After decommissioning a vSAN cluster (uninstalling vSAN from hosts and moving them to a new cluster), a "ghost" or "zombie" vSAN datastore object remains in the original Cluster object.
VMware vCenter Server 8.x
The issue is caused by stale references in the vCenter Database (VCDB). Even though vSAN has been disabled on the ESXi hosts and the hosts have been moved or formatted, the vCenter inventory database retains a record of the vSAN datastore object. Because vCenter believes the datastore exists within the cluster, it prevents the deletion of the cluster object to maintain referential integrity.
Note: This issue typically occurs after manually dismantling a vSAN cluster without fully decommissioning the datastore object first.
Since the datastore object does not exist on any physical hosts, the stale reference must be removed directly from the vCenter Database.
Verify Host Status: Ensure all hosts have been removed from the cluster or have vSAN completely disabled.
As this resolution requires direct modification of the vCenter Database (VCDB) to purge the specific "zombie" entity rows, please open a support request with Broadcom Support.
To prevent this in future decommission scenarios, ensure the vSAN datastore is fully destroyed/unmounted while the hosts are still in the cluster and before the cluster object is emptied. Follow the standard "Decommission vSAN Cluster" procedures which include removing the disk groups and disabling vSAN on the cluster object level first.