VERITAS Volume Manager (tm) 3.1 for Windows 2000 hotfix and service pack fixed incident history

Article:TECH29569  |  Created: 2005-01-27  |  Updated: 2005-01-27  |  Article URL http://www.symantec.com/docs/TECH29569
Article Type
Technical Solution

Environment

Issue



VERITAS Volume Manager (tm) 3.1 for Windows 2000 hotfix and service pack fixed incident history

Solution



This document provides a centralized historical reference of all issues resolved in each hotfix and service pack release of Volume Manager 3.1 for Windows  2000. This document will be updated with the release of each new hotfix or service pack of Volume Manager 3.1 for Windows 2000.

Note: For downloads for the hotfixes and service packs, go to the Related Documents section of this TechNote

The information in this document is taken directly from the readme.txt file of each hotfix or service pack and consolidated into this document. For complete details of any hotfix or service pack, VERITAS recommends reading the entire readme.txt file provided with the hotfix or service pack.

The list of issues resolved is provided in reverse chronological order:

HotFix 05 - Released April 22, 2005
HotFix 04 - Released June 23, 2004
HotFix 03d - Released December 29, 2003
Service Pack 1 - Released September 10, 2003
HotFix 02a- Released July 22, 2003
HotFix 01f- Released March 12, 2003

--------------------------
HotFix05 - Released April 22, 2005

ISSUES RESOLVED IN THIS HOTFIX

1. [DMP] Enhancement: Add DMP support for IBM FAStT and STK arrays (Incident 209639).
   After this hotfix is applied, DMP will be able to support the following arrays:
   - STK BladeCtlr B210 (FAStT100)
   - STK BladeCtlr B220
   - STK BladeCtlr B280

2. The vxstat utility may terminate abnormally (Incident 218027).

3. [DMP] New LUNs that were dynamically allocated arrive in VM successfully, but they are not kept by the system and cannot be used (Incident 220201).  Note: If the system uses the Emulex full port driver, the options "Report unknown SCSI device" and "Look for disappearing device" must be checked when configuring the HBA.

4. If a volume fails because the healthy plex fails during a mirror resynchronization operation and the volume is then reactivated, the mirror resynchronization operation completes successfully,     however the volume remains in a "Resynch" state (Incident 223164).  This issue is resolved by this hotfix so that the volume state changes to "Started" after the resynchronization operation completes.

5. [CLI/API] The command vxvol -v volinfo displays the incorrect subdisk offset (Incident 229755).  This issue is resolved by this hotfix so that vxvol -v volinfo displays the correct subdisk offset.

6. [MSCS] With vxclus enabled on a dynamic quorum cluster disk group, the disk group cannot be onlined with MSCS Cluster Administrator (Incident 230431).  In VSFW 3.1, even after running the vxclus enable command (vxclus enable -g<DynamicQuorumClusterDiskGroup>) on a target cluster disk group that contains the dynamic quorum volume, the MSCS service cannot be started successfully.

7. [DMP] The test path retry mechanism for monitoring I/O paths is not enabled by default installation (Incident 235099).  This issue is resolved by this hotfix so that the test path retry mechanism is enabled by default.

8. Performing a snapback operation on a very large volume may take as much time as the snapstart operation (Incident 236138).  In general, the snapback operation requires less time to complete than the snapstart operation on the same volume because only modified blocks are copied. This hotfix resolves this issue for large volumes so that the snapback operation requires less time than the snapstart operation.

9. Under conditions of heavy system stress, a system with RAID-5 volumes may encounter a fatal system error (BSOD) (Incident 260544).

10. Taking repeated snapshots of a volume may cause a snapshot error, resulting in failure of the snapshot operation (Incident 261387).

11. [FMR] If a mirror of a snapshot volume is created and the original source volume is deleted, the operation to break the mirror of the snapshot volume fails (Incident 267260).

12. A VCS MountV resource does not come online if the volume name or VM DiskGroupName is typed in the wrong case (Incident 268716).  This issue is resolved by this hotfix so that the volume name or VM DiskGroupName can be entered insensitive to case.

13. Upgrading or repairing a system that has VM 3.1 installed and has Windows Driver Verifier enabled may cause an abnormal termination (BSOD) (Incident 269095).

