Andrew's Blog

This post will document the procedure to fail-over to a replicant volume on a second EqualLogic Group, when using XenServer.

The recovery process is in a separate entry

XenCenter

• Stop the Virtual Machine
• Detach the Storage Repository (only detatch, do not destroy or forget). Answer “Yes” to the confirmation dialog, and wait until the detach process is complete (check the Storage Repositories logs).

If this is a planed failover, and all hardware is online and working, you will want to perform a final replication now, to copy changes since the last replication to the D.R. group.  If this is unplanned, and the production hardware is offline, then you will lose any data since the last replication.

D.R. EqualLogic Group Manager (replication target)

• Promote the Replica Volume: Click on the Replication tab, and select the production group under “Replication Partners”, then “Inbound Replicas”, and select the replica volume you wish to restore. Click on “Promote to volume”, this will promote the last replica to a volume.  To use another replica, you need to clone the replica to a new volume, which is outside this procedure.
• Click “Yes” to the confirmation dialog, which will prompt you that inbound replication will pause during the promotion.
• Make sure both the boxes (Set volume online, and Retain target name of the original volume) are ticked, and click Next.
• Set the iSCSI access (by IP address, use asterisks for wildcards), and access type (Set read-write, and Allow simultaneous connections is ticked), and click Next.
• Click Finish.

XenCenter

• Re-attach the Storage Repository using the D.R. Group address as the target host, “Discover IQNs” selecting the correct IQN from the list, then “Discover LUNs”, there should only be one.
• Click Finish.
• Click “Yes” to the confirmation dialog.
• Start the Virtual Machine.

Production EqualLogic Group Manager (when possible, i.e., when it is back online)

• Select the volume you have promoted to D.R.
• Select under Replication Failback, “Demote to replica set”.
• There will be a confirmation dialog, making sure you have completed the promotion steps above, select “Yes” to this dialog.
• If you recover from a disaster (i.e., failure of production, and it was an unexpected shut), it will suggest that there are changes to the volume that have not been replicated, we will select “No” here, and lose the changes, as we have already started using the replica, and are recovering from an unexpected shut, and know we were going to lose data.
• This process will move the volume to the inbound replicas on the production group.