This article outlines the correct, top-down sequence to cleanly remove all TMC-SM components across the architecture: starting with the managed guest clusters, moving to the registered Supervisor cluster, and finally removing the central TMC-SM control plane.
Administrators may need to completely uninstall and remove the TMC-SM footprint from their environments.
To ensure a clean uninstallation without causing Kubernetes garbage collection deadlocks, the teardown must follow this specific order:
Phase 1: Remove Agents from Managed Guest Clusters
Before touching the management layer, you must remove the TMC agents from all downstream workload clusters.
If your TMC-SM control plane is functioning normally and you intend to keep the workload clusters in your environment:
Note: If your TMC-SM control plane is non-functional, if a cluster is stuck attempting to unmanage, or if you intend to permanently delete the workload clusters rather than keep them, do not proceed. Please contact Broadcom Support for assistance.
Phase 2: De-register the Supervisor Cluster.
Once all guest clusters are unmanaged and clean, sever the connection between the vSphere Supervisor and TMC-SM.
Phase 3: Uninstall the TMC-SM Control Plane
After all downstream agents and Supervisor registrations have been cleared, you can safely tear down the central TMC-SM hosting environment.
You have two options depending on your cluster utilization: