When managing a vSphere with Tanzu environment, administrators may observe the following behavior:
1. The vSphere Client Task Console displays one or more entries for "Uninstalling Namespace Node operation" stuck in a Pending or In-Progress state.
2. These tasks do not progress over time and cannot be manually canceled through the vSphere UI.
3. The affected ESXi hosts may have recently been moved between clusters or placed into Maintenance Mode while Workload Management was enabled.
VMware vSphere: 7.x, 8.x
vSphere with Tanzu
This issue occurs due to a synchronization mismatch between the vCenter Server Task Manager and the ESXi Agent Manager (EAM).
When an ESXi host is removed from a Supervisor-enabled cluster—or moved while in Maintenance Mode—vCenter attempts to decommission the Namespace Node components (Spherelet and CRX). If the host is unavailable or in a state that prevents the "handshake" from finalizing, the task record remains orphaned in the vCenter VPX_TASK database table. Because the task is no longer tied to an active process on the host, it remains in a permanent "Pending" state at the management layer.
To resolve this issue and clear the orphaned tasks from the vSphere UI, the vCenter Server management service must be restarted to refresh the task queue.
Step 1: Verify the Host State
Ensure that the ESXi host is no longer part of a Supervisor Cluster.
Step 2: Restart the vpxd Service
Restarting the vpxd service will flush non-persistent orphaned tasks.
1. Log in to the vCenter Server Appliance (VCSA) via SSH as root.
2. Access the BASH shell.
3. Run the following command to restart the VMware vCenter Content:
service-control --restart vpxd
(Note: This will temporarily disconnect the vSphere Client; ensure no critical vCenter-level management operations are running.)
Step 3: Verification
Once the service has restarted, log back into the vSphere Client. The "Uninstalling Namespace Node" tasks should no longer appear in the Task Console.
To avoid orphaned tasks in the future, follow the recommended lifecycle procedures for Supervisor Clusters:
1. Ensure Workload Management is fully disabled or the host is properly decommissioned before moving ESXi hosts out of a Tanzu-enabled cluster.
2. Avoid moving hosts that are in Maintenance Mode directly out of a Supervisor Cluster without first confirming the "Uninstalling Namespace Node" task has completed successfully.
Related Information: Broadcom Documentation: Remove Hosts from a Supervisor Cluster