data Mover. The problem occurred if specific bytes of a gidmap file were zero ( 2th, 3th and 512th bytes), resulting in the checksum being calculated incorrectly and the check during mount failing. The data type for the calculation was corrected and the checksum is now calculated correctly. 377385 / 35859262 Code now limits the buffer overflow that can potentially happen with an incorrect IP address format. 370824 / 35147654
FileMover
The following entries describe fixed issues related to the Celerra FileMover:
EMC Celerra Network Server Version 6.0.70.4 Release Notes
259
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 385097 / 36947738
1
While using FileMover, high memory utilization used in EDRS handles caused CIFS threads to hang and rendered the data unavailable.
The high memory utilization was because the memory used in EDRS read handles was not being freed quickly enough as they were getting blocked by another EDRS process. This issue was resolved by putting the close/free operations into their own thread pool. Code has been fixed to check all existing tree IDs resulting in proper path display after a directory rename from the client.
1
When having non-sequential tree IDs on the file system, renaming one of the directories representing one of the quota tree path (having an ID after the gap in the list) from the client did not reflect the change in the quota tree list using the nas_quotas command. The issue occurred because code was checking the sequential IDs only to the first gap. DHSM connections (using Kerberos for authentication) tried to re-use Kerberos tickets in the credential cache. The code used to check to see whether a ticket was available in the cache did not check to see if the ticket was expired, so the specified password was deleted. Subsequent code tried to use the expired ticket, then was unable to get a new ticket because the password had already been deleted, resulting in DHSM reconnection failure. For XML API clients, a query was made to get Shares with a filter of name="value", but the value field was specified as caseinsensitive. Code enforced that values had to be case -sensitive, which caused older queries by XML API clients to fail if they did not specify the value correctly.
386777 / 37133416
1
Added code (in Kerberos) to check for expired ticket in credential cache, and to delete the ticket if it is expired.
375592 / 35601816
2
The C3 code was modified to accept case in-sensitive values for share names in order to remain backward compatible.
384012
260
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 401210 / 37512728
2
User could not setup a NetApp server as a secondary server for DHSM. Netapp always returned unspecific error .
This has been fixed, and the CIFS client can now connect to NetApp server.
Installation/Configuration/ Upgrades
The following entries describe fixed issues related to installations, configurations and upgrades:
Severity Symptom Description Fix Summary Tracking Number 385933 / 36553396
2
The sys_log showed the following repeatedly: Enclosure 0 failed to respond to "ping" at management switch A. These alerts were due to slow response from system Management Switches to Control Station ping requests. If the partition table was corrupted, the PUHC did not capture the error and the upgrade failed at step "Create partitions for upgrade".
Code was fixed to prevent false reboot messages and to harden some boxmonitor communication. A new PUHC check was added to ensure that the disks are consistent so that the issue could be captured prior to the start of the upgrade. Added new check for secondary CS to verify whether root file system have enough space prior to upgrade.
1
391103 / 37606144
1
Control Station upgrade failed due to insufficient space on the "root" file system. In PUHC, there was no check available for the secondary Control Station (CS1) to verify the "available space" on "root" file system before upgrading.
375108
EMC Celerra Network Server Version 6.0.70.4 Release Notes
261
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 381208
2
Post Upgrade health check failed with error: Error HC_SYS_14505213994: "nas_storage -check" command timed out.
Code has been modified to avoid this timeout condition. The checks "cs::check_if_Symapi_ data_is_pres ent" and "cs::check_if_Symapi_i s_synced_with_Storag e_System" are shifted to the end of the "cs" checks in checks.txt file to gain some time. This error is no longer displayed.
1
An install of version 6.0.36.4 in an NSG2 system failed with the following: cmd: /nasmcd/sbin/cs_dhcpd_monitor single_execution output: Unable to start cs_dhcpd_monitor -single_execution, already running! If user's PATH variable was missing and they tried to run clariion_mgmt, the iptables file was not updated. This caused the following error during PAHC/PUHC: Storage System : Check SPA SPB communication Error HC_BE_14505017473: At least one iptables rule is missing. Action : Run "spa_spb_comm -setup" to fix the problem. While performing an out-of-family upgrade from 6.0.x to 7.0.x, a VG8 system was incorrectly reported as VG2. During an in-family upgrade of a dual Control Station system, on the secondary CS, the task 'create IDE cache' was not run , therefore the files were copied directly to the common cache and wbem directories, thus filling up the Control Station root. In an out-of-family upgrade from 5.6 to 6.0, if the NAS service failed to stop
381753
2
Adding full path to commands in clariion_mgmt using iptables-save to prevent this condition.
384764
1
The system now reports the correct model. The task 'create IDE cache' is now run during upgrades to prevent this condition.
402851
2
380619
1
A fix has been provided to ensure that
389477
262
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
properly and some NAS file systems were left mounted at the end of the task. Upon Control Station reboot, the upgrade could not complete since the nas file systems were mounted to restore pre-reboot system state. Error - Partition table on /dev/hda is inconsistent. 2 After installing new language packs on top of an existing language pack in Unisphere, error messages were returned in English. When the file /nas/sys/nas_install_version did not exist, the upgrade failed with the following error: /nas/sys/nas_install_version: No such file or directory. There was no reference to which version of NAS code the system was originally installed with.
no NAS related file systems are mounted by the upgrade for state restoration once the NAS service has been stopped during the upgrade. With this fix, error messages are returned in the installed language. Added /nas/sys/nas_install_ve rsion into nasdb_backup code. 395453
2
387041
iSCSI
The following entries describe fixed iSCSI-related issues:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
263
Version 6.0.41.3
1
In a cluster environment, the iSCSI server hung and stopped responding. Host A had heavy I/O traffic to a LUN, and Host B tried to reserve the same LUN, using the SCSI-3 Persistent Reservation command. The reserve command hung, causing a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../dskdump.cxx at line: 1836 : ESX iSCSI access issue. The stack trace showed: 0x1f055ae0: 0x140ce4 waitForReboot+0x90 0x1f055b00: 0x140f69 fault_dump+0x67 0x1f055b20: 0x140e51 PANIC+0x29 0x1f055b30: 0xa2da32 _Z11paniccfgcmdR6Option+0x17e 0x1f055c70: 0x1c9f67 _ZN11cfgcmd_desc6invokeEPKcPP8CC MD_Msgii +0x1a7 0x1f055ce0: 0x2ae4a2 _Z10xml_cfgcmdP7macDataRP8CCMD_Msg +0x392 0x1f055d30: 0x2a9418 _ZN17macCCMDCommand_DB6invokeEPK cP7macDa taRP8C 0x1f055d60: 0x2a991e _Z17mac_CCMDDocParserP7macDataRP 8CCMD_Ms g+0x20
The problem was caused by coding bug which created a small unprotected windows where an I/O completion callback was skipped. This was corrected and the unprotected window has been eliminated.
377382
NDMP
The following entries describe fixed issues related to Network Data Management Protocol and/or Backup:
Severity Symptom Description Fix Summary Tracking Number 387236 / 36692786
2
McAfee antivirus applications reset the last accessed time (atime) of a file after a clean scan. The Data Mover code updated the ctime when the atime was reset which impacted other features such as incremental NDMP backups. A race condition could occur during NDMP backup, when abort is called ,
The ctime is no longer updated as a result of an antivirus client resetting the atime.
1
Code has been fixed by using mutex lock to
395387 / 38052144
264
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
which results in a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 1d32a4 Err code: 0 Target addr: 0 *. DART stack trace: 0x2c4ba44: 0x14124c waitForReboot+0x90 0x2c4ba64: 0x1414d5 fault_dump+0x67 0x2c4ba84: 0x13a398 handle_fatal_fault+0xa 0x2c4baf0: 0x11fa5cd _ZN9NdmpdData7abortOpEv+0x253 0xf80e7ae0: 0x11fb42c _ZN16NdmpdDataAbortMH14messag eHandlerEPv +0x94 0xf80e7b10: 0x1223aec _ZN14NdmpConnection15processM essagesEi+0 x2c6 0xf80e7b40: 0x122275c _ZN14NdmpConnection15processRe questsEv+0 x14 0xf80e7b60: 0x11f29c3 _ZN21ConnectionFileHandler11h andleEventE v+0x1
protect the Data Mover from panicking under this condition
EMC Celerra Network Server Version 6.0.70.4 Release Notes
265
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 380146 / 36341620
1
At the end of NDMP backup/restore, two threads accessed the same object during cleanup. One thread was reset while the other was using it. This led to a Data Mover panic with the following Dart/panic fault message: *** Page Fault Interrupt. Virt ADDRESS: 9eed24 Err code: 2 Target addr: ce **. The stack trace showed: 0x2c3ab24: 0x14124c waitForReboot+0x90 0x2c3ab44: 0x1414d5 fault_dump+0x67 0x2c3ab64: 0x13a398 handle_fatal_fault+0xa 0x2c3ac50: 0x9ee354 _ZN13PaxDataModuleD0Ev+0x4c 0x37a91c30: 0x11f3192 _ZN22NdmpdDataModuleManager10de allocateEP1 5Nd 0x37a91c40: 0x11f145e _ZN9NdmpdData16deallocateModule EP15NdmpdDa taM 0x37a91c50: 0x11ede7e _ZN9NdmpdDataD1Ev+0x34 0x37a91c70: 0x12020e9 _ZN12NdmpdSessionD1Ev+0x63
Code now protects the cleanup during NDMP backup/restore to prevent this condition.
2
Running security scanner against NDMP connections using port 10000 and having connection closed not cleanly, left that connection open on Data Mover. Eventually, when 2048 such connection will be left open, that will result in the following server error: NDMP: 3: Active NDMP session number reaches the limit 2048, No more NDMP sessions can be started. Contact your Customer Support representative and any NDMP backup will fail from this point. A directory read by an NDMP backup triggered blocks to be flushed to the file system, which in turn caused the changes to be snapped to the SavVol. The oldest
Code has been fix to properly close the NDMP connection when network connection closed abnormally eliminating the problem.
380377 / 36311850
1
Code now allows the NDMP readdir operation to be aborted
375343 / 35697814
266
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
checkpoint had to be inactivated because the SavVol was full. However, the lock held by the NDMP backup on the file system prevented the unmount of the oldest checkpoint. This deadlock persisted until the panic was forced by the pending I/O. DART panic/fault message: >>PANIC: I/O not progressing LastVol touched Sn188 Kind 5 (ptr=0xace4c404). The stack trace showed: 0x6db1e90: 0x13fab4 waitForReboot+0x90 0x6db1eb0: 0x13fcb1 fault_dump+0x67 0x6db1ed0: 0x13fbb1 PANIC+0x29 0x6db1ee0: 0x822a30 _ZN9RawIO_IRP13checkProgressEv+ 0x11a 0x6db1fd0: 0x82290c _ZN17IrpProgressThread5startEv+ 0x6 0x6db1fe0: 0x14424c Sthread_startThread_internal+0x e
and retried when the checkpoint operation has completed.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
267
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 376319
1
During NDMP backup, in some rare conditions, a tape error led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../include/stream.hxx at line: 979 : datab::freeRef: db_ref <= 0 . The stack trace showed: 0xbccc5e50: 0x13f288 waitForReboot+0x90 0xbccc5e70: 0x13f485 fault_dump+0x67 0xbccc5e90: 0x13f385 PANIC+0x29 0xbccc5ea0: 0x1b407a _ZN5datab7freeRefEv+0x32 0xbccc5eb0: 0x96dc41 _ZN11MsgLinkList5EmptyEv+0x71 0xbccc5ed0: 0x970a76 _ZN9rw_thread5startEv+0xcb0 0xbccc5fe0: 0x143a08 Sthread_startThread_internal+0x e
Code was modified to properly handle media errors and prevent this panic.
