HP StorageWorks SAN Virtualization Services Platform 3.0.
Legal and notice information © Copyright 2008-2010 Hewlett-Packard Development Company, L.P.
Description HP StorageWorks SAN Virtualization Services Platform (SVSP) is a hardware and software solution that enhances the storage capabilities of approved storage arrays. Using Virtualization Service Manager (VSM) software, redundant VSM servers and Data Path Modules (DPMs) map application server I/O to locations on storage systems. Versions HP SAN Virtualization Services Platform 3.0.4, which includes: • Virtualization Services Manager software version 1.2.583 supported by LSI HBA driver version 9.1.
Enhancements This release of the SVSP product introduces the following enhancements: • Pre-installations of the VSM server operating system will use Microsoft Windows Server 2008 R2 instead of Windows Server 2008 SP2. • The VSM server hardware changes to an HP ProLiant DL360 G7. Information on the features of this server can be obtained from: http://www.hp.
• The SNMP modify community trap name only changed value for trap generation, not the MIB queries. • DPM failure occurred when a virtual disk was deleted and smaller virtual disk was created with same LUN. • DPM failure occurred after a synchronous mirror virtual disk was deleted due to delayed synchronous mirror cleanup. • When the LUN from an HP XP array was added to the storage pool, another LUN appeared. • VSS over VMware issues: • Hardware provider supports HSV100 device types.
At the location HKEY_LOCAL_MACHINE\System\CurrentControlSet\Control\ CrashControl\CrashDumpEnabled, set the registry key to 0x1 (instead of 0x2). Host operating systems and multipath drivers See http://www.hp.com/storage/spock for a listing of supported operating systems and required multipath drivers. Site registration is required. Also see “Operating system issues” on page 12 for issues with supported operating systems. To obtain the HP MPIO Full Feature DSM and DSM Manager: 1. Go to http://www.hp.
SVSP system capabilities and limitations • • • • Maximum capacity per system: 2 PB Maximum number of VSMs per system: 2 Maximum number of DPMs per system: 4 DPM groups with maximum of 2 DPMs per group Maximum number of array ports: 128 For example: • 32 arrays with 4–ports each • 16 arrays with 8–ports each • Any combination totalling up to 128 ports • Maximum number of servers (with 2 HBA ports/server) per DPM pair: 512 (2048 with 4 DPM pairs) • Maximum number of tasks per domain (total snapclones, async
• Maximum number of synchronous mirror virtual disks per DPM group: 128 (512 with 4 DPM groups) • Maximum number of total active asynchronous remote copies, snapclones, and concurrent migration jobs per SVSP system: 512 • Maximum number of asynchronous mirror tasks (remote copies) per source virtual disk: 3 • Maximum number of cross-connected systems or domains for management over IP: 4 SVSP back-end capabilities and limitations • • • • • • • • Back-end array capacity of the SVSP system: 2 PB Total back-e
• The Windows system disk of the VSM server and the VSM setup virtual disk must not be converted into a dynamic disk. • You should not upgrade (or accept automatic updates for) operating-system software components such as iSCSI initiators and QLogic HBA drivers. Only explicit VSM components are approved for use. Operating system security updates are supported.
6. The snapshot is presented to a host. 7. The second I/O stream is initiated to the snapshot. If you decide to delete the virtual disk, you should remove or halt all the operations that were performed from the time the virtual disk was created, by working backwards. In this case the process would be: 1. Halt the second I/O stream to the snapshot (initiated in step 7 above). 2. Unpresent the snapshot to its host (presented in step 6 above). 3. Delete the snapshot (created in step 5 above). 4.
• When performing remote asynchronous mirror split and merge operations, after the split, and before the merge operation, you need to perform a VSM failover in the destination domain. Afterwards, you can perform the merge operation. Otherwise, the operation fails. • Do not set the LUN range for a UDH when only one LUN is assigned to the UDH (for example, when the same LUN is assigned to both the first and last LU).
DPM group. The newly added DPM will not join the old DPM group. To recover from this issue, perform the following steps: 1. Manually delete the old (partial) DPM group. 2. Verify the front-end HBA list is the same for both DPMs, their rank is the opposite, and they have the same WWNN. 3. Manually create a new DPM group consisting of the two DPMs. 4. Verify the correct exposure of the virtual disks in both DPMs. 5.
disk results in data loss on imported LUNs or having the setup volumes erased. Do not use Windows to format LUNs or volumes when you receive this message. HP is working with Microsoft to resolve the issue. Miscellaneous Windows issues The following are currently issues with Microsoft Windows: • For a Windows HyperV environment, the only supported guest operating systems are the Windows operating systems—Windows Server 2003 SP2, Windows Server 2008 SP1/SP2, and Windows Server 2008 R2.
VSM issues Thin virtual disks must have read/write permission and cluster flag Do not create thin virtual disks with: • Only read-only permissions (in other words, without any read-write permissions) • Read-write permissions to both VSMs without the cluster flag. If either operation is performed, the operation is blocked, but the virtual disk is created in a Delete state. The thin virtual disk is deleted only upon restart or failover of the active VSM.
Manual failover of loaded system with VSM GUI causes a non-expected system state A manual VSM failover from the VSM GUI with 512 asynchronous mirror data movers causes a non-expected system state (NESS). The scenario is as follows: 1. VSM1 is active. 2. A failover is initiated from VSM1 to VSM2. 3. VSM2 becomes active, but experiences a NESS after a period of time and reboots. 4. VSM1 becomes active again and VSM2 becomes passive after it boots up. 5.
Front-end hosts not able to work with objects When a back-end array or LUN in the DPM back-end disappears or is masked out for a long time and then returns, the front-end hosts are not able to work with the objects (virtual disks, snapshots, and so on) that use the disappearing device (array or LUN). The workaround is that you need to execute an HBA disable/enable cycle or reboot the front-end host when the device returns.
5. Select the user account to which you are logged in (usually the manage account). 6. Change the base preference to base 2 or base 10, as desired. 7. Click the Modify User button, and click OK on the Success pop-up screen. From the CLI: 1. Log in using telnet or ssh. 2. Enter the set cli-parameters base command, where is 2 or 10. Error with Continuous Access license capacity when replicated differently The SVSP 3.0.
back-end logical unit in step 3. The reason is that the disk may contain some SVSP signatures left over from step 1, which may be recognized in step 3. ForceActive with VSM server in unknown state If the ForceActive utility is running and the VSM server enters an unknown state, stop the VSM GUI, stop the ForceActive utility, restart the VSM, and then run the utility again. Deletion of last DPM in a DPM group Do not delete the last DPM in a DPM group (even though it is enabled for recovery purposes).
2. Log in to the Windows server, right-click on the My Computer icon, and select Manage. 3. Click Device Manager. In the right pane, right-click Disk Drives, and select Scan for hardware changes. 4. New LUNs will appear under Disk Drives. New back-end LUNs require rescan To discover new back-end LUNs, perform a rescan from Computer Management on the VSM server. Do not use the Rescan devices option from the Windows Management GUI. Use this general procedure when assigning a new EVA LUN: 1.
Asynchronous mirrors and jobs do not fail when the source site is unavailable Asynchronous mirror groups and jobs on the destination system do not fail when the source system is down. It is expected VSM behavior that a job stays in a Normal state when the source site is down. Accessing VSM may require Java Web Services cleanup It may be necessary that Java Web Services be run (using Start > Run > Javaws) in order to access VSM through Internet Explorer or Mozilla. Delete all entries in the windows.
Potential for I/O loss with synchronous mirrors due to active DPM failure I/O loss could occur with synchronous mirroring using “Sync Always” groups (not with “Continue on Fail” groups) when an active DPM is completely disconnected from the domain. Ensure that members of the synchronous mirror group are failed over to the other DPM before disconnecting an active DPM from the SAN.
Change active switch command does not work with Sun Solaris hosts The change active switch command is an operation that you can initiate on the VSM to force a virtual disk to become active on the other DPM. This is typically used in maintenance operations to proactively failover virtual disks to one of the DPMs so that the other DPM can be serviced.
DPM NTP configuration does not work with Windows 2000 domain controllers The DPM only supports NTP servers that support network time protocol query (ntpq) and the full RFC-1305 specification. In particular, Windows 2000 only implements 'Simple NTP' and is not supported by the DPM as an NTP server.
• The personality or OS type of a user-defined host (UDH) cannot be changed in the VSM while virtual disks are exposed. All virtual disk permissions must be removed before changing the personality or OS type of a UDH. Not doing so could lead to unexpected results, a crash, or a reboot of the DPMs. • Persistent reservations and SCSI-2 reserve-release do not work correctly in some cases: • Persistent reserve completes successfully on passive SCSI-2 on a reserved virtual disk.
immediately marked by the VSM as failed. A message in the VSM will state Snapclone Tasks XXX has changed status to FAILED, reason: Host doesn't exist. The snapclone group needs to be deleted and created again with a valid host name. Error received with GetDpmInfo command when using –SyncGroup argument An error is received when using the GetDpmInfo command with the -SyncGroup argument. Do not use the -SyncGroup argument. The -VirtualDisk , -Snapshot, and -VDGroup arguments works correctly with this command.
The ResumeLocalAsyncMirrorTask command does not work in HP-UX the first time The ResumeLocalAsyncMirrorTask command may not work until the second time it is executed in HP-UX. Therefore, either try rerunning the command, or run the ResumeAllLocalAsyncMirrorTasks command instead. HP is working to resolve this issue.
VSS hardware provider issues The following are known restrictions and limitations for the Volume Shadow Copy Service (VSS) hardware provider: • The hardware provider cannot create a snapshot on a synchronous mirror virtual disk. • The hardware provider always creates user PiTs instead of standard PiTs on asynchronous mirror virtual disks. • When deleting a PiT that is in a "precommitted" state, a This will delete the snapshot set warning appears.
Under storage, select Storage Software and then select HP StorageWorks SAN Virtualization Services Platform under Storage Virtualization Software. For product white papers, browse to http://www.hp.com/go/svsp and click the White papers link.
