Search code examples
kubernetesupgradekubectlkubeadm

"kubeadm upgrade apply v1.18.0-00" does not seem to upgrade the master node

I'm quite confused here on this kubeadm upgrade.

As you can see from my screenclips below, my serverVersion is stuck at v1.17.2 when displaying kubectl version -o json from my Mac.

When I'm on my MASTER node, I issued the kubeadm upgrade plan and the system displays the kubeadm upgrade apply v1.18.0 to upgrade.

But, after the upgrade, my CURRENT version is still showing v1.17.2.

You can see from the 3rd screenclip with kubectl get nodes -o wide that the nodes are all showing v1.18.0.

But, kubectl version -o json, still shows the serverVersion at v1.17.2.

Also, my k9s continues to show my K8s Rev: v1.17.2

Adding to the question.

Why on the MASTER node, when entering kubectl version would the Client & Server versions be DIFFERENT? Which is shown in the 2nd to last screen clip below.

Isn't the kubectl on the MASTER node both the CLIENT & SERVER? I mean if I am ssh'd into the MASTER node, and I use kubectl, am I not issuing a CLIENT request ON the SERVER itself? So, how can the CLIENT & SERVER versions be different?

Can someone tell me what I am doing wrong?

Thanks in advance. sc1 sc2 sc3 sc4 sc5 sc6

Solution

  • The problem was that emacs had created its normal backup files in the /etc/kubernetes/manifests files, and this was creating the problem. I'm not sure WHY kubectl had an issue with these backup files, but after I removed these backups from the directory, you can see my kubectl version for the Client & Server versions match.

    pic1 pic3 pic2

    Thanks Nick!