When adding a new host/witness or re-adding a host to a vSAN cluster via vSphere Web client the host becomes network partitioned.
The vSAN Skyline Health Alert "vCenter state is authoritative" is triggered.
To confirm the mismatch of vSAN UUID between vCenter and ESXi do the following:
esxcli vsan cluster get output as seen below in the next step. The string of alphanumeric characters after the ':' is the vSAN cluster UUID and should align with the Sub-Cluster UUID as seen in the below esxcli vsan cluster get output.SSH to one of the hosts in the cluster that is part of the vSAN cluster (not partitioned) and run esxcli vsan cluster get , as seen below the vSAN cluster UUIDs don't match up with what is seen on the vSphere Client. Refer the below snippets.
[root@host] esxcli vsan cluster getCluster InformationEnabled: trueCurrent Local Time:YYYY-MM-DDTHH:MM:SSZLocal Node UUID: 5c48d76f-07f0-fa60-81ac-############Local Node Type: NORMALLocal Node State: AGENTLocal Node Health State: HEALTHYSub-Cluster Master UUID: 5c4f4d53-d623-d510-####-############Sub-Cluster Backup UUID: 5c48d99b-573c-d974-####-############Sub-Cluster UUID: 52f1b2fb-fe25-3210-####-############Sub-Cluster Membership Entry Revision: 71
Unicast Mode Enabled: true
Maintenance Mode State: OFF
Config Generation: 36979731-f1b3-4ea6-####-############ 166 YYYY-MM-DDTHH:MM:SS.SSS
Mode: REGULAR
VMware vSAN 7.0.x
VMware vSAN 8.0.x
One of the ways the vSAN UUID goes out of sync between vCenter and vSAN is:
When the vSAN cluster is "Turned OFF" instead of "Shutdown".
To bring vSAN back ON, the hosts are joined back to the vSAN cluster as per Improper vSAN Shutdown and Accidental turning off vSAN leads to Cluster Configuration Loss and Service Disruption.
Now the new vSAN cluster would have a different vSAN cluster UUID on vSphere client, while the hosts would have a different vSAN cluster UUID.
If the above symptoms and issue matches, please contact Broadcom Support to investigate the issue.