1
If the user created an NDMP account with text format, and put letter "A" into the account comment part, when DMA sent NDMP_CONNECT message with MD5 format, the decryption algorithm caused a Data Mover panic because the dividend became 0. DART panic/fault message: *** Divide exception. Virt ADDRESS: 11c036b ***. The stack trace showed: 0x2baf8ec: 0x140cec waitForReboot+0x90 0x2baf90c: 0x140f71 fault_dump+0x67 0x2baf92c: 0x139eec handle_fatal_fault+0xa 0x2baf9a0: 0x11c006e _ZN18NdmpdConnectAuthMH14mess ageHandlerE Pv+0x4 0xef84bb10: 0x11ec6f2 _ZN14NdmpConnection15processM
This is a wrong configuration. Code now handles this exception correctly and returns INVALID_ARGUMENT error.
390481 / 37563732
268
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
essagesEi+0 x2ba 0xef84bb40: 0x11eb388 _ZN14NdmpConnection15processR equestsEv+0 x14 0xef84bb60: 0x11bbee3 _ZN21ConnectionFileHandler11h andleEventE v+0x1 0xef84bb80: 0x11ed956 _ZN19NdmpEventDispatcher8disp atchEim+0x7 ce
Network
The following entries describe fixed network-related issues:
Severity Symptom Description Fix Summary Tracking Number 385326 / 36986690
1
A PCI bus failure caused the Neterion 10G NIC device to stop working. Undetected hardware failure of the Neterion 10G device resulted in an unusable port.
Code was modified to periodically check the hardware state of the Neterion device, and issue restart command if the device stops responding to I/O requests.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
269
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 382462 / 36615128
1
When using virtual ethernet comprised of multiple devices (in this case configured an FSN over two 4-port trunks), memory allocation triggered a watchdog panic during power up. DART panic/fault message: >>PANIC in file: ../sched.cxx at line: 840 : SYSTEM WATCHDOG. The stack trace showed: 0x20141ae4: 0x13fb94 waitForReboot+0x90 0x20141b04: 0x13fda9 fault_dump+0x67 0x20141b24: 0x13fc91 PANIC+0x29 0x20141b34: 0x141b41 system_watchdog+0x77 0x20141b44: 0x138966 _t_nmi+0x68 0x20141ba0: 0x30a072 _Z7LM_HaltP16_LM_DEVICE_BLOCK+0 x1ea 0x20141bd0: 0x307b20 _Z15LM_ResetAdapterP16_LM_DE VICE_BLOCK+0 x48 0x20141c00: 0x306392 _Z20LM_InitializeAdapterP16_L M_DEVICE_BL OCK+0x
Code now separates the start into individual threads, causing the RT lock to be released between each device's start processing, thus avoiding the watchdog panic.
2
User reported slow response time when mapping a Data Mover CIFS share or when mounting an NFS export. This issue only occurred when the Data Mover DNS client was explicitly configured with the TCP protocol (server_dns configures UDP by default). The DNS client used for discovering Domain Controllers and query for hostname addresses did not properly interpret two DNS responses included in the same TCP packet. In some instances, the device "opaque" fields (ones that the device is supposed to copy unmodified) experienced single bit corruptions. DART panic/fault message: *** Page Fault Interrupt. Virt
The DNS client code was modified to properly handle two DNS responses received in a same TCP packet.
399644 / 38312216
1
The fix is to replicate the information so that it can be quickly verified when received.
377651 / 36005286
270
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
ADDRESS: 322e53 Err code: 0 Target addr: c **. The stack trace showed: 0x2ba3de4: 0x140ce4 waitForReboot+0x90 0x2ba3e04: 0x140f69 fault_dump+0x67 0x2ba3e24: 0x139ee4 handle_fatal_fault+0xa 0x2ba3e90: 0x31c7a8 _Z20LM_ServiceInterruptsP16_L M_DEVICE_BL OCK+0x1 0x4e81f80: 0x30fecf _ZN12bcmnicDevice5bcmRTEiPv+0x4 5 0x4e81fb0: 0x14226f _ZN15SchedRT_Initial5startEv+0x f3 0x4e81fe0: 0x1455cc Sthread_startThread_internal+0x e 1 Under heavy load, the Neterion 10G NIC hung, which led to loss of CIFS/NFS access, and logged many errors similar to the following in the server log: 2010-07-07 21:43:00: DRIVERS: 6: fxg-1-0 : Rescheduling RT. Enhanced recovery mechanism to fix the issue on Neterion 10G with faulted hardware. 375769 / 35590150
EMC Celerra Network Server Version 6.0.70.4 Release Notes
271
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 386273
1
In a rare case, during the disconnection of an SSL session, a certain sequence of network packets in the SSL module stream caused a Data Mover panic with the following: DART panic/fault message: Assertion failure: '_RDp != __null' on line 695 in file '../../dart_ssl/ssl_str.cxx'>>P ANIC in file: ../assert.c at line: 26 : assert failed. The stack trace showed: 0x77ddea0: 0x14a464 waitForReboot+0x90 0x77ddec0: 0x14a703 fault_dump+0x67 0x77ddee0: 0x14a5d5 PANIC+0x29 0x77ddef0: 0x23de29 _assert+0x51 0x77ddf10: 0x11c9a4b _ZN10ssl_strCtx7wputSslEv+0x37 0x77ddf60: 0x11c9e99 _Z11sslstr_wsrvP5queue+0x4d 0x77ddf80: 0x20b593 _Z10STRserviceiPv+0x1af 0x77ddfa0: 0x14bba6 _ZN15SchedRT_Initial5startEv+0x 132
SSL module stream function now handles the sequence of network packets correctly.
NFS
The following entries describe fixed issues related to Network File Systems:
Severity Symptom Description Fix Summary Tracking Number 388304
2
Write performance was reduced when file system free space was limited.
Enhanced the search for free space by selecting cylinder groups by their availability instead of a predefined order.
272
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 385026
2
When a file system was mounted with option “ntcredential”, the UNIX command chgrp failed on objects with owners that were members of more than 16 groups. The Data Mover took into account only groups that were part of NFS credentials, ignoring other groups specified in NT credentials.
The Data Mover now takes into account both NFS credentials and NT credentials when file system is mounted with option “ntcredential” and parameter cifs.acl.extendExtraGid is set. The limit of total number of groups is defined by parameter security.maxgroups. Modified the way files are allocated to optimize performance. Made the removal of these locks faster, so that the system watchdog is no longer triggered under this condition.
1
File system performance dropped when the files in the file system were not allocated correctly. When an NFSv3 client notifies the Data Mover that it has just rebooted, the Data Mover must revoke all the locks owned by the previous instance of this client. The removal of these locks triggered a system watchdog panic when the count of locks was huge (30K). DART panic/fault message: >>PANIC in file: ../sched.cxx at line: 846 : SYSTEM WATCHDOG. The stack trace showed: 0x5085f24: 0x141254 waitForReboot+0x90 0x5085f44: 0x1414dd fault_dump+0x67 0x5085f64: 0x1413c5 PANIC+0x29 0x5085f74: 0x1432d1 system_watchdog+0x77 0x5085f84: 0x139f72 _t_nmi+0x68 0x5085fe0: 0x145bd6 Sthread_startThread_internal+0x e
397730
1
401962 / 38796804
EMC Celerra Network Server Version 6.0.70.4 Release Notes
273
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 393021
1
If an NFS export was created with readonly host ip as 99.99.99.99, creating the NFS export failed with the error message: Modify NFS export server_2:/sanity_f123. server_2 : Invalid argument Export Error: Unknown option details. When the same process was done for Access hosts, it created the NFS export successfully. User could not set permissions via NFSv4 clients on files and folders with ACL containing ACE of Windows local SYSTEM account. The following error appeared in server_log: NFS: 3: nfs4_checkSpecialAce: SID S-1-5-12 cannot be translated to NFSV4 : ACL can't be modified.
Code was modified to ensure that exporting a path as read-only to specified hosts work correctly.
2
A new parameter nfsv4.ntAllSpecialUser s was added. When this parameter is set to 0 (default), Windows special users that are not listed in the RFC 3530 are returned as 'nobody' and changing permissions on objects with ACL containing local SYSTEM account will fail. When this parameter is set to 1, Windows special users are translated to a unique identifier, this allows modification of ACL for such objects from NFSv4 clients. Code has been modified and a better mutex protection has been added to export database initialization.
387582
1
During NFSv4 export database initialization, the server panicked with the following DART panic message: *** Page Fault Interrupt. Virt ADDRESS: 0x000179fb2e Err code: 0 Target addr: 0x0000000078 **
396771
1
An NFSv4 client was sending a request to confirm the client ID to the server. In some cases, the Data Mover had already confirmed the ID, and didn’t know how to handle this situation. The Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 631ae3
Code has been fixed to handle this request by returning the following error message to the client: ID already in use and the Data Mover no longer panics in this situation.
384098 / 36843082
274
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
Err code: 0 Target addr: 28 **. The stack trace showed: 0x2acfb34: 0x13fb94 waitForReboot+0x90 0x2acfb54: 0x13fda9 fault_dump+0x67 0x2acfb74: 0x138d94 handle_fatal_fault+0xa 0x2acfbc0: 0x62f76c _ZN16NFSv4_ClientList7Confirm EyPA8_KhR8n fs4_ctx 0x433fdcb0: 0x62822b _Z24nfs4_setclientid_confirmR P4msgbS1_RP hR8nfs 0x433fdd10: 0x622de0 _Z23nfs4_convertNfsv3Statusm+0x 656 0x433fde90: 0x657949 _ZN13NfsGssRequest8dispatchEP 4msgbPvP17A ccess_ 0x433fdee0: 0x605e16 _ZN17nfs_serverInitial5startEv+ 0x5d0
EMC Celerra Network Server Version 6.0.70.4 Release Notes
275
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 385131 / 36960608
1
During nfsv3_lookup, if the path name length field in nfs header was corrupted, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 5e5e39 Err code: 0 Target addr: ffee8693 **.The stack trace showed: 0x28a6c64: 0x13fa98 waitForReboot+0x90 0x28a6c84: 0x13fc95 fault_dump+0x67 0x28a6ca4: 0x138cfc handle_fatal_fault+0xa 0x28a6e20: 0x5e2436 _ZN17nfs_serverInitial5startEv+ 0xac2 0xdefd1fe0: 0x144230 Sthread_startThread_internal+0x e
To prevent this panic, string length verification was added to verify the length field of NFS header of the request received from the network.
Replication
The following entries describe fixed problems encountered while using Celerra Replicator, SnapSure or Checkpoints:
Severity Symptom Description Fix Summary Tracking Number 387170 / 37192564
1
A RepV2 session could not be stopped when the transfer was waiting for the destination SavVol to be extended.
