The Service With No Endpoints

Priority: high · Reporter: on-call · Service: store

Since the last deploy, the storefront is returning nothing — clients hitting the store Service in the default namespace get no response. The backend Pods look healthy and are serving on port 80. "It worked yesterday."

You've been paged. The backend is up; something between the Service and its Pods is broken. Your job is to restore the store Service so it routes traffic to the running backend again.

No one has told you the root cause — that's the point. Investigate, form a hypothesis, fix it, and verify.

Where to start

A Service that returns nothing but whose backend is healthy almost always comes down to endpoints: is the Service actually selecting any Pods?

kubectl get pods -n default --show-labels
kubectl get endpoints store -n default
kubectl describe service store -n default

Useful aliases

k is aliased to kubectl and completion is configured. The jumpbox also has k9s, jq and yq.