Namespaces stuck in a terminating state in RKE2 v1.32.5+rke2r1 clusters with the Calico API Server enabled

Article Number: 000021943

Environment

A standalone or Rancher-provisioned RKE2 v1.32.5+rke2r1, with the Calico CNI, in which the Calico API Server (disabled by default) is enabled

Situation

Namespaces are stuck in a terminating state when attempting to delete the, but disabling the Calico API Server allows the namespaces to be deleted successfully.

Cause

This is a result of an upstream Calico issue affecting the Calico API Server on Calico v3.30.0 when running in a Kubernetes v1.32 cluster. Calico v3.30.0 was shipped in RKE2 v1.32.5+rke2r1.

Resolution

This upstream Calico issue is fixed in Calico v3.30.1 which is shipped in RKE2 v1.32.6+rke2r1. Affected clusters should be upgraded to resolve the issue.