WSFC File Share Witness

availability-groupsclustering

I've got a multi-site 2008 R2 WSFC with a file share witness. I noticed that the possible owners for the FSW is every node in the cluster. I thought the concept of a FSW was that no node of the cluster owns it. Its a witness that sits outside of the cluster. So what is the purpose of having "owners" of a FSW resource and what would be the effect if I limited the possible owners to just those nodes at the primary site?

This secondary site of this cluster only exists to support a SQL Always On Availability group that's using asnyc commit for DR. Failover to the DR site will be manual. All nodes at the DR site are set to NodeWeight = 0 so they don't participate in quorum anyway.

Best Answer

The concepts of quorum, and owners are separate topics. Just because a member of the WSFC doesn't get a vote in quorum does not mean it can't own a resource. Additionally, SQL Server doesn't really play a role at all--the same concepts apply regardless of what type of clustered resource you're dealing with.

Quorum: The quorum is the number of votes necessary to transact business on your WSFC. Depending on your WSFC configuration, voters can be nodes (servers), a drive, or a file share. You need more than 50% of your votes in order for the WSFC to be online. If you lose 50% or more of your voters, then the WSFC and all clustered services (including your FCI) will go offline and not come back until you have (or force) quorum.

In your configuration, you have a File Share Witness (hopefully in an impartial location reachable by both primary & DR sites), and you've also changed the NodeWeight to 0 for your DR servers. Rather than thinking of NodeWeight as "The DR servers don't get to vote," you should think of it as "The DR servers get to vote, but their vote doesn't count." They are still there, they're still part of the WSFC, it's just that the WSFC doesn't listen.

Even though the FSW isn't hosted on the cluster (or, perhaps, more accurately, because the FSW isn't hosted on the cluster!), it still has to be "present" to vote. Which brings us to the concept of ownership...

Cluster Owner/Host Server: Your WSFC has a network name and an IP address. That name & IP has to be tied to a machine that is part of your cluster. More specifically, it can be tied to any one machine in your cluster. This is part of your WSFC.

In your scenario, your DR servers have no vote for quorum, but they still must be possible owners of the WSFC Host Server. If you manually fail over to your DR site (which will involve forcing quorum because you have zero voters in DR), then one of your DR servers must host the cluster name & IP. If it doesn't, then your cluster can't come online.

Your FSW is "owned" by the same server that owns/hosts the Cluster itself. Therefore it must have all servers, including DR, as possible owners. When you do force quorum and come online in your DR site, the WSFC is going to want to continue talking to the FSW.

Your original question:

...what would be the effect if I limited the possible owners to just those nodes at the primary site?

I suspect that you would have problems when you tried to force quorum and bring your WSFC online in the DR site--though, that's just a guess.

From a SQL Server perspective, you just care that you have quorum and the WSFC is reliably up. If you're having issues in that area, I'd look at the specific issues to your reliability. In reality, you probably don't really care which server is hosts your WSFC (and FSW).