14. [FMR] If a system fails over or is rebooted during a mirror resynchronization operation, the resulting mirror is unusable after the system fail over or reboot completes (Incident 271629).  In addition, the Reactivate Disk command on the context menu of the mirror is grayed out and unavailable so that the mirror cannot be resynchronized (Incident 271637).  This issue is resolved by this hotfix so that the mirror is usable after system fail over or reboot and the Reactivate Disk command is available in the context menu.

15. [DMP] Running vxassist on the active cluster node after a failover may cause an abnormal termination (BSOD)(Incident 273295).

16. The move subdisk operation may fail for a striped volume that has been extended in such a way that the size of the volume is not a multiple of the stripe size (Incident 275207).  This issue is resolved by this hotfix so that the subdisk operation does not fail for striped volumes.

17. [DMP] Creating a basic partition on a system with DMP installed fails (Incident 302398).

18. Creating a primary partition on VM 3.1 fails when connecting remotely from VSFW 4.1 or VSFW 4.3 (Incident 315340).

   
HotFix 04 - Released June 23, 2004

ISSUES RESOLVED IN THIS HOTFIX
The issues section is divided into three topics:
A. General Issues
B. Cluster Issues
C. DMP Issues  

A. General Issues
1. If you run two vxassist rescan commands during the same time frame to rescan storage objects, the second vxassist rescan command will not complete. (Incident 102033). This issue is resolved by this hotfix so that the second vxassist command will complete successfully.  

2. Snapback may take an extended period of time to complete when there is very little I/O following a snapshot operation (Incident 102800).  

3. If you increase the size of a volume, select the Disk Regions tab on the right pane of the GUI, and attempt a drag and drop subdisk move involving the volume, the subdisk move may fail (Incident 102898). This issue is resolved by this hotfix so that performing a drag and drop subdisk move will complete successfully when there is adequate space.  

4. Even though the maximum size is selected when creating a volume, the resulting volume may be less than the maximum size (Incident 102898).  

5. If a snapback operation fails because the snap volume cannot be opened, a misleading snap record not found message is returned (Incident 102941). This issue is resolved by this hotfix so that the correct error message is returned.  

6. The Microsoft Perfmon function incorrectly displays duplicate objects for dynamic disk and logical disk counters (Incidents 102965, 123970). This hotfix resolves this issue so that Perfmon correctly displays dynamic disk and logical disk counters.  

7. Enhancement: The vxvol volinfo command now provides a list of all snapshots associated with a volume (Incident 102970). After this hotfix is applied, the vxvol volinfo command will provide a list of all the snapshots of a volume that exist.  When vxvol volinfo is used on a snapshot volume, it will provide the name of the original volume.  

8. Enhancement: The vxscrub utility is made available to remove unneeded registry entries resulting from adding and deleting volumes (Incident 103022). This hotfix installs the vxscrub utility. A Windows 2000 system  may not boot up when the system hive of the registry becomes too large. The exact size of the system hive that causes this problem depends on the configuration of the system. Repeatedly adding and deleting volumes or disk devices will increase the size of the system hive. For example, performing many off-host backups may add and delete many volumes and increase the size of the system hive. The increase  in size of the system hive can be seen in the growth of the system file (%SystemRoot%\system32\config\system). Running the vxscrub utility as a part of regular system maintenance will shrink the registry size and avoid this problem. Rebooting the system after running vxscrub is mandatory. With the -p option, the utility can uninstall and clean up unused volume, disk, and MountManager registry entries. It also purges the registry to physically shrink the size of the registry. Run vxscrub -p from the command line and then immediately reboot the system to perform this operation. Additional vxscrub options can be displayed by running vxscrub from the command line without any options.  

9. When a mirror resynchronization operation is aborted, an incorrect success alert is raised (Incidents 103131, 103302). After this hotfix is applied, when a resynch operation is aborted, a failure alert is raised.

10.Enhancement: Add a force option to the vxdg join command to forcibly join two dynamic disk groups when some volumes in the disk groups are being used by another application (Incident 103301). Joining two dynamic disk groups will fail when some of the volumes in the disk groups are being used by another application. After this hotfix is applied, the vxdg join command with the force option will complete the operation successfully even when some of the volumes are being used by other applications.  

11.Adding a mirror to a striped volume that was extended so that its size is no longer a multiple of the stripe size creates a plex configuration that will cause an I/O failure for both reads and writes involving the last stripe (Incident 107903).  

12.Removing disks that contain mirror volumes during synchronization of the mirrors may cause a fatal system error [BSOD] (Incident 130136).  

