After removing a duplicate vCenter Server (Compute Manager) from NSX to resolve a duplicate Virtual Distributed Switch (VDS) UUID conflict (error code: 9539), all NSX segments may disappear from the vCenter Server inventory.
When viewing the vCenter Server inventory, the NSX portgroups for the segments on the VDS are missing. Several objects display natively as nsx.LogicalSwitch:<UUID>. Additionally, when viewing the Edit Settings menu for an NSX-attached virtual machine, the assigned network configuration is completely blank.
This issue is an artifact of a previous VDS UUID conflict. Resolving the duplicate UUIDs by removing the conflicting Compute Manager results in orphaned identifiers for the NSX logical switches within vCenter Server.
The KB article Troubleshooting missing NSX-T segments in vCenter networking inventory suggests restarting vCenter services to force a rebuild of network object mapping. If this fails you can work around the issue by repreparing a transport node for NSX.
Unprepare and impacted ESXi hosts in the cluster from NSX.
Note: To limit impact, temporarily move required virtual machines to standard port groups before proceeding.
Remove the ESXi hosts from the cluster in vCenter Server.
Re-add the ESXi hosts back into the vCenter Server cluster to fully clear the previous state.
Re-prepare the cluster for NSX. This will successfully recreate the segment portgroups in the vCenter Server inventory.
Migrate the impacted production virtual machines to another prepared ESXi hosts using vMotion.
During the migration wizard, manually reselect the correct target network (segment) for each virtual machine. Once the virtual machines are correctly reconfigured and bound to the healthy segments, the orphaned nsx.LogicalSwitch:<UUID> objects will disappear from the inventory.