Skip to main content
Service has no endpoints

Kubernetes Service has no endpoints

A Kubernetes Service currently has no ready backend addresses to receive traffic, even though its virtual IP and DNS name may still exist.

Service and EndpointSlice controllersService networkinghigh
Exact message
Endpoints: <none>

Also searched as: no endpoints · EndpointSlice empty

What it means

The Service selector may match no Pods, matching Pods may be unready, or a selectorless/external endpoint definition may be absent. The Service virtual IP can exist while every request still has nowhere to go.

Typical trigger conditions

  • Selector does not match Pod labels
  • All matching Pods fail readiness
  • Wrong namespace
  • Manual EndpointSlice missing or malformed
Evidence before intervention

First diagnostic checks

1

Inspect selector and endpoints

Compare the Service selector with EndpointSlices generated for it.

kubectl describe service <service> -n <namespace>; kubectl get endpointslices -n <namespace> -l kubernetes.io/service-name=<service>
2

List matching Pods

Apply the exact selector and inspect readiness rather than listing all Pods.

kubectl get pods -n <namespace> -l '<selector>' -o wide
3

Trace readiness failures

A matching but unready Pod is intentionally excluded from normal Service endpoints.

Version and platform notes

EndpointSlice is the current scaling mechanism; legacy Endpoints behaviour and deprecation status depend on Kubernetes version.

Engineering signal

Follow Kubernetes failures and fixes

One useful weekly email with new error references, tools, integration notes and production lessons. No daily noise.