Added a check for the stop request to continue during SavVol extension. This will allow a stop to proceed when the session is waiting for the destination SavVol to be extended. Code was fixed to prevent the checkpoint on the destination file system from becoming
1
Repv2 suspends the writes to the destination file system when the SavVol is being extended. However, if the current SavVol usage value was equal to the
388664 / 37379978
276
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
HWM (high watermark), writes were not suspended as expected. This caused the destination checkpoints to become inactive and if the file system on the destination was being accessed, an inconsistent view of the file system led to a Data Mover panic with the following DART panic/fault message: mangled directory ent. The stack trace showed: 0xcfdd56c0: 0x14124c waitForReboot+0x90 0xcfdd56e0: 0x1414d5 fault_dump+0x67 0xcfdd5700: 0x1413bd PANIC+0x29 0xcfdd5710: 0x778757 _ZN14UFS_FileSystem28markCorr uptedFsForU nmount 0xcfdd58d0: 0x7e272c _ZN9UFS_Vnode6dirbadEmPKc+0xea 0xcfdd5990: 0x7c0ee7 _ZN20UFS_DirHashInfo_dir3C1EP 11Local_Vno de+0x5 0xcfdd5a10: 0x7e5b04 _ZN15UFS_DirOps_dir311initDir HashEP9UFS_ Vnode+ 0xcfdd5a40: 0x7d6ef6 _ZN15UFS_DirOps_dir37scandirE P9UFS_Vnode PK18Fi 1 When using NFSv4, users were unable to copy directories on Data Mover shares from AIX clients using relative pathnames (for example, using “cp –r” command). The following error was reported by the client: No such file or directory. The issue occurred due to incorrect MOUNT_FILEID value returned to AIX client in NFSv4 request.
inactive under this condition.
Data Mover now returns the correct MOUNT_FILEID value to NFSv4 clients allowing successful directory copies.
389170 / 37129124
EMC Celerra Network Server Version 6.0.70.4 Release Notes
277
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 394547 / 37882060
1
Replication recovery tasks were not marked as complete if the destination Data Mover could not communicate with the source Data Mover during a VDM replication failover. A subsequent reboot of the Data Mover failed to mount the root file system of the VDM due to the presence of the incomplete tasks.
The code now checks whether a remount of the file system is needed during taskManager recovery, remounts the required file systems only, and correctly updates the status of the tasks. Code has been modified to handle the un-aligned 8KB Read/Write I/Os to writable checkpoints correctly.
1
When using writable checkpoints and when un-cached I/O was used (in cases such as with VMWare/VAAI), Data Mover panicked with the following DART panic/fault message: >>PANIC: I/O not progressing LastVol touched Sh148108 Kind 8 (ptr=0x57d42204). The stack trace showed: 0x7c3de90: 0x14124c waitForReboot+0x90 0x7c3deb0: 0x1414d5 fault_dump+0x67 0x7c3ded0: 0x1413bd PANIC+0x29 0x7c3dee0: 0x8ad1b6 _ZN9RawIO_IRP13checkProgressEv+ 0x11a 0x7c3dfd0: 0x8ad092 _ZN17IrpProgressThread5startEv+ 0x6 0x7c3dfe0: 0x145bfa Sthread_startThread_internal+0x e
386603
1
When the virtually provisioning options were modified on a file system replicated with Repv2, the replication database was not updated with the new settings. Therefore, after the Data Mover was rebooted, the old settings were applied to the file system. This resulted in the virtually provisioned file system not being extended.
The nas_fs - modify command now modifies the mount options correctly.
384264 / 36813726
278
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 115393
2
The nas_replicate -stop and -delete operations required the user to enter a "mode" (set to both, source, or destination).
The nas_replicate command was enhanced so that the stop and -delete options default to 'both'. Code has been fixed to use the correct schedule for mounting the checkpoint.
2
When creating a checkpoint schedule, if a user changed the relative checkpoint from time based to named and then tried to mount the checkpoint, the checkpoint was mounted with the time based schedule instead on the named schedule.
378004
EMC Celerra Network Server Version 6.0.70.4 Release Notes
279
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 391982 / 37718380
1
During the merge of an empty checkpoint into an older checkpoint, after all the needed mappings were moved and the stableEntry was 2048*N. After a reboot, the wrong table of contents (TOC) buffer was flushed and the TOC became corrupt. This led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../DeltaVol.cxx at line: 7550 : Value in TOC is different from the value in btree. The stack trace showed: 0x9cb85df0: 0x14124c waitForReboot+0x90 0x9cb85e10: 0x1414d5 fault_dump+0x67 0x9cb85e30: 0x1413bd PANIC+0x29 0x9cb85e40: 0xfb1e7c _ZN19PoolManagedDeltaVol14verifyTO CValue Emy+0x 0x9cb85ea0: 0xfb1805 _ZN19PoolManagedDeltaVol10ver ifyTOCsEym1 1Verif 0x9cb85f00: 0xfb14fe _ZN19PoolManagedDeltaVol5merg eERP5Event+ 0xcf2 0x9cb85f80: 0xfc6d3c _ZN11mergeThread5startEv+0x1b4 0x9cb85fe0: 0x145bfa Sthread_startThread_internal+0x e
Code has been fixed to avoid the panic and to make the TOC pointer point to the right TOC buffer to be flushed.
3
The nas_replicate-info command always shows the "Next Transfer Size" field to have the same value as "Current Transfer Size" field.
The Next Transfer Size (KB) field in the output of command nas_replicate -l was removed. Code now skips the empty signatures check during one-tomany replication session to avoid this
370770 / 34923460
1
During 1-to-Many replication sessions, empty signatures check prevented the source VDM from being changed to mounted state.
386277
280
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
issue. 1 A deadlock condition occurred when RDE layer queries for SavVol stats while a checkpoint restore is in progress. The deadlock occurs during the dereference process in the DpLayer. Code no longer requires the RDE thread to need a version set lock during the dereference process, hence avoiding a deadlock. When unknown clients attempt to connect to port 8888, the sessions are now properly closed and cleaned up. 395215
1
If an unknown client (e.g. a port scanner) connected to RCP 8888 port too fast, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0 Err code: 0 Target addr: 0 ** . The stack trace showed: 0x2ba3d10: 0x140ce4 waitForReboot+0x90 0x2ba3d30: 0x140f69 fault_dump+0x67 0x2ba3d50: 0x139ee4 handle_fatal_fault+0xa 0x2ba3db0: 0x1d10d4 CloseStream+0x1b6 0x1ff3deb0: 0x12ce0b8 _ZN18RCP_SessionManager6accep tEPvP11sock addr_ 0x1ff3df40: 0x12ca690 _ZN10RCP_Server5startEv+0x3e6 0x1ff3dfe0: 0x1455cc Sthread_startThread_internal+0x e
384423 / 36899914
1
Replication transfer did not make progress as the SavVol on the destination could not be extended. User did not have enough information to understand why replication transfer did not progress.
An event is now posted to inform the user of the SavVol extension progress.
380166 / 36340862
EMC Celerra Network Server Version 6.0.70.4 Release Notes
281
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 373743
1
A checkpoint was stuck at Frozen state when the following three conditions occurred simultaneously: a PFS was being extended, checkpoint was being refreshed and SavVol was full. This led to blocking of snap IOs and triggered the following: DART panic/fault message: >>PANIC: I/O not progressing LastVol touched not Known The stack trace showed: 0xdfe5df0: 0xc6c7bd _ZN9RawIO_IRP13checkProgressEv+ 0x3e9 0xdfe5fc0: 0xc6cb3e _ZN17IrpProgressThread5startEv+ 0x4 0xdfe5fd0: 0x16ad51 Sthread_startThread_internal+0x
Code was modified to ensure that when there are pending unmounting events for a checkpoint, the initFS allows mounting. This prevents the failure of checkpoint thaw for potential ongoing checkpoint refresh operation.
2
There was potential for slower than expected network transmit of RepV2 data when manually setting RCP TCP params.
Corrected an arithmetic error, and the decision logic for when to change the network transmit window. The "Max Out of Sync Time" setting is now preserved and will be applied to the session when it is restarted after a failover/switchover. In the HTTPS server of the Data Mover, there was no timeout during the establishment of a SSL connection (between the acceptance of the TCP connection and the first packet sent by the client). The Data Mover was waiting for
381470 / 35658014
1
When a failover/switchover is executed on a replicationV2 session followed by a start in the reverse direction, the session will be set for manual refresh regardless of original "Max Out of Sync Time" setting.
389459
1
While replicating one Celerra to two other Celerras, the Data Mover on one of the destination Celerras did not clean up the connections to port 5085 and eventually the server_log became full of Too many connections on ports 5085 queue. The Data Mover on the destination file system panicked with the following DART panic/fault message: >>PANIC in file: ../dskdump.cxx at line: 1836 : Too many connections on
381594 / 36464168
282
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
port 5085. The stack trace showed: 0x1facbae0: 0x141254 waitForReboot+0x90 0x1facbb00: 0x1414dd fault_dump+0x67 0x1facbb20: 0x1413c5 PANIC+0x29 0x1facbb30: 0xa44c58 _Z11paniccfgcmdR6Option+0x178 0x1facbc70: 0x1cbb8b _ZN11cfgcmd_desc6invokeEPKcPP 8CCMD_Msgii +0x1a7 0x1facbce0: 0x2b2b3e _Z10xml_cfgcmdP7macDataRP8CCMD_ Msg+0x392 0x1facbd30: 0x2adaf0 _ZN17macCCMDCommand_DB6invoke EPKcP7macDa taRP8C 0x1facbd60: 0x2adff6 _Z17mac_CCMDDocParserP7macDat aRP8CCMD_Ms g+0x20 1 After rebooting, the Data Mover would hang during the checkpoint recovery process and remain in "Loaded" state. Data Mover failovers would also fail.
this first packet (the SSL clientHello), and until this packet was received, the HTTPS didn't process any new connection request. The HTTPS server was blocked and unavailable. This has been fixed.
A deadlock occurred while trying to correct a rare form of checkpoint corruption. The code now unlocks locked memory correctly, avoiding the deadlock.
378406 / 35911666
EMC Celerra Network Server Version 6.0.70.4 Release Notes
283
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 383291 / 36431708
1
A function in the code was consuming stack memory, which caused a stack overflow and led to a Data Mover panic during replication. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 1fa53e Err code: 0 Target addr: 0 **. The stack trace showed: 0x2acb158: 0x13fb94 waitForReboot+0x90 0x2acb178: 0x13fda9 fault_dump+0x67 0x2acb198: 0x138d94 handle_fatal_fault+0xa 0x2acb1c0: 0x56e267 krb5_cc_resolve+0x17 0x1eca92b0: 0x56e609 krb5_cc_default_suffix+0xa3 0x1eca92e0: 0x53e1db _Z26krb5_gss_release_principalP jPPv+0x46b 0x1eca93a0: 0x53ed90 _Z25krb5_gss_acquire_cred_ext PjPv22gss_b uffer_ 0x1eca9400: 0x5456ff gss_acquire_cred_ext+0x4b
Code was modified to reduce memory consumption on the stack by using heap memory instead, preventing this condition.
1
When a checkpoint was being refreshed or deleted the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../DeltaVol.cxx at line: 7550 : Value in TOC is different from the value in btree. The stack trace showed: 0x2971dd70: 0x141254 waitForReboot+0x90 0x2971dd90: 0x1414dd fault_dump+0x67 0x2971ddb0: 0x1413c5 PANIC+0x29 0x2971ddc0: 0xfaae5e _ZN19PoolManagedDeltaVol14ver ifyTOCValue Emy+0x
The checkpoint merge process now correctly handles the case when the merge has been interrupted by a Data Mover reboot.
