Stale NSX-T "nsx.LogicalSwitch" on vCenter network view after supervisor restore from backup.

Products

• VMware NSX

Issue/Introduction

Below is a screenshot of stale NSX-T logical switch:

From the localhost access logs we observe DELETE api triggered to the NSX-T manager:

nsx_manager/var/log/proton/localhost_access_log.txt.7:2026-02-04T03:50:34.729Z - "DELETE /nsxapi/api/v1/infra/segments/vnet_########-####-####-####-############_0 HTTP/1.1" 200 - 927 927 +

From wcpsvc logs:

2026-02-04T03:50:30.595Z info wcp [kubelifecycle/restore.go:91] K8s Nodes was upgraded as part of Supervisor ########-###f-4##8-####-############ restore. Mark restore as done.

From configure-wcp.log on master-vm:

Feb 04 03:41:33 Online supervisor restore complete

Environment

VMware NSX
vSphere Supervisor

Cause

If the TKG clusters were created after the supervisor backup was taken. When the Supervisor restore was performed, the etcd database was rolled back to the state captured in that backup - a point in time before these clusters existed. As a result, their Kubernetes CRs (VirtualMachines, VirtualNetworkInterfaces, etc.) were absent from the restored Supervisor state, causing the removal of the nsx segments from NSX-T and remained on vCenter Server.

Resolution

Please contact Broadcom support  for assistance in Cleaning Up Orphaned Objects After Supervisor Restore from a Backup