calico-policy-container on the worker node is on a restart loop. how can i check why?

Question

I have two coreos stable machines (with latest stable version installed) to test Kubernetes. i installed kubernetes 1.5.1 using the script from https://github.com/coreos/coreos-kubernetes/tree/master/multi-node/generic and patched it with https://github.com/kfirufk/coreos-kubernetes-multi-node-generic-install-script.

I installed controller script on one and worker script on the other. kubectl get nodes shows both servers.

kubectl get pods --namespace=kube-system shows that calico-policy-controller-2j5dn restarts a lot. in the worker server I do see that calico-policy-controller restarts a lot. any idea how to investigate this issue further?

how can I check why it restarts? are there any logs for this container?

Answer 1

kubectl logs --previous $id —namespace=kube-system

i added --previous because when the controller restart it has a different random characters appended to it.

in my case that kube-policy-controller what started on one server, and requested the etcd2 certificates that where generated on a different server.