Tanzu guest cluster re-provisions after deletion due to legacy TanzuKubernetesCluster (TKC) object.

Products

• VMware vSphere Kubernetes Service

Issue/Introduction

• Unable to permanently delete a Tanzu Guest Cluster or TKG Cluster using the standard v1Beta1 Cluster object.
• After successfully deleting the cluster with the below command, the cluster gets automatically re-provisioned and returns to a Provisioned state within 30-60 seconds.

kubectl delete cluster <cluster_name> -n <cluster_namespace>

• The yaml file of the guest cluster confirms that the cluster object is a legacy v1alpha1 TanzuKubernetesCluster (TKC) object. Validate with the below command:

kubectl get cluster <cluster_name> -n <cluster_namespace> -o yaml | less

Environment

• VMware vSphere Supervisor 1.27
• VMware vSphere Kubernetes Service 1.26

Cause

The initial command (kubectl delete cluster) only deletes the newer, lower-level Cluster API (CAPI) Cluster object. As the TKC object still exists, the TKC controller immediately reconciles the state by recreating/re-provisioning the missing Cluster object.

Resolution

• To ensure a complete deletion, you must delete the TanzuKubernetesCluster (TKC) object with the below command: 

kubectl delete tanzukubernetescluster <cluster-name> -n <namespace>

• This instructs the TKC Controller to initiate the full cluster cleanup process, preventing re-provisioning.