381357
284
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
0x2971de20: 0xfaa7e7 _ZN19PoolManagedDeltaVol10ver ifyTOCsEym1 1Verif 0x2971de80: 0xfac73f _ZN19PoolManagedDeltaVol11cha ngeToTOCEv+ 0x8b 0x2971deb0: 0xfac110 _ZN19PoolManagedDeltaVol13fin dFirstFreeE RyRP5E 0x2971df00: 0xfa9c62 _ZN19PoolManagedDeltaVol5mergeE RP5Event+0x 474
EMC Celerra Network Server Version 6.0.70.4 Release Notes
285
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 377960 / 35997088
1
There was a race between SnapSure smart snap and delete command. Smartsnap traversed from the latest ckpt to oldest ckpt to decide whether a block is a smartsnap block, while at the same time, the oldest ckpt was deleted by user command. During smart snap, code needed to find mapping from the oldest ckpt and only found it was destructed, then panicked the Data Mover with the following DART panic/fault message: *** GP exception. Virt ADDRESS: b877dd. Err code: 0 *** . The stack trace showed: 0x172f458: 0x13db94 waitForReboot+0x90 0x172f478: 0x13dd91 fault_dump+0x67 0x172f498: 0x137d54 handle_fatal_fault+0xa 0x172f4e0: 0xb91b91 _ZN15snapshotContext10BlockIn UseEP8Delta VolyP8S 0x242e9be0: 0xb7d6dc _ZN19PoolManagedDeltaVol18wri teBlockFilt eredEP 0x242e9c90: 0xb8d232 _ZN10SnappedVol18writeBlockFi lteredEP10D iskIO_ 0x242e9cb0: 0xb8d1ed _ZN10SnappedVol18writeBlockIn ternalEP10D iskIO_ 0x242e9cd0: 0x731731 _ZN11HyperVolume18writeBlockI nternalEP10 DiskIO
This fix is to close this race by suspending smart snap if there is a delete operation in progress.
1
The Data Mover experienced repeated Watchdog panics when an attempt was made to create the first ReplicationV2 session. This was because the Control Station to DART interconnect did not exist. DART Server183ic/fault message: >>Server183IC in file: ../sched.cxx at line: 846 : SYSTEM WATCHDOG. The stack
During a Data Mover reboot, a check will be performed to verify that the Control Station to DART interconnect is present. If the interconnect is missing it will be added by running "nas_cel -
371904 / 35235954
286
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
trace showed: 0x5085f24: 0x14124c waitForReboot+0x90 0x5085f44: 0x1414d5 fault_dump+0x67 0x5085f64: 0x1413bd Server183IC+0x29 0x5085f74: 0x1432c9 system_watchdog+0x77 0x5085f84: 0x13IP_ADDR_nmi+0x68 0x5085fe0: 0x145bfa Sthread_startThread_internal+0x e
update id=0".
EMC Celerra Network Server Version 6.0.70.4 Release Notes
287
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 370442
1
A SavVol was almost full, and the most recent ckpt was about to be inactivated to save space. A snap request on this ckpt happened just before the deactivation, which caused an invalid function in the code and led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../DeltaVol.cxx at line: 3792 : doSavVolIO failure. The stack trace showed: 0x236025e60: 0x1309df4 _ZN19PoolManagedDeltaVol10doS avVolIOEyjP ci16 0x236025eb0: 0x130ed59 _ZN19PoolManagedDeltaVol22Flus hSnapShotL ogLo 0x236025ef0: 0x130f0c7 ZN19PoolManagedDeltaVol16Flus hSnapShotLo gE1 0x236025f20: 0x12fcdf5 _ZN12IntentLog_CB8allocateEv+0x 15 0x236025f40: 0x1348c65 _ZN21ckptBlockAllocatorThr5st artEv+0x375 0x236025fd0: 0x15519f Sthread_startThread_internal+0x 7
This issue was fixed by adding a hdr check condition to FlushSnapShotLogLoc ked, thus avoiding a panic.
1
There was a race condition between ckpt restore complete routine and ckpt delete command. SavVol was full during ckpt restore process, it erased all the ckpts on the SavVol, while a ckpt on this SavVol was being deleted by user's command, it double-destructed the ckpt which led to a panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: f8d2d0 Err code: 0 Target addr: 0 **. The stack trace showed: 0x2ba3bd4: 0x140ce4 waitForReboot+0x90
A lock mechanism was added to avoid the race condition.
159333
288
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
0x2ba3bf4: 0x140f69 fault_dump+0x67 0x2ba3c14: 0x139ee4 handle_fatal_fault+0xa 0x2ba3c70: 0xf8cf26 _ZN19PoolManagedDeltaVol18Rel easeChunkBi tMapEv+ 0x1f453d60: 0xf7f332 _ZN19PoolManagedDeltaVol15Rel easeSnapsho tEv+0x 0x1f453dc0: 0xf7aa74 _ZN19PoolManagedDeltaVol19SetSnap ShotInA ctiveE 0x1f453e40: 0xf94af2 _ZN15snapshotContext17BroadCa stInactiveE PKcP8D 0x1f453e80: 0xf802a8 _ZN19PoolManagedDeltaVol11Res toreStopEv+ 0xa4 1 A replication stop/delete/failover operation task hung on the source side of replication, if the destination side was already stopped or deleted. The task indicated the following error: Error 13160415328: Registration request rejected. Code has been modified to ensure replication commands such as stop/delete/failover no longer hang when the destination side is stopped or deleted. 380942
EMC Celerra Network Server Version 6.0.70.4 Release Notes
289
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 388310 / 37319436
1
A Data Mover hung when an SMB thread tried to close and compress a file on a file system to which a replication thread was trying to unmount. DART panic/fault message: >>PANIC in file: ../dskdump.cxx at line: 1836 : cifs block. The stack trace showed: 0x2128bae0: 0x14124c waitForReboot+0x90 0x2128bb00: 0x1414d5 fault_dump+0x67 0x2128bb20: 0x1413bd PANIC+0x29 0x2128bb30: 0xa450c8 _Z11paniccfgcmdR6Option+0x178 0x2128bc70: 0x1cbdab _ZN11cfgcmd_desc6invokeEPKcPP 8CCMD_Msgii +0x1a7 0x2128bce0: 0x2b205a _Z10xml_cfgcmdP7macDataRP8CCMD_ Msg+0x392 0x2128bd30: 0x2ad00c _ZN17macCCMDCommand_DB6invoke EPKcP7macDa taRP8C 0x2128bd60: 0x2ad512 _Z17mac_CCMDDocParserP7macData RP8CCMD_Ms g+0x20
Replication code now verifies that the file system is not locked by any other threads under such conditions before trying to unmount a file system.
1
A bug which allowed a null value to be returned to SnapSure led to the following Data Mover panic: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: fa59c6 Err code: 0 Target addr: 4 **. The stack trace showed: 0x2c3b184: 0x14124c waitForReboot+0x90 0x2c3b1a4: 0x1414d5 fault_dump+0x67 0x2c3b1c4: 0x13a398 handle_fatal_fault+0xa 0x2c3b260: 0xfbabe0 _ZN10SnappedVol18writeBlockFi
The code has been fixed so that a NULL value is no longer returned, and no panic results.
382916 / 36634308
290
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
lteredEP10D iskIO_I 0xe9871240: 0xfbb091 _ZN10SnappedVol16aioInternalW riteEP10Dis kIO_IR 0xe98712a0: 0xfbaca8 _ZN10SnappedVol11aioInternalE P10DiskIO_I RPP8ai 0xe98712d0: 0x8c2361 _ZN11HyperVolume11aioInternal EP10DiskIO_ IRPP8a 0xe98713b0: 0x8d4007 _ZN17UnCachedIO_Object10lio_listi oE7LioM odePP8
EMC Celerra Network Server Version 6.0.70.4 Release Notes
291
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 383407 / 36730552
1
If an application label containing spaces was used to create an iscsi LUN replication, when user issued a reverse command, the Data Mover panicked with the following Assertion failure: 'dicSessionp != NULL' on line 931 in file '../DpTaskManager.cxx' >>PANIC in file: ../assert.c at line: 26 : assert failed. The stack trace showed: 0xe0031e30: 0x14124c waitForReboot+0x90 0xe0031e50: 0x1414d5 fault_dump+0x67 0xe0031e70: 0x1413bd PANIC+0x29 0xe0031e80: 0x1ff975 _assert+0x51 0xe0031ea0: 0x10c63c8 _ZN13DpTaskManager20createSyn cDicSession ERP15 0xe0031ee0: 0x10cba57 _ZN33DpTaskManagerCopyReverse Initiator8s tartS 0xe0031f30: 0x10caa07 _ZN33DpTaskManagerCopyReverse Initiator14 start 0xe0031f50: 0x10c844e _ZN33DpTaskManagerCopyReverse Initiator14 Start
The application label is now validated and rejected if it contains spaces.
Security
The following entries describe fixed Security related issues:
Severity Symptom Description Fix Summary Tracking Number 393878
1
When simultaneous access by two threads to the same SSL certificate was attempted, the Data Mover panicked with the following DART panic/fault message:
Added a mutex lock to protect the access to the SSL certificate and
292
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
>>PANIC in file: ../malloc.cxx at line: 351 : Memory corrupted. The stack trace showed: 0xe5fefc40: 0x140ce4 waitForReboot+0x90 0xe5fefc60: 0x140f69 fault_dump+0x67 0xe5fefc80: 0x140e51 PANIC+0x29 0xe5fefc90: 0x1fbdb7 free+0x69 0xe5fefcc0: 0xda5b38 CRYPTO_free+0xc 0xe5fefcd0: 0xd3fe2d BUF_MEM_free+0x2c 0xe5fefcf0: 0xd918ba X509_NAME_dup+0xe5 0xe5fefd10: 0xd8c7a1 ASN1_item_ex_free+0x171 2 The server_ldap command gave no indication whether the LDAP Service configuration succeeded. User needed to do additional queries or check the server logs to find out.
prevent this condition.
A new alert-level event was added for LDAP Service configuration failures. If a configuration fails, the following message will be in the sys_log and (if configured) emailed to the user: An attempt to configure the LDAP Service has failed. See the server log for applicable error messages. Correct the error, then retry the configuration command..
149215 / 29904638
EMC Celerra Network Server Version 6.0.70.4 Release Notes
293
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 385248
1
In version 6.0, code was added to log an event if an LDAP service configuration attempt failed. However, the code was added to several error loops that included 20 second delays before retrying the configuration attempt. the result was that the 20 second delay is always run whether there is an error or not, causing a delay in the service configuration. A race condition between periodic check for reconfiguration and manual reconfiguration may lead to an LDAP deadlock.
Fixed the error while loops to only run the 20 second delay if an error occurs.
1
This issue occurred when performing continuous LDAP reconfigurations in a short period of time. The LDAP timer no longer signals the event if already signaled.
383944
SRDF
The following entries describe fixed Symmetrix Remote Data Facility related issues:
Severity Symptom Description Fix Summary Tracking Number 371759
3
The fs_timefinder -M off command running against fs_group failed with the following error message. Error 13431668739: File system group has a file system which is backup of a Primary File System (PFS) which is actively mirroring to file system <fsname>. During a nas_rdf -restore, a gatekeeper lock condition should have forced the command to exit, the command was allowed to continue due to a unique condition. This caused the nas_rdf restore command to hang.
Removed the check which caused this issue. The fs_timefinder -M off command with fs_group now works correctly. The nas_rdf command will now exit when excessive lock contention is experienced. If a gatekeeper lock will not release, identify the lock and manually
1
382490 / 36173416
294
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
release it. Identify the pid holding the lock with the following command: /usr/symcli/bin/symcf g list -sem -v Release the lock with the following command: ipcrm -s <pid> Then rerun the nas_rdf -restore command.
Storage
The following entries describe fixed storage-related issues:
Severity Symptom Description Fix Summary Tracking Number 382625 / 36666666
1
When the source storage system was modified after the destination had been activated, the nas_mview -restore command failed.
Enhancements to the nas_mview command have been applied to detect storage system changes allowing the restore to complete successfully.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
295
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 376685
1
When paths to a device failed, and some (administrative) action caused a disk signature to be verified, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x0000a43da9 Err code: 0 Target addr: 0x0000000010 **. The stack trace showed: 0x3680760: 0xbf6450 _ZN7DskDump9checkdumpEPKc+0x7a 0x36807a0: 0xbf665e dump_return+0x4 0x36807b0: 0x14c7d2 _ZN12Memory_Frame13setProtect ionEm19Memo ry_Prot 0x36807d0: 0x14ce5d fault_dump+0x89 0x36807f0: 0x13378d handle_fault+0x1d 0x689fd20: 0xa4aec7 _ZN11BasicVolume21insertToIoQ ueueCommonE P8SgIO_ 0x689fd60: 0xa53518 _ZN11BasicVolume15IOBlockInternalEP 10Dis kIO_IRP 0x689fdf0: 0xa57ab9 _ZN11BasicVolume18writeBlockI nternalEP10 DiskIO_
Fixed a locking bug which caused stale queue data to be referenced.
1
When using UDP-based applications with large xfer sizes or using TCP and network environmental conditions caused fragmentation to occur, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 0x00003cb317 Err code: 0 Target addr: 0x0000000006 **. The stack trace showed: 0x3680760: 0xbf6490 _ZN7DskDump9checkdumpEPKc+0x7a 0x36807a0: 0xbf669e dump_return+0x4
Updated code to correct issues with fragment list management and prevent this pan
388143 / 37288224
296
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
0x36807b0: 0x14c7d2 _ZN12Memory_Frame13setProtect ionEm19Memo ry_Prot 0x36807d0: 0x14ce5d fault_dump+0x89 0x36807f0: 0x13378d handle_fault+0x1d 0x6880d90: 0x3c08e7 ip_mxrput+0x10d 0x6880dd0: 0x37b4bd _Z7bcmreadPvP4msgbtjj+0x4e7 0x6880ee0: 0x38b2b2 _Z20LM_ServiceInterruptsP16_L M_DEVICE_BL OCK+0x3 1 Periodically the Data Mover NIS client was unable to connect to the NIS server and switched to another NIS server. The server log reported an error like: The NIS client is unable to connect to the NIS server w.x.y.z : RPC failure. In some instances the NIS server can be slow to respond so the NIS RPC retry timeout is increased to 5 seconds and the NIS RPC request timeout increased to 15 seconds. This will prevent switching to another NIS server when there is a brief delay in the response from the NIS server. Checks were added in the code and error messages enhanced to determine what needs to be done to diskmark different configurations. 375293 / 34806928
1
User removed 5 LUNs from Celerra to change HLU’s (host LUN unit) and was unable to import them back even though no errors were logged. LUNs were part of a MirrorView/S consistency group and the command nas_mview -init needed to be used to diskmark them, but the Celerra error message gave no indication of this.
126624 / 25702698
EMC Celerra Network Server Version 6.0.70.4 Release Notes
297
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number 372043 / 35302478
1
The Data Mover received incorrect inquiry data from a backup tape device, which later triggered a panic due to dereferencing of a null volume pointer. The Data Mover panicked with the following DART panic/fault message: ** Page Fault Interrupt. Virt ADDRESS: 8b341f Err code: 0 Target addr: 0 **. The stack trace showed: 0x2c2b494: 0x141254 waitForReboot+0x90 0x2c2b4b4: 0x1414dd fault_dump+0x67 0x2c2b4d4: 0x13a3a0 handle_fatal_fault+0xa 0x2c2ba60: 0x296c17 _Z16macthr_sysconfigP4msgb+0x1b 5f 0x21e25fa0: 0x27c52a _ZN11Mac_Initial5startEv+0x3e0
The fix avoids nullpointer dereferencing when invalid Inquiry data is returned from a device, preventing the panic.
1
An attempt to specify more than 8 paths on a volume disk command resulted in a Data Mover panic with the following DART panic/fault message: *** GP exception. Virt ADDRESS: 13990eb. Err code: 0 ***. The stack trace showed: 0x2c4bcb8: 0x14124c waitForReboot+0x90 0x2c4bcd8: 0x1414d5 fault_dump+0x67 0x2c4bcf8: 0x13a398 handle_fatal_fault+0xa 0x2c4bd20: 0x1399061 _ZN14XmlElementNodeD0Ev+0x1d 0x25429cf0: 0x13991ca _ZN14XmlElementNode14removeCh ildrenEv+0x 2c 0x25429d10: 0x1399059 _ZN14XmlElementNodeD0Ev+0x15 0x25429d20: 0x13991ca
An attempt to specify more than 8 paths did not fail but caused memory corruption and a Data Mover panic. The code that parses volume disk commands has been modified to explicitly check and fail with the following error if more than 8 paths are specified: logIO::logmsg (LOG_STORAGE, LOG_ERR, 5825 logIO::logmsg (LOG_STORAGE, LOG_ERR "Cannot create Basic Volume %s, numpaths parameter invalid %d\n .
378239
298
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.41.3
Severity
Symptom Description
Fix Summary
Tracking Number
_ZN14XmlElementNode14removeCh ildrenEv+0x 2c 0x25429d40: 0x1398fc1 _ZN14XmlElementNodeD2Ev+0x15 1 When performing a PFS restore from the oldest checkpoint, if the SavVol became full leading to oldest checkpoint inactivation, the Data Mover panicked with the following DART panic/fault message: >>PANIC: I/O not progressing LastVol touched D6853119_0 Kind 8 (ptr=0x115ed0038. The stack trace showed: 0xc3addf0: 0xc566bd _ZN9RawIO_IRP13checkProgressEv+ 0x3e9 0xc3adfc0: 0xc56a3e _ZN17IrpProgressThread5startEv+ 0x4 0xc3adfd0: 0x16b4b1 Sthread_startThread_internal+0x 7 Code has been fixed to prevent the panic for this condition. 370887
Impact Level Symptom Desc.
Severity 1 Celerra kept reporting the trespassing of HDDs from SPA to SPB. Server log showed many CAM errors like this: 2010-01-11 13:11:42: 81874518019: CAM: 3: 3: Warning: The SCSI HBA 16 is inaccessible. 2010-01-11 13:11:42: 13155041280: CAM: 3: I/O Error: c16t0l0 Irp 0x57c25f04 CamStatus 0x05 ScsiStatus 0x08 Sense 0x05/0x04/0x00 2010-01-11 13:11:42: 13155041280: CAM: 3: I/O Error: c16t9l15 Irp 0x3c641104 CamStatus 0x05 ScsiStatus 0x08 Sense 0x05/0x04/0x00 and sys_log showed errors like this: Dec 8 23:40:28 2009:CS_PLATFORM:NaviEventMonitor:CRITICAL:101::::1260312028: Disk d8 has been trespassed. Dec 8 23:40:28
EMC Celerra Network Server Version 6.0.70.4 Release Notes 299
Version 6.0.41.3
Fix Service Request No.
2009:CS_PLATFORM:NaviEventMonitor:CRITICAL:101::::1260312028: Disk d11 has been trespassed. Dec 8 23:40:28 2009:CS_PLATFORM:NaviEventMonitor:CRITICAL:101::::1260312028: Disk d13 has been trespassed. Dec 8 23:40:28 2009:CS_PLATFORM:NaviEventMonitor:CRITICAL:101::::1260312028: Disk d14 has been trespassed. Enhanced the error status handling code to handle busy luns without forcing a trespass. 31963984
UFS
The following entries describe fixed UNIX File System issues:
Severity Symptom Description Fix Summary Tracking Number 382495 / 36659580
1
During the UTF-8 conversion process for a file system, if a Data Mover attempted to convert a directory owned by a user who was exceeding quotas, the Data Mover panicked with the following DART panic/fault message: Cannot read unnamed stream inode for container. The stack trace showed: 0xdf591310: 0x14124c waitForReboot+0x90 0xdf591330: 0x1414d5 fault_dump+0x67 0xdf591350: 0x1413bd PANIC+0x29 0xdf591360: 0x778b27 _ZN14UFS_FileSystem28markCorr uptedFsForU nmount 0xdf591520: 0x7cabba _ZN9UFS_Vnode18setUnnamedVnodeP trEv+0x192 0xdf591570: 0x777d74 _ZN14UFS_FileSystem16findExis tingNodeEmR 11File 0xdf5915c0: 0x777976 _ZN14UFS_FileSystem8findNodeE
Code was updated to be more tolerant of this error, and will no longer panic under this condition. The following error is returned if the conversion fails: UFS_Vnode::setUnna medVn odePtr -- fs <fsid> inum <inode number> Quota Exceeded\n where 'Quota Exceeded' can be user, group, tree quota, or just general file system out of space. The node being accessed will not be initialized, and the user will not be able to read the affected file/directory.
300
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
mR11File_St atust8 0xdf591610: 0x7e12ea _ZN15UFS_DirOps_dir38getVnode EP9UFS_Vnod emPP9V.
Version 6.0.40.8
Celerra Management Applications
The following are fixed problems related to the Celerra Management Applications, i.e., Unisphere, Celerra Startup Assistant, or Celerra MMC Snap-ins:
Severity Symptom Description Fix Summary Tracking Number 395097
1
User was unable to add or manage newer versions of Celerra systems to legacy domains in Unisphere due to restrictive version check.
Version check was modified to allow Celerra systems with newer version to be added to or managed in the domains of current version systems using Unisphere. Added 10 gigabit network devices for inclusion in Celerra Manager network graphs.
2
Network statistics were unavailable through Celerra Manager when Celerra had 10 gigabit network devices. Graph always showed "Data Loading..." and graph options menu didn't include 10 gigabit network devices.
379658 / 36124794
EMC Celerra Network Server Version 6.0.70.4 Release Notes
301
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 374949 / 35327164
2
When a Celerra was in production mode, Unisphere took a considerable time to load. The loading symbol (rotating arrows) was persistently shown near the IP address of the Control Station in the Unisphere left pane.
An underlying command was used to find the model of the Celerra system which caused serious latency for the loading of the Celerra pages in Unisphere. The command model –x was modified to include caching of the model of the system which speeds up the Unisphere pages. Code has been fixed to ensure that Jserver parsing code never fails on unexpected output lines and ensures that “server_cifs” processes spawned by Jserver always complete and prevent excessive memory utilization from occurring.
2
Jserver fails to parse output of the server_cifs server_x -o audit command due to some extraneous lines coming from Data Mover response resulting in the process hanging on Control Station. Over time, lots of hanging server_cifs processes consumed all the memory on the Control Station and the Control Station had to be rebooted.
375192 / 35638732
CIFS
The following entries describe fixed Common Internet File System problems:
Severity Symptom Description Fix Summary Tracking Number 384136 / 36849782
1
The Windows group Policy was maintained on the domain controller in unicode files. If a user replaced the unicode file with a non-unicode security template, then the file was not read properly and if large enough (greater than 54Kbytes) the Data Mover panicked with the following DART panic/fault message: NIC in file: ../malloc.cxx at line: 525 :
Updated the GPO code to check if the file is unicode (first two bytes of the file are 0xFFFE), then process the file either as unicode or nonunicode, retrieving and maintaining all settings
302
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Memory corrupted..
that are applicable to the Data Mover. Code now uses the FQDN of the first interface of the default CIFS server for the CIFS server. Code now frees the configuration and the streams used to send the heartbeat when VC is stopped. 365843 / 34585998
2
When implementing CAVA using a CIFS server with disjoint DNS and Windows domains, CAVA would not go online unless the default CIFS server was specified in viruschecker.conf. When doing a lot of CAVA /CEPA configurations while the CAVA service was stopped, the configuration was not freed and the streams used to send the heartbeat were not freed. This led to a memory leak which could result in a Data Mover panic.
2
381971 / 36571200
EMC Celerra Network Server Version 6.0.70.4 Release Notes
303
Version 6.0.40.8
1
Under heavy load, when the CEPP feature was enabled, Data Mover panicked with the following due to a race condition between two CIFS threads: DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: c601c2 Err code: 0 Target addr: 0 **. The stack trace showed: 0x2c3bc54: 0x14124c waitForReboot+0x90 0x2c3bc74: 0x1414d5 fault_dump+0x67 0x2c3bc94: 0x13a398 handle_fatal_fault+0xa 0x2c3bcd0: 0xc5b803 _ZN8fid_info15attachCloseCepp EP9CIFS_CEP P+0x59 0xfa6ffca0: 0xc60ada _ZN9CIFS_CEPP17setDeferredActio nEv+0x2a 0xfa6ffcc0: 0xc5262a _ZN8AppLibNT5closeEP8fid_info iR11File_St atusP1 0xfa6ffe80: 0xbf8302 _ZN13smb_streamCtx16closeOpen edFilesEt+0 xd2 0xfa6ffed0: 0xb09df2 _ZN13smb_streamCtx10credRemov eER11cRefer encerI
This race condition has been eliminated.
385161 / 36978078
1
After upgrading to 6.0 release, when using NTLM authentication, users from trusted domains might no longer be able to access resources (server, share, file, directory) granted through Universal group membership. DHSM connections (using Kerberos for authentication) try to re-use Kerberos tickets in the credential cache. The code used to check to see whether a ticket was available in the cache did not check to see if the ticket was expired, so the specified password was deleted. Subsequent code tried to use the expired ticket, then was unable to get a new ticket because the password had already been deleted, resulting in DHSM reconnection failure.
Code now correctly processes extra SIDs returned by the Domain Controller.
385626 / 36963164
1
Added code (in Kerberos) to check for expired ticket in credential cache, and to delete the ticket if it is expired.
375592 / 35601816
304
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
DART panic/fault message: >>PANIC in file: ../dskdump.cxx at line: 1836 : DHSM recall issues. The stack trace showed: 0x21201ae0: 0x14124c waitForReboot+0x90 0x21201b00: 0x1414d5 fault_dump+0x67 0x21201b20: 0x1413bd PANIC+0x29 0x21201b30: 0xa44e08 _Z11paniccfgcmdR6Option+0 x178 0x21201c70: 0x1cbdab _ZN11cfgcmd_desc6invoke EPKcPP8CCMD_Msgii +0x1a7 0x21201ce0: 0x2b205a _Z10xml_cfgcmdP7macDataRP 8CCMD_Msg+0x392 0x21201d30: 0x2ad00c _ZN17macCCMDCommand_DB6in vokeEPKcP7macData RP8C 0x21201d60: 0x2ad512 _Z17mac_CCMDDocParserP7 macDataRP8CCMD_Ms g+0x20
1
When users in Active Directory were in "lastname, firstname" format, the resulting distinguished name included an escaped comma; i.e., "\,". The Data Mover search filter processing for LDAP did not process escaped commas correctly, and returned incorrect search results.
Code was modified to allow an escaped comma to be included in the search filter string.
369461 / 3480317
EMC Celerra Network Server Version 6.0.70.4 Release Notes
305
Version 6.0.40.8
2
Data Mover stopped parsing nsswitch.conf file when encountering an invalid entry in it, the rest of the configuration lines were rejected and no error was reported about it which might prevent certain functionality from working, such as LDAP for CIFS UID/GID mapping. User tried to disable file filtering temporarily through the cifs.enableFileFiltering parameter. The parameter description says it applies immediately but it was found that changing the value had no effect if the connection was still opened. An attempt by client to create file "\nasadmin\New Microsoft Word Document.doc" failed with STATUS_ACCESS_DENIED. The Data Mover allowed the write after the Data Mover was rebooted.
Code has been fixed to parse the whole file and ignore invalid entries, but report an error message in the server_log about the invalid entry. The parameter was modified to not require any user action (rebooting), even for connection active when the parameter has been changed.
372383 / 35287884
1
373375 / 35092392
Control Station
The following entries describe fixed problems related to the Celerra Control Station:
Severity Symptom Description Fix Summary Tracking Number 370795 / 35139496
2
The Data Mover did not failover after the panic. This occurred because the code did not notify all the processes to release the process locks, and a checkpoint related process was holding a lock which prevented the failover. Users observed the following error in the GUI alert log: Parsing error. Invalid XML. The message did not contain any specific information about the parsing error.
Code now ensures that all processes are appropriately notified to exit and release the locks.
2
The function that logs the messages containing the reason for the parsing failure was altered to log the entire message with the specific reason of the parsing error in /nas/log/nas_log.al. mgmtd file. This
361847 / 33292418
306
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
message will contain the entire xml string, including in this case, the invalid characters. 2 When NTP services were started on a Data Mover with the interval greater than 1440 minutes specified, it was set on the Data Mover but then the Data Mover panicked with the following DART panic/fault message: *** Divide exception. Virt
ADDRESS: a2edaf ***. The stack trace showed:
Code now prevents a value greater than 1440 minutes (or 24 hours) for NTP sync interval.
361893 / 34311052
0x2ba3d0c: 0x140ce4 waitForReboot+0x90 0x2ba3d2c: 0x140f69 fault_dump+0x67 0x2ba3d4c: 0x139ee4 handle_fatal_fault+0xa 0x2ba3ed0: 0xa2fa0c _ZN8TimeSync5startEv+0xb2 0xfdd37fe0: 0x1455cc Sthread_startThread_internal+0xe
CORE
The following entries describe fixed problems related to Celerra Data Movers:
Severity
Symptom Description
Fix Summary
Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
307
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 374681 / 35598996
1
A rolling panic occurred when a local netgroup file contained a nested netgroup. Users had CIFS access, but no NFS access. DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 14dd7a Err code: 2 Target addr: 8f **. The stack trace showed: 0x2c2cce4: 0x141254 waitForReboot+0x90 0x2c2cd04: 0x1414dd fault_dump+0x67 0x2c2cd24: 0x13a3a0 handle_fatal_fault+0xa 0x2c2cd50: 0x1c54a3 _Z13CollectStreamPvS_+0x77 0x5081e30: 0x1d577a ReadPutq+0x2b2 0x5081e60: 0x6350eb nfs_rput+0xa47 0x5081eb0: 0x634b8c nfs_rput+0x4e8
Code now processes only ten levels of nesting in a netgroups. If there exist netgroups nested to a depth greater than ten, the following error message is now printed in the server log: NETGROUP netgElem::IsMember Looping on <netgroup name> (loop in nested netgroups) \n.
1
When a Data Mover was mistakenly zoned with more than 32 other initiators, the Data Mover was unable to see the storage and failed to boot.
The Data Mover will now detect when too many devices are incorrectly zoned into the Data Mover and will log diagnostic information that will help quickly identify the issue. The following messages will appear in the server log under such conditions: Failed to add device, no resources and Failed to add device, unsupported login type
366770 / 34764834
308
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Deduplication
The following entries describe fixed issues pertaining to Celerra Deduplication:
Severity Symptom Description Fix Summary Tracking Number 375789 / 35737678
1
On a system with heavy deduplication activity, the Data Mover did not release memory in the EDRS (EMC Data Reduction Services) portion of the code and caused a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../addrspac.cxx at line: 289 : Out of memory. The stack trace showed: 0xde21b0a0: 0x14124c waitForReboot+0x90 0xde21b0c0: 0x1414d5 fault_dump+0x67 0xde21b0e0: 0x1413bd PANIC+0x29 0xde21b0f0: 0x13bef6 _ZN19Memory_AddressSpace10all ocPagesEi12 Memory 0xde21b110: 0x1fe726 _Z11more_memoryjP9free_list+0x4 a 0xde21b130: 0x1fe7ec malloc+0xa0 0xde21b160: 0x152dbc2 csx_rt_os_mem_raw_alloc_maybe+0 x14 0xde21b180: 0x152bc16 csx_rt_mem_raw_alloc_maybe+0x14
Code was fixed to correctly release memory and prevent this panic.
1
Under rare condition, removing a deduplicated file, creating a checkpoint and performing a restore on the same file system at the same time, resulted in a deadlock due to lock interdependencies resulting in blocked threads on the Data Mover and eventually causing access issues.
Code has been fixed to properly handle interdependencies lock so no deadlock would occur under this condition.
372616 / 35375114
EMC Celerra Network Server Version 6.0.70.4 Release Notes
309
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 374310 / 35478060
1
File systems with deduplication enabled (including NDMP backup checkpoint file systems) failed to unmount. Messages similar to the following were logged to the server log: 2010-07-28 19:24:11: DEDUPE: 3: openFileData failed: fsid 483 status RDE_STATUS_DECOMPRESSION_ER ROR. When a read of a dedupe file system resulted in sending a larger request than the fibre driver could process, the Data Mover panicked with following DART panic/fault message: >>PANIC in file: ../BVolumeIrp.cxx at line: 169 : Problem building SG list. Panicking Processor: 2. The stack trace showed: ee2a90: 0xa9e735 _ZN8SgIO_IRP10initializeEP10Dis kIO_IRPP11B asi ee2ad0: 0xa54b3c _ZN11BasicVolume11aioInternalEP 10DiskIO_IR PP8 ee2ba0: 0xa55cd7 _ZN11HyperVolume11aioInternalEP 10DiskIO_IR PP8 ee2c70: 0xa5f213 _ZN11SliceVolume11aioInternalEP1 0DiskIO_IR PP8 ee2ca0: 0xa55cd7 _ZN11HyperVolume11aioInternalEP 10DiskIO_IR PP8 ee2d70: 0xa55cd7 _ZN11HyperVolume11aioInternalEP 10DiskIO_IR PP8 ee2e40: 0x133ccb5 _ZN10SnappedVol11aioInternalEP1 0DiskIO_IRP P8 ee2e70: 0xa55cd7 _ZN11HyperVolume11aioInternalEP 10DiskIO_IR PP8 This occurred was because of a limitation in the fibre driver code to support up to 64
Code is updated to ensure files are closed when a DECOMPRESSION ERROR is detected.
1
Code now breaks larger requests into chunks of 64 entries to prevent this sort of panic.
379947 / 36247340
310
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
request blocks only.
