VMware Cloud Community
dhorghidan
Contributor
Contributor

Snapshot of cluster node with RDM disks

I need to have a snapshot created for one node in a 2 nodes active/active WSFC Win 2012 failover cluster . There is SQL on top of the clusters.  There is node 1 with SQL cluster 1 running on it as preferred node, and node 2 with SQL cluster 2 running on it as preferred node. Each of the 2 nodes contain C and D drives (vmdk). In addition each node have the cluster drives which are all RDM disks (SQL data drives, log drives, quorum). 

All I need to do is get a snapshot of node 1 since we need to install a software on its C drive (not RDM). I know that hot snapshot for a VM with RDM disk cannot be done.

However, for steps I consider doing a cold snapshot like below:

1.failover cluster resource from node 1 to node 2, now on node 1 remain just the C drive (OS) and D drive (application) and node 1 do not have any more RDM disks, all RDMs are on node 2.

2. Shutdown node 1 and cold snapshot.

3. Start node 1 and run software install. if software install successful then fail back cluster on node 1 and done change request.

4. if software install unsuccessful then shutdown node 1 , restore cold snapshot back to node 1 then restart and fail back cluster1 back to it. 

I am not experienced in this, I am the DBA for the environment and contacted our VMWare admin who told me that in order to get a snapshot, we need to disconnect all RDM from node 1 first, then take snapshot. But since I propose to take cold snapshot as per steps 1-4 then all RDM disks failover to node 2 , and no RDM on node 1 where I need to take cold snapshot remain.

Wondering if any of you can share a similar scenario.

I appreciate your help.

Thank you.

Moderator note by wila: Moved post from Technical Community Resources to vSphere discussions area

 

Labels (1)
Tags (1)
Reply
0 Kudos
1 Reply
pdirmann01
Enthusiast
Enthusiast

The disks are still a "part" of Node 1's configuration even though they aren't actively mounted in the OS. A snapshot will not consider if it the disk is "active on the cluster node". Typically, you could change the disks that you don't want to be part of the snapshot to 'Independent - Persistent" disks, which requires a shutdown of the VM and that would essentially exclude them from a snapshot. But there are other VM-level configurations that could also prevent it, e.g., if all of your disks are on one SCSI controller and it has physical bus sharing enabled (which is needed to a traditional WSFC). Are you able to post more about the VMs' configs?

Reply
0 Kudos