Readiness probe failed
The container is running, but its readiness probe fails, so the pod’s Ready condition is False and it’s listed as not ready in the Service’s EndpointSlices. Nothing restarts.
Warning Unhealthy kubelet Readiness probe failed: HTTP probe failed with statuscode: 503Diagnose in order
Section titled “Diagnose in order”- Read the probe failure message in the events: status code, timeout, or connection refused.
- Check the probe definition matches reality: path, port (name or number), scheme (HTTP versus HTTPS).
- Call the same endpoint from inside the pod and read the response body, which often says which dependency failed.
- Check the effect on traffic:
kubectl get endpointslices -l kubernetes.io/service-name=<svc> -o wide.
Usual causes
Section titled “Usual causes”| Message | Cause |
|---|---|
statuscode: 404 |
Wrong path |
connect: connection refused |
Wrong port, or the app isn’t listening yet |
statuscode: 503 |
The app reports itself not ready, often a dependency check failing |
http: server gave HTTP response to HTTPS client |
Probe scheme doesn’t match what the app serves |
Correct the probe, or fix the dependency the app is waiting for. If every replica is not ready because of one shared dependency, reconsider whether readiness should depend on it at all.
Read next: How a pod gets from YAML to running.