File System
The following entries describe fixed bugs pertaining to File Systems:
Severity Symptom Description Fix Summary Tracking Number 165384 / 33310518
1
If a frozen file system was unmounted while the system was out of vnodes, the naming node became inconsistent which led to a Data Mover panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 6d8dd7 Err code: 0 Target addr: 1d4 **. The stack trace showed: 0x2acb734: 0x13fb94 waitForReboot+0x90 0x2acb754: 0x13fda9 fault_dump+0x67 0x2acb774: 0x138d94 handle_fatal_fault+0xa 0x2acb7b0: 0x7dc2c4 _Z17MPDShadowCallBackP15File_ NamingNodeP K18File 0x24a45b00: 0x7cf46a _Z14ShadowCallBackP15File_Nam ingNodePK18 File_N 0x24a45c00: 0x7777a6 _ZN9UFS_Vnode19doDirChangeCal lBackEPK18F ile_Na 0x24a45c30: 0x694d7f _ZN12File_DirNode21lookupComp onentLocked EPK18F 0x24a45d00: 0x69b627 _ZN16File_RootDirNode21lookup ComponentLo ckedEP
The naming node is now appropriately replaced under this condition to avoid the inconsistency.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
311
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 370855 / 35172362
1
When user tried to mount a root file system in read-only mode on a stand-by Data Mover, the Data Mover panicked with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 794170 Err code: 0 Target addr: 40 . The stack trace showed: 0x2898554: 0x13fa6c waitForReboot+0x90 0x2898574: 0x13fc69 fault_dump+0x67 0x2898594: 0x138cfc handle_fatal_fault+0xa 0x28985e0: 0x75723c _ZN9UFS_Vnode10lookupCompEPK 18File_Name Componen 0xc48a77d0: 0xec4db4 _ZN10MGFS_Vnode10lookupCompE PK18File_Na meCompo 0xc48a7810: 0x6f7de3 _ZN18UFS_FileSystemKind6init FSEPKcP9IO_ Objecti 0xc48a7900: 0xec1f89 _ZN19DHSM_FileSystemKind6ini tFSEPKciR6O ptionP1 0xc48a7980: 0x6cc267 _Z25File_MountCommandInterna lR6OptionP1 8VFS_Fi
During a mount, code now checks whether the default log device is initialized or not. If not initialized, the following error is returned: Log device not initialized.
1
Issuing a quota check command at the same time as a quota operation from a Windows client resulted in a deadlock condition and rendered the file system inaccessible until the Data Mover was rebooted. 1. During any version 5.6 to 6.0.36.4 upgrade, if the quota policy was set to filesize (which is not the default blocks configuration), the task for upgrading Data Mover could timeout because the Data Mover takes too long to reboot due to an automatic tree quota check. 2. While running version 6.0.36.4,
The fix has the quota command from Windows abort requiring the quota lock while the FS is inactive. Code no longer forces a quota check during the reboot procedure, following a quota policy change. See Primus emc254145 for more details.
381651 / 36551080
1
388085 / 37305372
312
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
changing the quota policy and rebooting the Data Mover could cause a long reboot, due to the automatic tree quota check.
FileMover (DHSM)
The following entries describe FileMover-related fixed bugs:
Severity Symptom Description Fix Summary Tracking Number 378918 / 36077544
1
User saw incorrect data when reading stub files on loopback connections. This issue occurred because of the way that offsets and read sizes were calculated on read operations which led to read timeouts. A Data Mover exhibited high CPU usage when DHSM was enabled and in-use on one or more of the file systems. Errors similar to the following were displayed in the server log. 1. "error direct read block xxx of entry xxx, status 20 - EndOfFile", and/or 2. "File size mismatch: Stub file xxxx bytes, migration vnode xxxx bytes,ino=xxxx status=EndOfFile"
Code now correctly adjusts the read offset and size on error recovery read operations.
1
Fixed a logic error in the DHSM I/O (read/recall) path to prevent this infinite loop.
372101 / 34673002
Install/Config/ Upgrades
The following entries describe fixed issues related to installations, configurations and upgrades:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
313
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 372086
1
During an NS4-120 installation, the "setup_enclosure detectSingleDMSystem" failed with the following error: Discover 1 Mgmtswitches in Primary subnet A, 0 Mgmtswitches in Secondary subnet B Please check cables and Mgmtswitches on Secondary subnet B Error: detectSingleDMSystem FAIL (EMISMATCHCOUNT) Remove Dynamic IP range in DHCPD_CFG ... OK Restart DHCPD with updated DHCPD_CFG ... OK Error: UNDEFINED retval = -10 (EMISMATCHCOUNT). While performing one of the following Upgrade/Conversion procedures, setup_slot failed while adding a new I/O module in single Data Mover system: -Upgrade from one blade to two blades -Add a second SFP for second Tape support -Add an I/O module (to all blades simultaneously) -Changing to the faster I/O Module of the same type -Add a second Control Station. During Express install, the "set_up_Data_Mover" process took 40 minutes to complete on a 4 Blade setup. This was due to a built-in delay during a reboot.
Code was modified so that the install will not fail if it does not discover the switch on subnet B in a single Data Mover system. It will be discovered by the setup_enclosure initSystem command which is run later during the install process.
2
The wrong reason code was being monitored for these procedures, and because the system never reached that reason code (5), the command returned an error. This has been fixed.
374063
1
The fix eliminates the delay when issuing a reboot with NAS services down.
388839
iSCSI
The following entries describe fixed iSCSI-related issues:
Severity Symptom Description Fix Summary Tracking Number 377716 / 33424708
2
When client demoted an iSCSI snap LUN using the force flag, the iSCSI remove LUN operation logged an error instead of
The server log message for this case is not a warning. An
314
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
a warning message in the server log. Users are likely to see this when using EMC Replication Manager because ERM internally demotes snap with force flag on its replication destination Celerra.
error message is only logged when the remove LUN operation actually fails.
NDMP
The following entries describe fixed issues related to Network Data Management Protocol and/or Backup:
Severity Symptom Description Fix Summary Tracking Number 371282 / 35158192
1
An uncooperative peer closed the TCP receive window during an NDMP backup. This led to the following panic because the NDMP was waiting on a TCP stream completion that never came. DART panic/fault message: >>PANIC: I/O not progressing LastVol touched Sn128 Kind 5 (ptr=0xa26d6104). The stack trace showed: 0x79dbe90: 0x13fb94 waitForReboot+0x90 0x79dbeb0: 0x13fda9 fault_dump+0x67 0x79dbed0: 0x13fc91 PANIC+0x29 0x79dbee0: 0x8478e0 _ZN9RawIO_IRP13checkProgressEv+0 x11a 0x79dbfd0: 0x8477bc _ZN17IrpProgressThread5startEv+ 0x6 0x79dbfe0: 0x144344 Sthread_startThread_internal+0x e
This was fixed by setting the abort flag for 3-way VBB operation, should an unmount or internal error require abortion of the backup/ restore operation.
Network
The following entries describe fixed network-related issues:
EMC Celerra Network Server Version 6.0.70.4 Release Notes 315
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 371067 / 35186226
1
LACP could not be used with Dell Poweredge 6248 network switch and Celerra. The following error message appeared in the server_log: DRIVERS: 3: LAP_Port::receive_LACP len=114 expected_len=110. The verification of the PDU length was too strict, where it should have been verifying the PDU is a least the minimum size of 110. When the client sent a packet greater than the MTU size (between 1518 and 1530 bytes), the Broadcom device hung. The Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../dskdump.cxx at line: 1836 : network issue. The stack trace showed: 0x1f635ae0: 0x14124c waitForReboot+0x90 0x1f635b00: 0x1414d5 fault_dump+0x67 0x1f635b20: 0x1413bd PANIC+0x29 0x1f635b30: 0xa45118 _Z11paniccfgcmdR6Option+0x178 0x1f635c70: 0x1cbdab _ZN11cfgcmd_desc6invokeEPKcPP 8CCMD_Msgii +0x1a7 0x1f635ce0: 0x2b205a _Z10xml_cfgcmdP7macDataRP8CCMD_ Msg+0x392 0x1f635d30: 0x2ad00c _ZN17macCCMDCommand_DB6invoke EPKcP7macDa taRP8C 0x1f635d60: 0x2ad512 _Z17mac_CCMDDocParserP7macDat aRP8CCMD_Ms g+0x20
The PDU length verification has been changed to accept anything not less than the expected PDU length.
1
Fixed the Broadcom 10G hang issue by correcting the MTU (max transmission unit) size passed to the device.
376090 / 35799674
1
An exploit of the NFS mount protocol was identified where an attacker could spoof IP addresses that are normally used between the Celerra Control Station and Data Mover over a private IP network and gain unauthorized access to file systems on the Celerra.
Code has been fixed to reject all mount requests using the Control Station IP addresses that do not come over from a private network that is internal to the Celerra
118547
316
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
cabinet, eliminating this vulnerability.
NFS
The following entries describe fixed issues related to Network File Systems:
Severity Symptom Description Fix Summary Tracking Number 372545 / 35327318
1
When using NFSv4, a missing reference for NFSv4 state object on Data Mover caused a Data Mover to panic with the following DART panic/fault message: *** Page Fault Interrupt. Virt ADDRESS: 66fdf2 Err code: 2 Target addr: 24 **. The stack trace showed: 0x2c2aba4: 0x141254 waitForReboot+0x90 0x2c2abc4: 0x1414dd fault_dump+0x67 0x2c2abe4: 0x13a3a0 handle_fatal_fault+0xa 0x2c2ac60: 0x66e352 _ZN16NFSv4_StateTable5CloseEm P8stateid4P 8nfs4_c 0xe10b5d50: 0x65d129 _Z10nfs4_closeRP4msgbS1_RPhR8 nfs4_ctx+0x 18b 0xe10b5d90: 0x6592b0 _Z23nfs4_convertNfsv3Statusm+0x 626 0xe10b5ee0: 0x63b64d _ZN17nfs_serverInitial5startEv+ 0xad7 0xe10b5fe0: 0x145bd6 Sthread_startThread_internal+0x e
Code has been fixed to properly handle missing reference on the object and prevent this panic.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
317
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 377431 / 35858230
2
NFSv4 locks could not be reclaimed after a temporary network outage. If an NFSv4 client temporarily lost contact with the Celerra and attempted to reclaim locks after its lease expired (20 seconds by default), the operation would fail. Access to the file from the process that owned the lock could be impacted, though if the process that owned the lock specifically released the lock, access would be restored. This problem was introduced in 5.6.45. NFSv4 code did not check permissions when modifying the ACL based on the mode bits. This can result in access behavior inconsistent with permissions. Unable to write data to a File Level Retention Enterprise (FLR-E) files system if the file system was mounted using NFS V4.
Fixed a miscomputation of the NFSv4 owner sequence number in Open-Claim-Previous operation. Attempts to reclaim locks will now succeed.
1
Code now correctly applies permissions when changing the NFSv4 ACL. Code has been fixed to ensure users can write data to FLR-E file systems mounted using NFS V4. Code has been fixed to properly release allocated memory for dropped requests, preventing Data Mover panics in unhealthy network environments.
