docker-vernemq icon indicating copy to clipboard operation
docker-vernemq copied to clipboard

Published 20 hours ago • verified publisher icon vernemq

Fail to start if kubernetes cannot be contacted

Open SerialVelocity opened this issue 5 years ago • 5 comments

Currently, if k8s cannot be contacted (e.g. if DOCKER_VERNEMQ_KUBERNETES_CLUSTER_NAME is set incorrectly), then the service continues to start up and none of the vmq-admin commands work.

SerialVelocity avatar Nov 12 '20 17:11 SerialVelocity

@SerialVelocity operationally though, what should happen in that case?

ioolkos avatar Nov 24 '20 14:11 ioolkos

I would expect the pod to fail to start. If it was a flake, it will restart automatically and succeed eventually.

SerialVelocity avatar Dec 03 '20 16:12 SerialVelocity

@SerialVelocity What exactly do ypu mean with "k8s cannot be contacted"? Do you mean that the K8S API-Server is not reachable?

micw avatar Dec 28 '20 19:12 micw

Yes, if the api-server is not reachable.

SerialVelocity avatar Dec 30 '20 09:12 SerialVelocity

This just happened again. My pods started up and timed out talking to the API server. Now they are operating as independent nodes.

SerialVelocity avatar Feb 19 '21 11:02 SerialVelocity