Kubernetes New Added 8 September 2026

Kubernetes: Readiness probe failed, endpoints never become ready

The container is running but its readiness probe never succeeds, so the Service keeps it out of the endpoint list and traffic goes nowhere. Common causes are a probe pointed at the wrong port or path, an app that binds to loopback only, or a startup slower than the probe's tolerance.

Quick fix

Read the commands before running them. Anything that restarts a service, deletes data or changes permissions should be tried on a non-production system first.

Quick fix

kubectl describe pod POD | grep -A5 Readiness
kubectl get endpointslices -l kubernetes.io/service-name=SVC

# Probe from inside the pod's own network namespace
kubectl exec POD -- wget -qO- http://127.0.0.1:8080/healthz

# Slow start is a startupProbe, not a longer readinessProbe
startupProbe:
  httpGet: { path: /healthz, port: 8080 }
  failureThreshold: 30
  periodSeconds: 5

# The app must listen on 0.0.0.0, not 127.0.0.1

How to diagnose Kubernetes errors

Kubernetes errors are best read as a state machine that got stuck. Pending means the scheduler could not place the pod (resources, taints, or an unbound volume). ImagePullBackOff means kubelet could not fetch the image (name, credentials, or registry). CrashLoopBackOff means the container starts and exits, so the answer is in the container's own logs. OOMKilled means the kernel killed it for exceeding its memory limit. Each state points at a different subsystem, and kubectl describe almost always contains the exact reason in its events.

If the quick fix above does not resolve it, work through these steps. They apply to this whole class of error, not just to this one message, which is usually what saves the time.

Start with kubectl describe pod <name> and read the Events section at the bottom. It names the precise failure, including registry errors and scheduling constraints.
For CrashLoopBackOff, read the previous container's logs: kubectl logs <pod> --previous. The current container may not have produced output yet.
Check resource pressure with kubectl top pod and kubectl top node, and compare against the pod's requests and limits.
Test RBAC directly: kubectl auth can-i <verb> <resource> --as=system:serviceaccount:<ns>:<sa>. This answers permission questions definitively.
For networking, confirm the Service has endpoints (kubectl get endpoints) before suspecting DNS, Ingress or the CNI.

Tools worth reaching for

kubectl describe
kubectl logs --previous
kubectl auth can-i
kubectl top
kubectl events --sort-by=.lastTimestamp
k9s

Authoritative references

Primary documentation for this error, worth reading before applying any fix in production.

kubernetes.io

Related Kubernetes errors

See all 34 Kubernetes errors →

Browse other categories

Something missing or wrong?

This entry is maintained by hand. If the fix is out of date, incomplete, or you have a better one, email a correction and it will be reviewed.