vSAN: Mismatched Versions

[DrSeussFreak]

Note: This is all my home lab

I have been running 3 ESXi hosts on ESXi 6.7 and 1 host on ESXi 6.5. The 6.5 host is too old to ugprade, 6.7 will not install due to the old CPU. The 6.7 hosts have Virsto.DiskFormatVersion set to 5, so they are compatible with the 6.5 host.

Yesterday, I added a new 6.7 host, and I forgot to change the Virsto.DiskFormatVersion on this new host, so it was added to the cluster with Virsto.DiskFormatVersion set to 6, and now, the 6.5 host cannot see any VMs on the vsanDatastore.

I have removed the offending 6.7 host from the cluster, and removed the 6.5 host as well. I change the Virsto.DiskFormatVersion on the new hosts to 5, and re-added the 6.5 host, but it can still not see anything on the vsanDatastore.

Is there a way I can salvage this? I really need that 6.5 host, so I am hoping someone has a way to help me get this working again.

Thanks in advance,

DSF

Message was edited by: Dr. Seuss Freak fixed formatting

[TheBobkin]

Hello DrSeussFreak​

Welcome to Communities.

"I have removed the offending 6.7 host from the cluster, and removed the 6.5 host as well. I change the Virsto.DiskFormatVersion on the new hosts to 5, and re-added the 6.5 host, but it can still not see anything on the vsanDatastore."

How exactly was the 6.5 node removed and re-added to the cluster? e.g. does it still have its original Disk-Groups intact and healthy?

If you didn't recreate the Disk-Groups on re-joining the cluster check using:

#vdq -Hi

#esxcli vsan storage list | grep CMMDS

Are you positive you didn't perform On-Disk format for the remaining 6.7 nodes in the absence of the 6.5 node?

Can check this via GUI but more clear from RVC:

> vsan.disks_stats <pathToCluster>

> vsan.obj_status_report <pathToCluster>

Sounds like the 6.5 node is potentially isolated from the cluster - if the 'Remediate cluster' button is clickable via the 'vCenter is authoritative' Health check under Cluster then potentially you can remediate unicastagent entries automatically - otherwise make the necessary checks e.g. node has vSAN-enabled vmk(s) that can communicate with the other nodes vSAN-enabled vmk(s), all nodes in cluster have 6.5 node entry in their unicastagent lists, 6.5 node has unicastagent list entries for all other nodes, node is out of Maintenance Mode, node is not having some other issues such as controller physical/driver/firmware issue (#dmesg is your friend here).

Bob

[DrSeussFreak · Accepted solution]

I am just rebuilding, I appreciate the response though.  I already began.

I am well aware I am running in a very unsupported config, and all of my VMs became inaccessible.