13.Enhancement: Provide support for the new IBM OEM partition type 0xFE (Incident 110051). After this hotfix is applied, the IBM OEM partition type 0xFE will be supported.  

14.Enhancement: The vxassist version command will display the build version that is installed (Incident 208332).  

15.The file system of a volume, particularly striped volumes, may be corrupted after completing a subdisk move with active I/O (Incident 209505).  

16.If the Windows locale setting is changed from English to a language that uses a comma as the decimal indicator (such as German) performing a numerical sort of the objects in the right pane of the GUI (table view) fails (Incident 209525).  

17.An incorrect error message is displayed when a Snap Start or Snap Back fails (Incidents 210421, 212300). This issue is resolved by this hotfix so that the correct error message will be displayed.  

18.In the Create Volume wizard, the buttons in the Browse for drive path dialog (New Folder, Rename, OK, Cancel, and Help) do not work properly (Incident 210772).  

19.When launching Help from a wizard, the Help window that appears does not allow any navigation (Incident 211030).  

20.Using manual disk selection to extend a mirrored volume with two or more DRL logs can cause the Volume Manager service (vxsvc) to crash (Incident 213392).

21.Using Perfmon to monitor performance on a server with a large number of dynamic disks and volumes can cause the Volume Manager service (vxsvc) to crash (213401).

B. Cluster Issues  
1.[MSCS] - When moving cluster resources from node 1 to node 2, node 2 may encounter a fatal system error (BSOD) (Incidents 107584, 206715). This hotfix resolves this issue so that the cluster resources can be moved to node 2 successfully.  

2.[MSCS] VERITAS Storage Migrator (VSM) fails to detect volumes of a dynamic disk group that are in a cluster (Incident 116655). After this hotfix is applied, VSM will detect these volumes successfully.  

3.[MSCS] Dynamic disk group resource information does not get updated in the Cluster Manager Administrator during a failover (Incident 210662).  

4.[MSCS] Creating a Volume Manager Disk Group resource using the MSCS Configure Application wizard may cause the Cluster Administrator GUI to close or stop functioning (Incident 211038).  

5.[MSCS] An abnormal termination of the MSCS service (clussvc) may cause the Volume Manager service (vxsvc) to have an access violation (Incident 12260).

6.[MSCS] A cluster disk group with a DCO volume cannot be added as a resource under Cluster Administrator (Incident 212272).  

C. DMP Issues  
1.[DMP] The system may hang during a system boot (Incidents 103795, 108567). The system may hang as a result of a deadlock in the main DMP driver. After this hotfix is applied, a deadlock will not occur.  

2.[DMP] While running Driver Verifier, the system may encounter a fatal system error (BSOD) (bugcheck 0xC9 and error code 0x220 or 0x224) (Incidents 104154, 104247).  

3.[DMP] With IBM/STK/LSI FAStT storage arrays, I/O errors may occur during array controller failover (Incident 107994).  

4.[DMP] During system reboot or while enabling/disabling a path, the system may encounter a fatal system error (BSOD) (bugcheck 0xCD) (Incident 108570).

5.[DMP] The Compaq DMP ASL (Array Support Library) causes a memory leak (Incident 108775).  

6.[DMP] After excluding a disk on a Compaq/HP array, its paths may incorrectly become paths to another disk (Incident 109197). This hotfix resolves this issue for the Compaq/HP array so that the paths to an excluded disk will not become paths to another disk.

7.[DMP] Enhancement: Add DMP support for the HP23xx HBA in the DMP registry (Incidents 114102, 117003). After this hotfix is applied, DMP will be able to support the HP23xx HBA.  

8.[DMP] Enhancement: Change the IBM and EMC tunable default values to make DMP more flexible (Incident 117070). After this hotfix is applied, the IBM and EMC tunable retry default value will be set to 3 retries and the timeout default value will be set to 30 seconds. Note: This hotfix will not apply these default values if any of the tunable values have been modified before installing the hotfix. The hotfix will preserve all the existing tunable values even if only one of the values has been modified.  

9.[DMP] If BCV LUNs on an EMC Symmetrix array are not in a Ready state, all paths on a BCV LUN display as not healthy with a red error (Incident 209318). The hotfix resolves the problem by having DMP implement the Test Unit Ready command only when the BCV LUN is in Ready state.  

10.[DMP] Disabling one DMP path may cause cluster resources to fail over (Incident 210713). This hotfix resolves this issue so that disabling one DMP path will not cause a fail over.

