MS failover cluster shared disk fails validation with 'Failed while verifying removal of any Persistent Reservation on physical disk'

Question

Hi

I want to create a two-node Windows Server cluster on which I will install MS SQL Server in 'cluster mode'. But I'm having trouble creating/using a cluster shared disk.

I've deployed a pair of MS Windows Server 2022 Standard systems as VMs in a vSphere 7.0.3 environment. Following the instructions in https://www.microfocus.com/documentation/open-enterprise-server/2018-SP3/clus_vmware_lx/t4cxgj97pshr.html, I created a second disk and attached it via vSphere Client to both of the Windows Server VMs.

After I created the failover cluster, I ran the 'Validate Cluster' task. I noticed that the storage section of the report showed an error that the shared disk was raw, and could not be used. So I used Disk Management to initialize the disk. I then re-ran the Validate Cluster task This time, the output again reported failure in the Storage section.

In the List Disks section, for both cluster nodes, I noticed the following disk characteristics: "Disk is on the system bus. Disk partition style is MBR. Disk type is BASIC." This seemed like a 'good' status to me.

But in the 'List Disks To Be Validated' section, I see:

"Failed while verifying removal of any Persistent Reservation on physical disk 8940f725 at node xxx.lab.local."

I've spent several hours trying to solve this problem, without success. I've tried changing some of the values of the the shared disk in vSphere Client, without success.

While https://kb.vmware.com/s/article/79616 provides some guidance, it doesn't mention this error, nor any facts that would seem, to me, to lead to a solution to my problem.

What's the cure to this problem?

Thanks!

Answer 1

Hi Terry,

Hope you're doing well.

1. Ensure that the shared disk configured in the vSphere Client is VMware-compatible and properly attached to both Windows Server VMs. Check that the disk is set to "Independent Persistent" to ensure consistency between virtual machines.
2. Make sure that the partition style of the disk is GPT (GUID Partition Table), not MBR (Master Boot Record), because cluster disks usually require GPT.
3. Make sure the latest version of VMware Tools is installed on the Windows Server VM. VMware Tools ensure that VMs interact correctly with the vSphere environment.
4. In the vSphere Client, ensure that any persistent reservations for the disk have been removed. You may need to unmount the disk from the vSphere Client and reattach it to ensure there are no remaining persistent reservations.

Best Reagrds

Answer 2

Thank you for your help! Regarding your suggestions:

"1. Ensure that the shared disk configured in the vSphere Client is VMware-compatible and properly attached to both Windows Server VMs. Check that the disk is set to "Independent Persistent" to ensure consistency between virtual machines."

I used vSphere Client to check the values for this disk on both VMs, and ensured that 'Independent Persistent' is set.

"2. Make sure that the partition style of the disk is GPT (GUID Partition Table), not MBR (Master Boot Record), because cluster disks usually require GPT."

I had set the disk to MBR. I changed it to GPT and re-ran the Validate Cluster test; no change in behavior: the 'List Disk To Be Validated' test still fails with "Failed while verifying removal of any Persistent Reservation on physical disk".

"3. Make sure the latest version of VMware Tools is installed on the Windows Server VM. VMware Tools ensure that VMs interact correctly with the vSphere environment."

I have applied all Windows Updates to these two Windows Server 2022 systems.

"4. In the vSphere Client, ensure that any persistent reservations for the disk have been removed. You may need to unmount the disk from the vSphere Client and reattach it to ensure there are no remaining persistent reservations."

I've searched for, but don't see how to view persistent reservations in vSphere Client.

Thanks again for the help!

Answer 3

I finally found my own answer. Looking through https://kb.vmware.com/s/article/79616, I noticed:

Datastore

• Require the datastore capability Clustered VMDK support to be enabled.
  Note: Perennial reservations are required on all datastores with Clustered VMDK support enabled.

When I enabled the 'Clustered VMDK' value on the vSphere datastore that I used to store the shared cluster disk, the persistent reservation failures went away.

FYI

tl

Answer 4

Thanks very much for your detailed reply.

Answer 5

Hi - any other thoughts on this? Is there any other information I can provide?