Currently I have a two node FCI using a Node and Disk Majority quorum, both nodes are physical servers. An AlwaysOn AG was configured within that FCI. I have two stand-alone nodes, both VMs, that can hold the AOAG but not any of the FCI resources because of course they don't share the physical disks being presented to the two physical FCI nodes. One of the two standalone VM nodes is located in a remote DR location. I need to physically move the original two FCI nodes and the shared storage those two share to the same remote DR location.
Now obviously the cluster has 4 nodes and a quorum disk for a total of 5 votes. When the two physical nodes get shut off to get trucked to the DR location, the quorum disk is going along with them so the cluster will shut down completely as 3 of 5 votes are now offline. I see three possible solutions to prevent the AOAG from staying down for the duration of the physical move:
After failing over the AOAG to the remote DR site node:
1.) Shut down the cluster. Force quorum on the DR site's node. Once the physical servers are in place at the DR site, shut down the cluster and start it back up normally since all 5 votes are back online. This of course causes two outages (in addition to the outage caused by the initial AOAG failover to the remote DR node), one to shut down the cluster and force quorum, and one more outage to shut the cluster down and return it to normal.
2.) Change the current cluster to Node and File Share Majority with the file share pointing to a share that will be available while the two physical nodes and their shared storage are in transit. This results in 3 of 5 votes (2 nodes + share) staying alive and thus the cluster stays up and never goes down. Once the two physical nodes are back online, can either return quorum config back to Node and Disk Majority or simply leave it Node and File Share Majority, old quorum disk can now be removed if the latter.
3.) Change the NodeWeight value on the remote DR node to be 3 which increases the total number of votes to 7. When the physical nodes and quorum disk are taken offline, 4 of 7 votes (node 1 = 1 vote, node 2 = 3 votes) stay alive and the cluster stays up and never goes down. When the two physical servers and quorum disk are back online, return all nodes back to 1 vote each.
Any recommendation as to which method would be the best? I am leaning towards #2 or #3 as it would seem they would require less outages. Or maybe a 4th option that would be a combination of 2 and 3, like replacing the quorum disk with a file share and changing NodeWeight to 2 on one of the two VMs. I would like to get rid of the quorum disk as it is really only shared between the two physical nodes whereas the file share would be shared between all 4 nodes.
Any insight would be appreciated. Thanks!