11.[DMP] After including a disk array, the array may be incorrectly brought under DMP control as a generic JBOD and not by the appropriate ASL (Array Support Library) (Incident 211539).  

12.[DMP with EMC CX-600 array] If the tunable parameter for testing paths (TestPathRetryCount) is set to 1 or more retries, DMP displays a path's status as healthy, even when the path is failed or disconnected (Incident 213559).  

--------------------------

HotFix 03d - Released December 29, 2003

This hotfix should be installed only on servers running Microsoft Windows 2000 Server, Advanced Server, or Datacenter Server operating systems. The hotfix requires Microsoft Windows 2000 Service Pack 3.

Note: This hotfix includes all the fixes from Volume Manager 3.1 HotFix 1, HotFix 2, and Service Pack 1. For information on these fixes, see the Readme files for the hotfixes and service pack on the Technical Services Web site.

This section is divided into two topics:

I. General Issues
II. Cluster and DMP issues

I. General Issues

1. When two vxassist rescans are run at the same time, perhaps by running two scripts simultaneously, one of the rescan commands will not complete and the CLI will hang (incident 130126). This hotfix resolves this issue so that both rescan commands complete and the CLI will not hang.  

2. The Microsoft Perfmon function does not display information about all dynamic disks and dynamic volumes (incident 130963). Volume Manager allows the Microsoft Perfmon function to be used to display information about dynamic disks and dynamic volumes. However, information about some of these resources was not being displayed. This hotfix resolves this issue so that Perfmon correctly displays dynamic disk and volume information.

3. During the import/deport of cluster disk groups, trap messages were not mapped. The message, "missing trap mapping," indicating that the trap messages  were missing was reported during the import/deport operation (incident 131228). This hotfix resolves this issue so that the trap messages will be mapped  during the import or deport of cluster disk groups and the messages will be displayed properly.

4. Some Volume Manager commands on a Volume Manager 3.1 server may fail when they are issued from a VERITAS Storage Foundation (tm) for Windows 4.0 client  that is connected to and managing that server (incident 133139). This hotfix resolves this issue so that Volume Manager commands issued from a Storage  Foundation for Windows 4.0 client will not fail on Volume Manager 3.1.

5. When performing a snapback from the snapshot volume (rather than the original volume), the snapshot is not synchronized properly (incident 133172). This  hotfix resolves this issue so that the snapshot is synchronized successfully.

6. Snapstart fails when there is no space left to create the DCO volume on the same disk as the original volume (incident 135354). This hotfix resolves this  issue so that snapstart will not fail in this situation.

7. When an encapsulation operation fails for a disk, the Volume Manager service may crash (incident 135360). After this hotfix is applied, the Volume Manager  service (VxSvc) will not crash in this situation.

8. After repeated snapshot/snapback operations, the snapshot may not be synchronized properly (incident 135427). This hotfix resolves this issue so that the  snapshot is synchronized successfully.

9. Fast Mirror Resync (FMR): If a disk that contains a mirror plex that was being created and was partially synchronized fails, the plex may not complete the  synchronization process properly when the disk is recovered (incident 136323). This situation applies to an ordinary mirror with FMR or to a snap plex  created during the snapstart operation. This hotfix resolves this issue so that the mirror plex is synchronized successfully.

10. After the installation of the Volume Manager 3.1 Service Pack 1, rules cannot be successfully created because, near the end of the Rule Manager wizard  sequence, the Next button appears dimmed (incident 136723). After this hotfix is applied, rules can successfully be created with the Rule Manager wizard.

11. The snapback operation may take an extended period of time to complete for some volumes (incident 136899). This issue is caused by a problem in the  snapstart operation. This hotfix resolves this issue for original volumes that did not have I/O during their snapstart operations. Original volumes that had  I/O during their snapstart operations may take an extended period of time to complete the resynchronization during snapback. This remaining issue will be  addressed in a future hotfix or service pack.
       
II. Cluster and DMP issues

1. Enhancement - MSCS, VCS: Provide a means to remove the limitation that a cluster disk group cannot be created on the same bus as the disk that contains  the cluster node's system or boot disk (incidents 110297, 130943). Users have requested that this limitation be removed in order to facilitate booting from a  SAN.

