3

We are planning to expand an existing AAG setup which is a 2 node setup in DC and will be adding a 3rd node in San Jose(different subnet). The current setup is configured with Node and File Share majority. I was going through a Microsoft Whitepaper for AAG setups and want to have a clear idea about this comment which was mentioned

"Because the third node is in a separate data center and no longer has a vote, you should change this quorum model to Node and File Share Majority. Create a file share in the primary data center on a server node that will not be participating in the WSFC"

The current file share witness can actually be seen in the cluster core resources. Does the above sentence now mean that the file share shouldn't be a part of it? Please clarify

1

I'd need to see the article where that quote is from to understand the context, but the file share will still be part of the cluster and will still be configured as the witness.

Votes: Node1=Yes, Node2=Yes, Node3=No, File share=Yes

If you ever failover to the other data center, you'll need to create a quorum there. Maybe you'd spin up another node and file share when you failover. Those would be created in the San Jose data center. Once those are in place, add the VM to the cluster and then reconfigure the cluster to use the new file share in the San Jose DC for the witness. After failover, disable the votes for the DC data center. You don't want those resources deciding if you have quorum when there's a network glitch between the two data centers.

0

That is correct. The quorum resource (disk, share, etc) will no longer be listed as a "clustered resource", but it will be under Advanced Quorum settings.

Right-click on the Cluster name in Failover Cluster Manager to manage/change the quorum settings.

For a local datacenter cluster, all nodes/quorum should be local. If you are clustering across datacenters, use a 3rd location for quorum like Azure blob storage. That way, if your primary datacenter goes down, both quorum (Azure) and San Jose node can vote the SJ node the new owner.

You get into all sorts of issues if you expect automatic failover to SJ when it can't communicate to DC to reach both the other nodes and the quorum. If you are setting up SJ as a replica with manual failover - it will work well enough to add a fileshare quorum/witness to the DC cluster.

  • For now, I am planning for a manual fail over between DC and SJ. So my thought was that the existing file share witness which was already created for the AAG setup in DC – user2923332 Feb 21 '17 at 17:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.