Reputation: 11
Good day.
I'm new in kubernetes.
I have a kubernetes cluster with 3 hosts on Ubuntu 16.04. Recently I noticed that this hosts doesn't have space in /boot partition.
I used apt-get autoremove to free some space in /boot on first host. And then apt-get upgrade.
I use Rancher (as docker container) to visualize my cluster. So now when I start Rancher it's always Waiting for Kubernetes API.
Tried to restart Rancher container and no luck. Still Waiting for Kubernetes API.
I always found that there is no kube-system containers on first host.
Is there a way to get Rancher back to life?
Upvotes: 1
Views: 5673
Reputation: 1
In my case I just update the cluster by rancher and update the number of nodes to two then the cluster is updated and running
Upvotes: 0
Reputation: 1960
Did you get a chance to make some free space in /boot
? It's also possible etcd is not healthy since it needs a quorum of odd number of nodes. You could try deactivating two of the hosts, get the kubernetes stack to become healthy and then add the hosts back with enough disk space. It might also be a good idea to create a github issue with logs from etcd and other core k8s components.
Upvotes: 0