Kubeadm: kubernetes cluster inited by kubeadm does not start automatically after rebooting os

Created on 12 Jan 2017 · 4Comments · Source: kubernetes/kubeadm

kubeadm: v1.6.0-alpha.0.2074+a092d8e0f95f52
# docker images
REPOSITORY                                               TAG             
gcr.io/google_containers/kube-apiserver-amd64            v1.5.0          
gcr.io/google_containers/kube-controller-manager-amd64   v1.5.0          
gcr.io/google_containers/kube-scheduler-amd64            v1.5.0          
gcr.io/google_containers/etcd-amd64                      3.0.14-kubeadm  
gcr.io/google_containers/kube-discovery-amd64            1.0             
gcr.io/google_containers/pause-amd64                     3.0

and i can not find a subcommand like start to restart the cluster i have initialized by kubeadm.

os : a customized flavor of CentOS 7, 64 bit, but believe me, it is almost the same with CentOS

Source

bruceauyeung

👍4

Most helpful comment

What does journalctl -xeu kubelet say after you've rebooted?
It's the kubelet's task to automatically spin everything up again.
I assume you've run systemctl enable kubelet?

luxas on 13 Jan 2017

👍15

All 4 comments

What does journalctl -xeu kubelet say after you've rebooted?
It's the kubelet's task to automatically spin everything up again.
I assume you've run systemctl enable kubelet?

luxas on 13 Jan 2017

👍15

@luxas i tried again and found it worked as expected.
thanks.

bruceauyeung on 14 Jan 2017

I use sudo swapoff -a when I init k8s with kubeadm, then when I robot k8s hasn't started up.

journalctl -xeu kubelet shows some error when startup.

After sudo swapoff -a , I try systemctl enable kubelet again.

Everything going well now.