Hi
I have a two node cluster, which uses shared storage provided by StarWind (a virtual SAN which runs on the local storage of the servers) there is one CSV and one Witness disk. Everything in the cluster works correctly except under the scenario when the witness disk owner is suddenly reset with no clean shutdown, after a short period the CSV will switch over to the remaining node, but the witness disk seems to get stuck in the Online Pending state for a long time, this causes any virtual machines running on top of the server which is not reset to crash, stating that the virtual machines configuration can be found (even though the CSV has switched over already without issue). I have read about the check disk parameter to change it from 0 to 4 which I have done but this hasn’t made any difference, I have also ensured that the folder and sub-files in the witness disk are not read-only.
The switch over for the witness disk is extremely slow and causes the virtual machines to crash, is there anything else I can look at? All the settings for the cluster are configured in their default states such as the resource policies.
also there were no warnings or failures on my cluster validation, everything was passed perfectly.
Thanks
Steve