We need to migrate iSCSI LUNs from FC disks on one controller to SATA disks on another controller. I plan to use snapmirror. There is one thing special about this case: there is an MS Cluster. We need to migrate the quorum LUN as well. Anybody know about the procedure?
When it comes to the quorum/witness disk (Windows 2008 and later called the disk a witness disk. Windows 2003 and earlier calls the disk a quorum disk), my advice would be to re-create the witness disk in the new location instead of migrating it. The quorum/witness disk itself should contain no data other than the cluster configuration information. Each node in the cluster will also contain a copy of this data. While there is no quorum/witness disk, the fault tolerance of the cluster is reduced, but the quorum/witness disk would be replaced before the cluster is put back into production use, so the risk is well mitigates.
The process I would use is... 1. Change the cluster to not use a quorum/witness disk and remove the quorum/witness disk from the cluster group. 2. Migrate all other LUNs as planned.
3. After the migration is completed, add a new quorum/witness disk to the cluster group and change the cluster type to use the quorum/witness disk
If you are not using SnapDrive, I *highly* recommend using it. SnapDrive will add LUNs to all nodes within the cluster and assign the specified drive letter -- without taking the cluster nodes offline. SnapDrive also allows administrators to grow LUNs without stopping clustered services.
We MS-Clustering and NetApp storage for many of our systems. If you have any additional questions, I'd be glad to help where I can.
We do this currently with snapmirror and vmware srm between a primary site and dr site, but in your case it's much simpler.
We have snapmirror replicating the nfs volume where the vm sits, and also the volumes/lun where the iscsi databases and logs sit.
When we bring it up on the other side (there is a change of ip addresses) but I basically vol clone the snapmirror destination and repoint the host to the new san and it all comes up for testing.
In your case. Just snapmirror the volumes, once the snapmirror is at idle, take the cluster offline. Quiesce and then break the snapmirror relationship. Unmap the old luns and remap the new luns on the new controller. Make sure you have an iscsi session to the new controller. Check Disk Management to make sure you see the disks appearing (the disks will probably appear as reserved until the cluster starts) start the cluster up again and it will pick up where it left off.