Setting up a one node Kubernetes cluster - internal IP address and port closed after few minutes

bauger

I made several attempts to complete all the test tools; unfortunately after few minutes of use (I was able to bootstrap the one node Kubernetes cluster and set up the networking plugin) I am no longer able to continue the exercise as the kubectl command returns an error with the API endpoint connection and asking if I provided the correct IP address and port. This is strange as it was working just few second ago and this is a sudden change not related to any configuration. I wonder if Google prevents us to use an internal IP address for an application or service without declaring it first from one of their tools...

gouravshah

If Google is blocking the IP, it should happen always imho. If it works for some time, stops, and may be starts working again, it could also be a problem with one of the components.

One possibility is some component of control plane crashing. Have you tried checking if all the components are okay, and there are no restarts etc? You could also further check the logs of API servers and other control plane components to verify.

bauger

@gouravshah

I do not remember any component troubles and unfortunately I have already decommissioned the VM instance and can't check it out.

Thanks for your support.

joyepreach

i had to install the kubeadm and other components manually because i noticed they were failing from the script. It worked after that