Supervisor deployment stalls at 11/16 after NSX VPC configuration

Products

• VMware NSX
• VMware Cloud Foundation

Issue/Introduction

• When deploying a Supervisor cluster in VMware Cloud Foundation (VCF) 9.1, the deployment may fail or be intentionally deactivated.
• Although the deactivation process in vCenter reports success, stale networking objects remain in the NSX Manager.
• These leftover components prevent a clean redeployment of the Supervisor cluster.
•  The Supervisor cluster is no longer present in vCenter after deactivation.
•  NSX Manager continues to display Load Balancers, Virtual Servers, and Server Pools associated with the deactivated cluster.
•  High volume of stale objects (e.g., 70+ Virtual Servers) that cannot be deleted en masse via the UI.

Environment

VMware Cloud Foundation 9.1

Cause

• This behavior is a limitation of the vSphere Kubernetes Service (VKS) Supervisor deactivation workflow.
• The automated cleanup typically only removes components that were created strictly within the standard workflow.
• If networking components were provisioned via standalone or non-standard deployment methods, or if the workflow is interrupted, the cleanup process may skip these objects.

Resolution

To resolve this issue and prepare the environment for a fresh Supervisor deployment, the stale NSX objects must be manually removed in a specific order.
Prerequisites

• Take a backup of the NSX Manager configuration.
• Ensure you have Administrative access to the NSX Manager UI.

Procedure

1. Log in to the NSX Manager UI.
2. Navigate to Networking > Load Balancing.
3. Delete the objects in the following strict order (dependencies prevent deleting the Load Balancer first):
   • Virtual Servers: Identify and delete all Virtual Servers associated with the failed Supervisor deployment.
     Note: These must be deleted individually in the UI; bulk deletion is not supported for these objects.
   • Server Pools: Once the Virtual Servers are removed, delete the corresponding Server Pools.
   • Load Balancers: Finally, delete the Load Balancer instance itself.
4. Verify in vCenter that the Workload Management state is clear before attempting a redeployment.

Additional Information

• Managing vSphere Kubernetes Service with vSphere Supervisor
• For assistance with architecting a supported VCF 9.1 deployment workflow, consult your Broadcom Account Team or Professional Services.