I'm working on setting up a 3 node multi-subnet FCI+AG solution with two FCI nodes in the primary data center on one subnet, and a third node with a standalone SQL instance on another subnet for DR. I've used the Microsoft whitepaper https://docs.microsoft.com/en-us/previous-versions/sql/sql-server-2012/jj215886(v=msdn.10)?redirectedfrom=MSDN as a guide and have ran cluster validation and done testing along the way. I'm using CSVs for storage and can't get past this one issue: As soon as I add the third node to the cluster from the second DR subnet, CSVs can no longer move between the two primary data center nodes, either manually or when one of the primary data center nodes is rebooted. When I try to move a CSV, it goes to Offline Pending and remains in that state, and eventually the SQL role fails due to no disk access (as would be expected).
Quorum is configured using Node and Fileshare and is set to exclude the DR node. The DR node has also been removed as a possible or preferred owner of all resources, including the CSVs (using Powershell set-clusterownernode on the CSV resources). RegisterAllProvidersIP is set to 1.
The only way to recover is to stop the cluster service on the third node...once it stops, everything recovers. Since the third node isn't part of quorum and can't own anything, why can't the disk resources move between the primary data center nodes?
Event Logs and Clustering logs don't reveal anything, at least not from what I can interpret. I've done plenty of research and haven't come across anything on this specific issue. Also, I can add a third node from the primary data center subnet and this issue does not occur. So it's an issue somewhere with the multi-subnet configuration or something with the third node itself. Any input/assistance is appreciated.
Thanks,
Kevin