Caution: An array hardware that supports booting from a SAN must be possessed and great care must be taken while implementing this configuration. Refer to  Microsoft Knowledge Base article 305547. To remove the limitation referenced above, use the following command at the command line:

vxclus enable SANBOOT (This command changes the Microsoft registry.)

To reverse the command, use:

vxclus disable SANBOOT

This hotfix will enable these commands.

Tip: If booting from a SAN is preferred, it is also recommended that the page file be moved to a local disk.

2. MSCS: The cluster server erroneously interprets cluster disks as failed, deports them, and reimports them (incident 132305). A Microsoft Winsock bug  causes MSCS to interpret the disk group resource as failed. MSCS tries to recover the resource and reimport it. This hotfix resolves this issue with a  workaround (contained in VERITAS Enterprise Administrator (VEA) 274) so that MSCS will not interpret the disk group as failed.

3. MSCS: When MSCS calls the Volume Manager resource, an unneeded error, error code 234 (ERROR_MORE_DATA), is logged in cluster.log (incident 132633). This  hotfix removes the logging of this unnecessary error code.

4. MSCS DMP: When both paths of node 1 and the preferred path of node 2 fail, the cluster dynamic disk group does not fail over to the remaining path on node  2. This issue occurs after the installation of the Volume Manager 3.1 Service Pack 1 (incident 133352). This hotfix resolves this issue so that the failover  to the remaining path on node 2 is successful.

5. DMP: The system may hang during a system boot or while an include or exclude operation is being performed (incidents 134287, 134288). The system hang may  be the result of a deadlock in the core DMP driver. After this hotfix is applied, a deadlock will not occur.

--------------------------

Issues Resolved by Service Pack 1 - September 10, 2003

This section describes the issues that are resolved in this service pack. The issues are organized in the following categories:

I. General Issues
II. DMP Issues
III. DMP and Cluster Issues
IV. Cluster Issues
V. FlashSnap Issues
VI. Enhancements

Note: This service pack includes fixes for Volume Manager 3.1 HotFix 01 and HotFix 02.

I. General Issues

1. The error Could not allocate contiguous space or partition occurs when adding a mirror using 18-gigabyte (GB) hard disks (incidents 112605 and 121928).  Incident 121928 involves a volume that was encapsulated from a basic disk.

2. Mirrored and RAID-5 volumes with logs may become missing after import (incident 127944).

3. Access Violation after a Vxdg split command involving a missing disk (incident 127773).

4. After a silent install of Volume Manager, the program cannot be launched from Computer Management (incident 113260).

5. A client-only Volume Manager install or uninstall should not require a reboot (incident 114746).

6. The Volume Manager service terminates unexpectedly after a reboot (incident 113413). The Volume Manager for Windows service is also referred to as the  VERITAS Object Bus or vxob.

7. The Volume Manager service crashes, causing a script to terminate (incidents 121743, 123059).

8. Testing that involves disconnecting and reconnecting disks repeatedly can cause the Vxio driver to hang (incident 114163).

9. Not able to disable a CD-ROM drive from the Device Manager (incident 117989).

10. Disks may be missing after a disk group is imported, especially after a cluster failover, but the problem is not limited to clusters (incident 114187).  The problem was due to a diskset ID mismatch.

11. The vxio driver may cause the system to blue screen during a configuration change, such as creating a regular volume or a snapshot volume (incident  120370).

12. There should be a warning if the private region is not valid on a disk (incident 117353). The private region contains the Volume Manager database, which  occupies a specific area on each disk that is under the Volume Manager control.

13. Cannot upgrade a basic disk to a dynamic disk if the basic disk is bigger than a terabyte (incident 117555).

14. The system checks removable drive status for volume properties when it is not necessary (incident 114782).

15. A missing disk in an existing disk group. The problem was caused by a Microsoft bug that is described in the documentation for Microsoft HotFix 332023.  (incident 130019).

16. After a volume is moved to another disk, the size of the free space that was used by the volume is displayed, in the GUI, as a smaller number than it  should be (incident 124388).

17. Receive an invalid arguments error when trying to assign a drive letter to a migrated NT-4 stripe set on a basic disk (incident 124620).
       
II. DMP Issues

1. Hitachi Data Systems 9960 array: Failing over and back again repeatedly sometimes causes the system to hang (incident 115036).

2. IBM Shark array was recognized by DMP as being supported by IBM JBOD, but it should be supported by the IBM satellite driver (incident 114737). DMP  support for IBM Enterprise Storage Server (ESS) models E10/E20/F10/F20 shark arrays was added through a satellite driver in 2002, but it was not registered  in the Windows 2000 registry. As a result, IBM Shark arrays were seen as JBODs. Now, the workaround given in TechNote 253814 is no longer needed.

3. Path status is continuously changed from "Not Healthy" to "Healthy" while running a utility that accesses invalid sectors on disks (incident 119358).  Utilities, such as dmdiag or vxdiag, had this effect on DMP.

4. Sometimes, after installing Volume Manager with DMP and then rebooting, the DMP paths tab is not available in the GUI (incident 123424).

5. DMP should not be allowed to disable the last "healthy" path (incident 108126).

6. During failover, the path icon on the DMP path may continue to display as yellow (may be "unhealthy") or red ("unhealthy") even when the DMP path is in a  "healthy" state (incident 121615).

7. In a failure condition (like a malfunction on a path), DMP driver may hang in some cases (incidents 124604 and 120691).

III. DMP and Cluster Issues

1. DMP/MSCS/Cluster Server: After booting a system or enabling a second DMP path in a cluster environment, extra shadow disks show up in the GUI (incident  108144). The problem was described in TechNote 253382, but it has now been resolved with this service pack.

2. Cluster Server and DMP: After a Cluster Server cluster failover and failback with heavy writes, paths may appear to become Active/Active (instead of  Active/Passive) because of blinking light-emitting diodes (LEDs) (incident 108371). The paths did not actually become Active/Active. This problem has been  resolved with this service pack.

3. DMP/Cluster Server/MSCS: DMP should update the GUI to show current preferred path when a path failure has occurred (incident 121121). When a preferred  path fails, DMP changes the preferred path to an existing surviving path, but this change was not shown in the GUI. This problem has now been resolved.

4. DMP/Cluster Server/MSCS: Pulling out a path that owns a cluster resource when there is I/O on the cluster resource can cause I/O failure (incident  119933).

IV. Cluster Issues

1. MSCS: Dynamic quorum arbitration timing issue may cause arbitration failure (incident 114162).

2. MSCS: A missing disk may appear after a move group command (incident 118415).

3. MSCS: Security vulnerability when using cluster disk groups with Volume Manager 3.1 and MSCS (incident 123456). This vulnerability is described in detail  in the VERITAS Technical Support TechNote 260302. VERITAS Technical Support also provided a patch to remedy the problem. Now, with this service pack, the  problem has been resolved permanently.

4. Cluster Server: After a power shutdown on a cluster node, the cluster does not failover to the other node (incident 115803). The problem had to do with  mounting the volumes on the second node.

5. When the MSCS service hangs on a node, VMDG resources are not able to failover to the other nodes because the SCSI reservation thread is still running on  the first node (incident 116326).

6. In an MSCS cluster environment, sometimes, a refresh or rescan operation will hang the Volume Manager service (incident 127874).

7. MSCS: In a two-disk mirrored volume split over two arrays, when one array is turned off, the mirror does not remain online (incident 106187).

8. Cluster Server: Cluster disk group may not come online when doing a force import because the disks were offline (incident 113525).

V. FlashSnap Issues

1. If after doing a snapstart, if the original volume is extended, and then a snapshot is done, the mirrors do not match after the snapback. This action is  no longer allowed (incident 126657). Now, a snapshot operation cannot be done until the mirrors are completely synchronized.

2. Configuration problem causing the Volume Manager service to have an access violation when creating a snapshot from an existing mirror (incident 123433).

3. Error messages on some snap commands needed to be either improved or made more consistent on the GUI and CLI versions (incidents 130337 and 130338).

VI. Enhancements

1. Enhancement - Cluster Server/MSCS: Provide a means to remove the limitation that a cluster disk group cannot be created on the same bus as the disk that  contains the cluster node's system or boot disk. (Incident 110297). Users have requested that this limitation be removed so as to facilitate booting from a  Storage Area Network (SAN).

Caution: An array hardware that supports booting from a SAN must be present and great care should be taken in implementing this configuration. Refer to  Microsoft Windows Server Knowledge Base article 305547. To remove the limitation referenced above, use the following command at the command line:

vxclus enable SANBOOT (This command changes the Microsoft Registry.)


To reverse the command, use:

vxclus disable SANBOOT

This hotfix will enable these commands. Tip: If you want to boot from a SAN, it is also recommended that your page file be moved to a local disk.

