High Availability Problem - VMs on vVol not protected but are on VMFS

[guerrillataktix]

We have a couple new ESX 7.0.3 clusters having trouble with High Availability. They use a Pure Storage SAN for vVols and VMFS datastores. We found that those VMs on the vVol datastore are not protected but those on a VMFS datastore are. HA is not having any trouble configuring itself on the hosts, its just that machines are showing as unprotected. 

I have a support case open about the issue but figured I would ask the community for input as well.

Any ideas why the vVol would make a difference? 

Our second site has the same sort of setup and is not having the issue.

[Thomas_França]

The behavior you described, where VMs on vVol datastores are not protected by High Availability (HA) while those on VMFS datastores are, could be attributed to various factors. Here are some:

1. Storage Awareness: VMware HA uses storage heartbeat to determine the availability of a host. If there is an issue with storage communication or latency for VMs on vVol datastores, HA might consider those VMs unprotected. Ensure that the storage communication between the ESXi hosts and the Pure Storage SAN for vVols is stable.

2. Storage Policies: vVols provide more granular storage management with Storage Policies, which define features like replication and availability. If the storage policies assigned to vVol-based VMs do not meet the requirements for HA protection, VMs might not be considered protected.

3. Storage Configuration: Ensure that the storage configuration for vVols is correctly set up and aligned with VMware HA requirements. Incorrect storage settings might lead to issues in determining VM protection status.

4. VMware Version Compatibility: Verify that your version of vSphere (7.0.3) and the version of Pure Storage you're using are compatible and have been tested together. Incompatibilities could lead to unexpected behavior.

5. VM Configuration: Check the virtual machine settings for VMs on vVols. Make sure that these VMs are configured properly and adhere to best practices for HA protection.

6. Network Configuration: Proper network configuration is crucial for both storage and HA communication. Any network issues between ESXi hosts and storage can impact HA protection.

[guerrillataktix]

The solution ended up being to restart the VASA service on our Pure array.

[Jasemccarty]

Where does it indicate that they aren't protected on vVols?

[guerrillataktix]

On the High Availability status page at the cluster level under the monitor tab in Vcenter. There is a metric for Protected and UnProtected machines - it shows zero protected if the machine has N/A status (on the HA Protected column in VMs view) or unprotected in some other cases.

As mentioned above when looking at all VMs on the vVol datastore there is an HA Protected column and the status is N/A or Not Protected for all.

Thanks for taking time to inquire on the issue.

[guerrillataktix]

Another thing I have noticed is if trying to browse the vVol datastore in the Vsphere client it shows the .Vsphere-HA folder but if i try to check the contents it throws an error about how it cannot be found. 

I looked for the same folder on the hosts using SSH and it doesn't show there in the CLI or if I look using WinSCP. I am looking with hidden folders showing.

If i try to rename the .vsphere-HA folder through by CLI it doesn't give an error about how it can't find it so it seems like it is there.

Example:

[root@host:/vmfs/volumes/vvol:33b205d688d63048-997eb7c2b83a972e] mv .vSphere-HA .vSphere-HA_old
mv: can't rename '.vSphere-HA': Invalid argument
[root@host:/vmfs/volumes/vvol:33b205d688d63048-997eb7c2b83a972e] mv test .vSphere-HA_old
mv: can't rename 'test': No such file or directory