When a vCenter Server node is deleted or decommissioned from an Enhanced Linked Mode (ELM) environment without being gracefully unjoined, the remaining nodes retain stale entries in the VMware Directory Service (vmdir). This prevents the successful integration of replacement vCenter instances.
Symptoms include:
Removed vCenter Server still appears in the vSphere Client inventory.
Replication errors in /var/log/vmware/vmdir/vmdir.log.
Failure when attempting to repoint or join a new vCenter to the existing SSO domain.
vCenter Server 8
Stale Single Sign-On (SSO) metadata. The legacy vCenter Server object remains in the vmdir replication topology because the cmsso-util unregister process was not completed prior to the node's removal.
For the situation where a vCenter was removed from an ELM pair and then redeployed, but not in the same shared ELM SSO domain, follow these steps to prune the stale metadata and register the new vCenter node:
Preparation: Power off the new vCenter Server to ensure no replication conflicts occur during the cleanup.
Identify Stale Node: Log into the surviving vCenter Server via SSH and run: /usr/lib/vmware-vmdir/bin/vdcrepadmin -f showservers -h localhost -u administrator
Unregister Legacy Node: Execute the unregister command on the Production vCenter, targeting the PNID (FQDN or IP) of the old vCenter: cmsso-util unregister --node-pnid <OLD_VC_FQDN> --username administrator@<SSO_domain>
Verify Removal: Re-run the command from Step 2 to confirm the legacy node is no longer listed.
Rejoin New Node: Power on the new DR vCenter Server and run the repoint command to integrate it into the Prod VC domain: cmsso-util domain-repoint -m pre-check --src-emb-admin Administrator --replication-partner-fqdn FQDN_of_destination_node --replication-partner-admin PSC_Admin_of_destination_node --dest-domain-name destination_SSO_domain
For more details on managing ELM topologies, refer to the Broadcom TechDocs: