When the cluster has no metrics-server or it is not ready, kubectl top and HPA resource metrics (CPU/memory) are unavailable. Autoscaling and resource monitoring based on usage cannot work.
Warning
N/A
- Report shows metrics-server is missing or not ready
- No metrics-server pods Ready in kube-system
- Deploy metrics-server using the official manifest or Helm chart
- Ensure it can reach kubelet metrics port (e.g. 10250) and resolve TLS/network policy
- Verify:
kubectl top nodesandkubectl top pods -n <namespace> - If using HPA, ensure HPA can get resource metrics after metrics-server is ready