2. Enhancement: Add percentage completion milestones (25%, 50%, 75%) when a mirror is resynchronizing (incident 93472).

3. Enhancement: The -f force option was added for vxdg split. (Incident 125514).

4. Enhancement: If a volume is resized, but the file system fails to resize as well because of a malfunction, there is no way to resize the volume, so now a  command has been added to resize a file system (incident 117332).

vxvol growfs <VolumeName|DriveLetter> [<length>]

Where:

VolumeName: Name of volume. For example, \Device\HarddiskDmVolumes\DynamicGroup\Volume1

DriveLetter: Drive letter of the volume, such as C

Length: Size in megabytes

--------------------------

Issues Resolved by HotFix 02a - July 22, 2003

1. A snapback operation cannot be done if the snapshot volume is in use or if the original volume is in use and the user has selected to resynchronize the  original volume with the snapshot (incident 109968).

In a snapback operation, the detached snapshot mirror is reattached to its original volume. The user can choose to resynchronize the reattached mirror with  the data of the original volume (the usual situation) or the original volume can be resynchronized with the data of the snapshot. In any case, if the  snapshot volume is in use, the snapback command will fail. If the original volume is in use and the user has chosen to resynchronize the original volume with  the snapshot, the snapback command will fail as well. Because there are times when the operating system indicates that a volume is in use but it is not  actually being used by an application, a force (-f) parameter has been added to the vxassist snapback command to allow the user to force the command in the  situations described above.

Note: The snapback command will not fail if the original volume is in use and the detached snapshot mirror is resynchronized with the original volume.

For this hotfix, the force option is included only in the Command Line Interface (CLI) version of the snapback command. With HotFix 3, the force option will  be included in the GUI snapback command.

Caution: Do not force the command unless it is certain that the volume is not actually in use by an application. Data corruption can result, especially when  the original volume is in use and the snapshot volume is selected for synchronizing the mirror.

--------------------------


Issues Resolved by HotFix 01f - March 12, 2003
This section is divided into two topics:

I. General Issues
II. Cluster and Dynamic Multipathing (DMP) issues


I. General Issues

1. Dell OpenManage Array Manager: Simple Network Management Protocol (SNMP) alerts sent from Array Manager to a Management Console (such as HP OpenView) are  disabled (incident 107100). After this hotfix is installed, SNMP alerts can be sent from Array Manager version 3.4 or later.

2. An Access Violation (AV) in the command line interface may occur when the last remaining basic disk on a system is either removed or upgraded to dynamic  (incident 108019). After this hotfix is installed, an AV will not occur in this situation.

3. The vxassist -f unassign command to force the unassignment of a drive letter fails and returns the following message: Failed to complete operation  e509000z:  The volume is currently in use. Please close applications and try again (incident 108896). After this hotfix is installed, the command will work.  

4. When online monitoring is initiated, occasionally, the vxsvc process causes the CPU utilization to increase by 25-50% (incident 110343). This CPU  utilization increase will not occur after this hotfix is installed.

5. When a basic disk is physically added to a system with no existing basic disks, the basic disk group does not appear in the GUI until the system is  rescanned (although the disk displays in the GUI) and the command line vxdisk diskinfo does not display information about the disk (incident 110641). After  this hotfix is installed, the system will recognize the basic disk group immediately and display it in the GUI.

6. This hotfix adds a force (-f) option to the vxassist delete command (incident 110656). The command syntax is:

vxassist -f delete <VolumeName|DriveLetter>  

7. The Add Mirror and Snap Start commands fail on volumes larger than 1 terabyte (incident 111747). After the hotfix is installed, these commands will work  for volumes larger than 1 terabyte.

8. On a fault tolerant server with multiple I/O enclosures (such as a Stratus server) and disk groups containing disks located in more than one enclosure,  the following behavior may occur as a part of the failover and recovery process when one storage enclosure is failed over to another enclosure under heavy  I/O:

a. A blue screen may occur (incident 112426, Stratus case: winv_vrts_119).

b. System deadlock may occur during multiple mirror resynchronization (incident 112930, Stratus case: winv_vrts_127).

c. System deadlock may occur during resynchronization of multiple Fast Resync-enabled mirrors (and their attached disk change object (DCO) volumes) (incident  112931, Stratus case: winv_vrts_128). This hotfix resolves these issues.

9. FastResync of a mirror occasionally takes longer than expected (incident 112509). This issue is resolved by this hotfix.

