I have deployed a node js Application on K8 1.16 version.I notice that post deployment the backend pods are not registering the end points , The back end pods thus keep restarting and go into CrashLoopBackOff.
kubectl describe svc
Name: backend-xx-backend-svc
Namespace: default
Labels: <none>
Annotations: <none>
Selector: chart=backend-xx-backend,tier=backend
Type: ClusterIP
IP: 192.168.246.12
Port: <unset> 80/TCP
TargetPort: 8800/TCP
Endpoints:
Session Affinity: None
Events: <none>Any suggestions as to why the backend pod end points are blank.
As per result of describe command mentioned in the comment, it looks like Readiness probe is failing. Unless a pod is in Ready state, k8s won't forward traffic to that pod; so may be that's why end point in the Service object is blank as none of the pod is in Ready state. Check why Readiness probe i.e. Get 10.39.67.76:8800/api/health is failing. If your app takes time at the start then increase initialDelaySeconds or failureThreshold of Readiness Probe configuration.