TMC-SM Supervisor Stuck in Unknown Health State
search cancel

TMC-SM Supervisor Stuck in Unknown Health State

book

Article ID: 435181

calendar_today

Updated On:

Products

VMware Tanzu Mission Control - SM

Issue/Introduction

In a Tanzu Mission Control Self-Managed (TMC-SM) environment, a Supervisor cluster reports health as Unknown state indefinitely after a re-registration attempt.

Additional symptoms include:

  • AgentInstall is deleted using the command "kubectl delete agentinstall"
  • The Supervisor is not manageable within the TMC console.
  • Following a re-registration attempt the AgentInstall job on the Supervisor is marked as Complete or INSTALLED but the cluster fails to transition the status to Ready.

Cause

Manual deletion of the AgentInstall Custom Resource (CR) fails to fully uninstall components because the TMC Cluster Agent is a Managed Resource governed by a declarative reconciliation loop. Directly deleting the AgentInstall CR via kubectl delete removes the configuration object but does not trigger the proper, orchestrated cleanup routines by the TMC controller/agent.

Resolution

To resolve this, you must perform a clean uninstallation of the existing agents using the TMC agent uninstaller before attempting a new registration.

Step 1: Perform a Clean Uninstall

  1. Log in to the Supervisor cluster context.
  2. Create the AgentInstall configuration to use the UNINSTALL operation. Use the following YAML template (replace TMC-NAMESPACE with your actual namespace, e.g., svc-tmc-c#) with name uninstall-agents.yaml:

    apiVersion: installers.tmc.cloud.vmware.com/v1alpha1
    kind: AgentInstall
    metadata:
      name: tmc-agent-installer-config
      namespace: TMC-NAMESPACE
    spec:
      operation: UNINSTALL
  3. Apply the configuration to trigger the removal with kubectl apply -f uninstall-agents.yaml

  4. Verify that the agent pods and resources have been removed from the namespace with kubectl get all -n TMC-NAMESPACE

Source: Manually Remove the Cluster Agent from a Supervisor Running in vSphere

Step 2: Re-register the Supervisor cluster:

For detailed steps, refer to: Register a Management Cluster with Tanzu Mission Control