10. This hotfix will correct occasional problems that may have occurred because of the way the Volume Manager driver handled plug-and-play requests (incident  112580). As a result of this change, a prompt to reboot may appear for the second time when this hotfix is installed.

11. A third party application using the Volume Manager API to monitor event notifications will not receive them after a client is disconnected from the  server and then reconnected to it (incident 113066). This will not affect the Volume Manager command line or GUI interfaces. After this hotfix is installed,  the third party program will receive notifications after reconnection.

II. Cluster and DMP issues

1. VERITAS Cluster Server (tm): A Cluster Server-controlled disk group resource can be deported even if the disk group is online (incident 108621). After  this hotfix is installed, manually deporting a disk group resource, under Cluster Server control, will not be possible.

2. DMP: Vxio disconnect errors may occur and paths may be marked as "Not Healthy" during a bin file change to update the Fibre Adapter in an EMC Symmetrix  storage array (incident 108863). All I/O requests will be "frozen" for a period of approximately 5 to 10 seconds during the bin file change. During this  period, if DMP attempts to monitor the health of a path, the attempt will fail and the path will be marked as "Not Healthy." Although the built-in retry  mechanism works correctly, the interval may still be too short, causing the path to be incorrectly marked "Not Healthy." If all available paths are marked  unhealthy, any further I/O attempts will fail and vxio errors will appear in the event log.

Installing this hotfix creates three new Windows registry names and values so that the retry count and sleep time can be adjusted before a bin file change is  made on a Symmetrix array. This hotfix also installs a command line utility, vxemctune.exe, in the Volume Manager 3.1 main directory. The typical path for  this directory is C:\Program Files\VERITAS\Volume Manager 3.1. Vxemctune can be used to trigger the DMP driver to reload the tunable parameters from the  registry. By adjusting these values, the situation described above, which occurs while the I/O is frozen, can be avoided. The default values for the new  Registry entries are:

a. TestPathDelay = 200 milliseconds

b. TestPathRetryCount = 0

c. TestPathTimeout = 2 seconds

Caution: Increasing the retry count will increase the time to fail over when a path or port becomes disabled or is pulled.

For information on how to use Regedit.exe or Regedt32.exe, consult .

Warning: Incorrect use of the Windows registry editor may prevent the operating system from functioning properly. Great care should be taken when making  changes to a Windows registry. Registry modifications should only be carried-out by persons experienced in the use of the registry editor application. It is  recommended that a complete backup of the registry and workstation be made prior to making any registry changes.

After installing this hotfix, follow these steps to edit the Windows registry:

a. Change the values in the registry using Regedit.exe or Regedt32.exe. The ranges for the in-memory tunable minimum and maximum values are shown below:

         HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\VxDMPEmc\Tunable

         Value name = TestPathDelay
         Data type = DWORD
         Value = in milliseconds, Min: 1  Max: 10000

    Value name = TestPathRetryCount
    Data type = DWORD
    Value = number of retries, Min: 0  Max: 50

    Value name = TestPathTimeout
    Data type = DWORD
    Value = in seconds, Min: 2  Max: 25

b. To display the current values, from the command window, run vxemctune -d

c. From the command window, run vxemctune -r for the new tunable values to take effect immediately (without rebooting). To verify the new in-memory values,  use vxemctune -d

Note: After the bin file change is completed, it is recommended that the registry settings be returned to the default settings.

3. DMP: High CPU utilization will occasionally occur while the user is including an array under DMP control (incident 109070). This situation will not occur  after this hotfix is installed.

4. Cluster Server or Microsoft Cluster Server (MSCS): The Reactivate Disk command may be unavailable to bring the disks of a degraded mirrored volume online  (incident 109549).

This occurs in a cluster with a mirrored volume in this situation:

a. The primary node fails,

b. a force import is done on the secondary node to bring the viable portion of the disk group online,

c. the remaining disks in the disk group become available and display on the secondary node as foreign disks,

d. the Merge Foreign Disk command is used to bring the repaired disks back into the disk group on the secondary node. The mirrored volume is in a degraded  state but the Reactivate Disk command is not available on the menu to resynchronize the disks.

After this hotfix is installed, the Reactive Disk command will be available on the menu so that the degraded mirrored volume can be reactivated in the  situation described above.




Legacy ID



264655


Article URL http://www.symantec.com/docs/TECH29569


Terms of use for this information are found in Legal Notices