378959 / 36091074
2
368246
1
In an unhealthy network, where many network packets were lost, memory on the Data Mover was exhausted by those many dropped requests. This led to a Data Mover panic with the following DART panic/fault message: >>PANIC in file: ../strbufs.cxx at line: 1747 : out of msgb. The stack trace showed:
379200
0xe2d6960: 0x20adcc _ZN15Buffer_MsgbPool3getEv+0 xac
0xe2d6990: 0x20af9a allocb+0x54 0xe2d69c0: 0x6e5ec8 nfs_rput+0x2c0 0xe2d6ab0: 0x40aa3e udp_input+0x9fa 0xe2d6c00: 0x443879 _Z9ip6_inputP10IPDevice_tP4msgb i+0x10cb
EMC Celerra Network Server Version 6.0.70.4 Release Notes
318
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
0xe2d6d60: 0x4316f0 _Z10ip6_mxrputP5queueP4msgb+0x1 3a 0xe2d6dd0: 0x37b4bd _Z7bcmreadPvP4msgbtjj+0x4e7 0xe2d6ee0: 0x38b2b2 _Z20LM_ServiceInterruptsP16_L M_DEVICE_BL OCK+0x3
Replication
The following entries describe fixed problems encountered while using Celerra Replicator, SnapSure or Checkpoints:
Severity Symptom Description Fix Summary Tracking Number 360751 / 34233340
1
The Data Mover on the destination side of a ReplicationV2 session panicked after a failover issued using the session name. This happened because the replication database contained an old entry for an inactive session with the same name and the failover operation used the old entry rather than the new one. When attempting to create an iSCSI replication session, if version set was out of space, the task showed success when it actually failed.
The session ID is now used during the failover since it will always be unique.
2
The iSCSI replication create task status was changed to Severity=error. The following will be displayed in the task status: Version set out of space.
374003
EMC Celerra Network Server Version 6.0.70.4 Release Notes
319
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 366022 / 34665376
1
In a cascade replication scenario, a replication session for the middle hop to the third hop was created before creating a session for the first hop to the middle hop. The code allowed a file system that acted as replication source and destination (middle hop) to be mounted as read-write. This violated replication destination read-only property assumption and caused a corruption on that file system which lead to Data Mover panic with the following DART panic/fault message: >>PANIC in file:
../ufsdirhash.hxx at line: 623 : UFS_DirHashIter_dir3::getEntry: mangled entry. The stack trace showed: 0x2b1f7050: 0x140ce4 waitForReboot+0x90 0x2b1f7070: 0x140f69 fault_dump+0x67 0x2b1f7090: 0x140e51 PANIC+0x29 0x2b1f70a0: 0x7b045a _ZN20UFS_DirHashIter_dir38getEn tryEm14Bu fInten 0x2b1f70e0: 0x7c2e9d _ZN20UFS_DirHashIter_dir34findE mPKcm13Fi le_Nam 0x2b1f7110: 0x7d1dd4 _ZN20UFS_DirHashIter_dir34findE mPK18File _NameC 0x2b1f7150: 0x7c3ad7 _ZN15UFS_DirOps_dir37scandirEP9 UFS_Vnode PK18Fi 0x2b1f71d0: 0x7cd806 _ZN15UFS_DirOps_dir311getSynony msEP9UFS_ VnodeP
A new check was added to the code, which prevents the read-write mount of a file system that acts as replication source and destination in a cascade replication.
2
While trying to perform an out-of-family Rep V2 replication, whenever a server with a lower slot number was made as standby for a server with a higher slot number, the multicast session lost the lower slot number server from the target list. When user invoked the nas_fs -delete command, the following error was generated: Error 5005: DpRequest_Max_VS_SuspendedCheckp ointsReach ed.
When there is a change in the server state, CIC (Celerra Interconnect) multicast target list is now generated from scratch rather than being modified. This solves the problem of a target being missed due to state and relation change of the servers.
144857
320
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
2
While using Replication V2, if a Data Mover was rebooted during a nas_copy process, the data transfer completed successfully, but the command did not complete and hung after the reboot. During merge operations involving file system checkpoints, the TOC (Table Of Contents) became inconsistent with the trailer and map. This inconsistency led to the following panic which required an unmount of the bad checkpoint before the Data Mover could be stabilized. DART panic/fault message: D130832_0: Btree_Insert failed, mark ckpt corrupted. The stack trace contained various messages similar to the following: 0x20643f40: 0x141254 waitForReboot+0x90 0x20643f60: 0x1414dd fault_dump+0x67 0x20643f80: 0x1413c5 PANIC+0x29 0x20643f90: 0xfb36ad _ZN28IndexChainCorruptMarkReq uest13handl eReque 0x20643fc0: 0xfb3312 _ZN23CkptCorruptMarkerThread5st artEv+0x80 0x20643fe0: 0x145bd6 Sthread_startThread_internal+0x e
The nas_copy command will now complete after a Data Mover reboot.
376684
1
Code now verifies the TOC entries against the other 2 references before the merge, during the merge, and after merge completes. Code then corrects the TOC as needed.
366985 / 34775056
1
Network issues might cause RepV2 session hang.
Code now promotes the RCP_CLOSE message priority so it will not cause blocking.
156084
EMC Celerra Network Server Version 6.0.70.4 Release Notes
321
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 383673 / 36791874
1
If the internal checkpoint name exceeded 16 characters, memory corruption occurred which caused a system panic such as I/O not progressing and volume display error. DART panic/fault message:
>>PANIC in file: ../sched.cxx at line: 846 : SYSTEM WATCHDOG. The
stack trace showed:
0x5089f24: 0x14124c waitForReboot+0x90 0x5089f44: 0x1414d5 fault_dump+0x67 0x5089f64: 0x1413bd PANIC+0x29 0x5089f74: 0x1432c9 system_watchdog+0x77 0x5089f84: 0x139f6a _t_nmi+0x68 0x5089fe0: 0x145bfa Sthread_startThread_internal+0xe
Checkpoint internal names whose length exceeded 16 characters were truncated resulting in unexpected behavior for example the name D2354619762_10113 could be truncated to D2354619762_1011 which does not exist. The naming convention was changed to shorten the name. The new format of the name is "D<ckptFsID>_<ckptG enCnt>", which keeps the total length below 16 characters and maintains the uniqueness of the name. This checkpoint name is only used internally for debug/display purposes. It can be seen in the server_log during checkpoint refresh operations for example 2010-09-20 06:42:01: 26042826752: SVFS: 6: D2354619762_1011: Recovered(Stable Index) FsID:23546 state:Active SavVol:29390 FsVol:D2480619762_ 0 rsv=2 h=12322040 c=12322057 t=12322057, in 0 Msec 2010-09-20 06:42:01: 26042826752: SVFS: 6: D2354619762_1011:
322
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number
currDel:D235461976 2_101 13 st:Active prevDelta:1573888
1
Replication V2 sessions were not automatically synchronized. The timer being used to check if it is time to transfer was using an overflowed value.
The interval between checks of whether it is time to sync a session now uses a consistent value rather than a calculation which could overflow.
386995 / 37192564
Security
The following entries describe fixed security- related issues:
Severity Symptom Description Fix Summary Tracking Number
EMC Celerra Network Server Version 6.0.70.4 Release Notes
323
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 376153 / 35803348
1
The LDAP configuration on a Data Mover was using multiple LDAP servers, and LDAP was configured to use the Active Directory "member" attribute instead of "memberUid". When the network connectivity to all LDAP servers was bad, memory corruption occurred and the Data Mover panicked with the following DART panic/fault message: >>PANIC in file: ../malloc.cxx at line: 351 : Memory corrupted . The stack trace showed: 0xdb9477e0: 0x14124c waitForReboot+0x90 0xdb947800: 0x1414d5 fault_dump+0x67 0xdb947820: 0x1413bd PANIC+0x29 0xdb947830: 0x1fe8e7 free+0x69 0xdb947860: 0x1fefb8 _ZdlPv+0x10 0xdb947870: 0x4f6582 _ZN10LdapClientD1Ev+0x32 0xdb947890: 0x53f311 _ZN16LdapDomainClient13getgrb ymemberEPKc PPP12A 0xdb947bd0: 0x69b7f3 _ZN20Access_GroupDataBase15ad dNewExtraGi dsEP18
Code was modified to correctly handle failover between LDAP servers and prevent memory corruption.
1
NTP software on the Control Station was vulnerable to the denial-of-service attack described in US CERT VU568372 when NTP service is running (disabled by default).
Code has been fixed to pickup the nonvulnerable version of NTP software from RedHat which prevents this denial-of-service attack.
364874 / 34476786
Storage
The following entries describe fixed storage-related issues.
324
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Version 6.0.40.8
Severity
Symptom Description
Fix Summary
Tracking Number 369324 / 35030006
1
A Data Mover watchdog panicked due to FC environmental conditions. This can happen when the processing used to clear some FC conditions takes too long due to unnecessary reads across the PCI bus. Direct attached tape drives connected with cable lengths greater than 50m may not be discovered after upgrade to 5.6.47.11 and above. The issue was caused by a race condition that was triggered by increased latencies using longer cables.
The unnecessary reads on the PCI bus have been removed during FC processing.
1
The code has been fixed to avoid race condition when latency is increased due to long cables.
381342 / 36335402
UFS
The following entries describe fixed UNIX File System issues:
Severity Symptom Description Fix Summary Tracking Number 361139 / 34143216
1
In an environment with VMware ESX server that was using Celerra iSCSI LUNs for its VMFS file systems, the ESX server had aborted an iSCSI session and disconnected from iSCSI LUNs while a large iSCSI snap was being deleted on the Celerra. This caused an outage on the ESX virtual machines. The following messages in server_log indicate the issue: SCT: 6: Task Management Function received: Abort Task, I_T Nexus: [<iqn>] ISCSI: 3: IscsiSession::closeSession: I_T nexus lost, sess=0xe0b6dc04 from [<iqn>]
The Data Mover code has been updated to improve response time during the iSCSI snap delete process. From the ESX server side, VMware updated ESX server code to allow longer timeout before aborting an iSCSI session.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
325
Troubleshooting and getting help
Troubleshooting and getting help
This section provides specific instructions for contacting EMC Customer Support and obtaining additional information about EMC products.
Where to get help
EMC support, product, and licensing information can be obtained as follows. Product information: For documentation, release notes, software updates, or for information about EMC products, licensing, and service, go to EMC Online Support (registration required) at: http://Support.EMC.com Troubleshooting: Go to EMC Online Support. After logging in, locate the appropriate Support by Product page. Technical support: For technical support and service requests, go to EMC Customer Service on EMC Online Support. After logging in, locate the appropriate Support by Product page and choose either Live Chat or Create a service request. To open a service request through EMC Online Support, you must have a valid support agreement. Contact your EMC Sales Representative for details about obtaining a valid support agreement or to answer any questions about your account.
Note: Do not request a specific support representative unless one has been assigned to your particular system problem.
326
EMC Celerra Network Server Version 6.0.70.4 Release Notes
Copyright © 2013 EMC Corporation. All Rights Reserved. Published in the USA. Published February 2013 EMC believes the information in this publication is accurate as of its publication date. The information is subject to change without notice. THE INFORMATION IN THIS PUBLICATION IS PROVIDED "AS IS." EMC CORPORATION MAKES NO REPRESENTATIONS OR WARRANTIES OF ANY KIND WITH RESPECT TO THE INFORMATION IN THIS PUBLICATION, AND SPECIFICALLY DISCLAIMS IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Use, copying, and distribution of any EMC software described in this publication requires an applicable software license. EMC2, EMC, and the EMC logo are registered trademarks or trademarks of EMC Corporation in the United States and other countries. All other trademarks used herein are the property of their respective owners. For the most up-to-date regulatory document for your product line, go to the technical documentation and advisories section on the EMC online support website.
EMC Celerra Network Server Version 6.0.70.4 Release Notes
327
