We have setup a 3 node AAG in our staging environment as a POC. The 3rd node has been configured in a different subnet and also as an asychronous replica. The Windows Server version is 2012 R2 and the SQL Server version is 2014 SP2. After I changed the availability mode to synchronous commit and tried to perform a manual failover across subnets, this is what I get. I had removed the node weight of the 3rd node. Am I missing something here? Please advice.
-
That image just has a warning. Did the failover not work as expected? Have you read the article the message links to? What is your actual vote configuration?– LowlyDBA - John MCommented May 3, 2018 at 19:00
-
The 2 nodes in one subnet have a vote and witness has a vote. This 3rd node in the other subnet dosent have any votes..– buddinggeekCommented May 3, 2018 at 19:33
-
Please let me know what is your quorum setting, I guess it will be Node and fileshare majority ?.– ShankyCommented May 4, 2018 at 6:52
-
Yes it’s node and file share majority– buddinggeekCommented May 4, 2018 at 11:17
1 Answer
I had removed the node weight of the 3rd node. Am I missing something here?
You're receiving the informational message because you've removed the possible vote (node weight of 0) of the replica. Since it won't ever have a vote, it also won't ever be able to stand on its own should any issues arise with the cluster, since this effectively removed the dynamic quorum aspect of clustering when this node owns anything.
This truly may be what you want, for example as a DR node whose availability you don't want impacting the production nodes. In that case, as long as you understand the situations you're protecting against and meeting those then I don't see a problem setting it like this (which is effectively that you know better than the cluster).