Managing Shared Disks

Making changes to twin-tailed volume groups

This section describes how to make a logical volume configuration change to your virtual shared disks, such as extending a volume group, changing the size of your virtual shared disks, or adding a new physical disk, and coordinate the timestamps so the IBM Recoverable Virtual Shared Disk subsystem does not incur unnecessary overhead during recovery processing. If you make a change to your virtual shared disks on twin-tailed volume groups that are managed by the IBM Recoverable Virtual Shared Disk subsystem without resetting the timestamps, any later recovery processing will cause the volume groups to be exported and then imported to the secondary nodes. This will occur even if you have made the updates to the virtual shared disks at both nodes.

Timestamps are maintained on the primary node, the secondary node, and the virtual shared disk itself.

Making a change to only one side of a twin-tailed volume group

This procedure describes how to make a change to one side (side A) of a twin-tailed volume group without varying off the volume group on that side.

1. Make the change.
2. Issue

   vsdvgts volume_group_name
   

   on side A. This reads the volume group timestamp from the disk and saves the value. The next time the IBM Recoverable Virtual Shared Disk recovery scripts vary on this volume group on this node, the timestamps will be the same and the overhead of importing the volume group will be avoided. If a failover occurs, the timestamps will be different on side B, so the volume group will be exported and then imported.

3. If virtual shared disks have been added, use the virtual shared disk Perspective to configure and start the virtual shared disks on all the nodes that need to be aware of this virtual shared disk, or issue the appropriate virtual shared disk commands.
4. Start IBM Recoverable Virtual Shared Disk if it is not running, using the ha_vsd reset command.

Making a change to both sides of a twin-tailed volume group

This procedure describes how to make a change to both sides (sides A and B) of a twin-tailed volume group without varying off the volume group on side A.

1. Make the change, issuing all the appropriate AIX commands on side B to reflect the changes made on side A (the volume group on side A is varied online). If a disk is being added and does not have its PVID defined on the secondary node, Recoverable Virtual Shared Disks will not be able to recover the volume on the secondary node if a failover should occur. You can use the chdev command to get a PVID assigned. Ensure that the PVIDs on the primary and secondary nodes match.
2. Issue

   vsdvgts -a volume_group_name
   

   on side A. This reads the volume group timestamp from the disk and saves the value for both sides A and B, since the same changes have been made on both. The next time the IBM Recoverable Virtual Shared Disk recovery scripts vary on this volume group on either node, the timestamps will be the same and the overhead of importing the volume group will be avoided.

3. If virtual shared disks have been added, use the virtual shared disk Perspective to configure and start the virtual shared disks on all the nodes that need to be aware of this virtual shared disk, or issue the appropriate virtual shared disk commands.
4. Start IBM Recoverable Virtual Shared Disk if it is not running, using the ha_vsd reset command.

Making a change to both sides of a twin-tailed volume group with the volume group varied off

This procedure describes how to make a change to both sides (sides A and B) of a twin-tailed volume group where both sides are varied off and you intend to explicitly export and import the volume group.

1. Make the change on side A. On side B, export the volume group and import it (issue exportvg and importvg). If a disk is being added and does not have its PVID defined on the secondary node, Recoverable Virtual Shared Disks will not be able to recover the volume on the secondary node if a failover should occur. You can use the chdev command to get a PVID assigned. Ensure that the PVIDs on the primary and secondary nodes match.
2. Issue

   vsdvgts -a volume_group_name
   

   on either side. This reads the volume group timestamp from the disk and saves the value for both sides A and B, since the changes made on A have been exported to B. The next time the IBM Recoverable Virtual Shared Disk recovery scripts vary on this volume group on either node, the timestamps will be the same and the overhead of importing the volume group will be avoided.

3. If virtual shared disks have been added, use the virtual shared disk Perspective to configure and start the virtual shared disks on all the nodes that need to be aware of this virtual shared disk, or issue the appropriate virtual shared disk commands.
4. Start IBM Recoverable Virtual Shared Disk if it is not running, using the ha_vsd reset command.