Windows Fail Over Cluster and Always On

I created a cluster with 2 nodes, both VMs with SQL servers, in a data centre, A. All versions were 2012, OS and SQL.  The quorum was setup between the 2 SQL server VMs and a shared network location in a different data centre, B. Then I added an Always On Group (AO) on the cluster. I had the AO setup with synchronous mirroring between primary and secondary and it was working just fine.

To this setup I wanted to add 2 new nodes, also VMs but in the B data centre under the same domain but different subnet. The intention was to have the databases from data centre A mirrored asynchronously to centre B to the 2 new nodes through the AO group.

As soon I added the 2 new nodes to the cluster, going through the validation and such, the cluster was brought down in the sense that in the both 2 original nodes the E and F drives (for SQL data and log files) were put offline, which caused SQL severs to fail and stop. Any attempted to start them was futile due to the missing drives that went offline.

I could not bring back online the E and F drivers on any of the 2 boxes before I destroyed the cluster itself. If I tried that before there was an error message saying that they "cannot be brought online due to a policy set by an administrator which is controlled by the fail-over cluster". The message is not exact. After I destroyed the cluster I was able to bring back online the E and F drivers and start the SQL server on both VMs.

My question is how and why was possible to bring down a cluster by adding 2 new nodes to it.
LVL 27
ZberteocAsked:
Who is Participating?
 
lciprianionutConnect With a Mentor Commented:
Because when you add 2 nodes and did cluster validation in that test is a test for disk system, meaning taking disks offline and online. This should never be done on SQL as will lead to downtime, it should be properly planed.
0
 
ZberteocAuthor Commented:
Are the disk taken offline and back online only because of the test or it is done anyway because of cluster change? Another question would be if this happens on all the nodes in the cluster or only on the new ones?
0
 
lciprianionutCommented:
As far as I know when you add nodes into a cluster first thing to do is to validate the cluster configuration with cluster validation tool, to be in a supported state by Microsoft. Well this validation will take disks offline and online on all the nodes that are participating in this validation. Validation is happening on nodes that were selected to take part on the test.
0
 
ZberteocAuthor Commented:
Mersi Cipriene. Numai bine! ;)
0
All Courses

From novice to tech pro — start learning today.