We have renamed the volume folder as supported here:
The volume folder name can be renamed, but the ClusterStorage folder cannot be renamed. An example of a path with a volume folder with a changed name is as follows:
C:\ClusterStorage\VMGroupA
https://technet.microsoft.com/en-us/library/dd630633%28v=WS.10%29.aspx#BKMK_paths
Everything is working well, we can reboot without problem. The issue occur when the connection to our iSCSI storage is lost (Mainly caused by a NIC issue).
When the Host is up, the cluster service is started and we are able to resolve the problem on the NICs. After the NICs problem resolved, the connections to the iSCSI storage is restored.
At this time, the cluster service rename itself the volume folders to Volume1 and Volume2... As read in another post, it seems the order of Volume1, Volume2... can change. (https://social.technet.microsoft.com/forums/windowsserver/en-US/269cfac1-e72a-4e10-a6c5-07e94c1f1732/ws2012-csvs-changed-volume-name-mount-point-and-need-to-fix)
Then, the VMs are failing to start with the message "at least one file is missing".
Why this rename occur?
To allow our VMs to runs, we have to rename the Volume folder as it was named before the lost of connection and everything is getting back to normal.
Additionally, before we renamed the volume folders, we tried to import the VMs using the Hyper-V manager because the VMs hosted on the automatically renamed volumes were disappeared from the list of VMs. All VMs still visible in in the Failover Cluster Manager with the "Fail" Status.
In this case, the import failed because some paths are adequately changed to the volume folder, some are not. It seems:
- the path of the VM files is not changed and remains with the post NIC failure volume foder name
- The HD paths are changed adequately with the automatically renamed volume folder
- The checkpoint folder is changed
- The Smart Paging folder is changed