The coreDNS and kube-proxy containers within the target Tanzu Kubernetes Cluster (TKC) are in an abnormal, non-functional state

Products

• VMware vSphere Kubernetes Service

Issue/Introduction

Cluster checks confirm that the kube-proxy pods cannot pull images and the coreDNS pods remain stuck in a Pending state.

 

Environment

TKC

Cause

- The coreDNS pods are pending because their nodeSelector targets Kubernetes version 1.31.14, whereas the actual TKC cluster is running version 1.30.11, which in turn prevents kube-proxy from resolving image registry domains.

- Configuration analysis shows that the mismatched nodeSelector criteria prevent coreDNS from scheduling onto any available nodes, completely breaking cluster DNS resolution and causing subsequent image pull failures for kube-proxy.

Resolution

Temporarily patch the coreDNS deployment to remove the incorrect nodeSelector

kubectl patch deployment coredns -n kube-system --type=merge -p '{"spec":{"template":{"spec":{"nodeSelector":{"run.tanzu.vmware.com/kubernetesDistributionVersion":null}}}}}'
kubectl get deployment coredns -n kube-system -o jsonpath='{.spec.template.spec.nodeSelector}'

Perform a rollback (kubectl rollout undo) to revert coreDNS to its correct, stable Release Condition (RC) version.