I'm trying to navigate the following situation:
I have two servers running Windows 2008 and Exchange 2007 configured as a CCR (Continuous Cluster Replication) pair. All SAN storage for node 1 is on a DMX4, all SAN storage for node 2 is on a Clariion. The quorum is a remote file share.
I need to migrate both of these nodes to new storage (VMAX, yay) and I'd prefer to use PowerPath ME with host copy.
I get that if this were a "normal" windows 2008 failover cluster, we'd fail services to one node, run PPME, bring the cluster resource groups offline, commit, and then bring the cluster resources online again.
would that be the case here, even though the storage is separate/independent on both nodes? If yes, I'd anticipate a) migrate node 2 (take an outage), and then b) migrate node 1 (take another outage).
The reason for bringing the resources offline/online in a 2008 cluster is to migrate the SCSI3 persistent reservation placed on the disk by the cluster. However, it doesn't sound like you have this. Sounds more like two stand alone servers. If that is the case and the disks are not in Failover Cluster Manager then you would perform the migration using the process for a stand alone server which means no outage and no offline/online.
If the disks are in Failover Cluster Manager and act as two single node clusters then the offline/online would be necessary to move the reservation to the new storage.
Hope this helps
In Microsoft Exchange CCR Cluster, there are two sets of disks. One set presented to the Active Node and the other set of disks presented to the Passive Node which get the data replicated continuously from the set of disks from Active Node. Cluster service still will be running. So The passive node may have to be powered off as in all cluster migrations using PPME (though not tried this). Once PPME migration is used to migrate all the disks, you would have to set their relationship on the passive side with the new disks on VMAX and start the replication.
Hi! The migrations should be treated independently of each other using the migration process for a stand alone, rather than the cluster method/procedure as mentioned in the other reply to your question.