Fresh microk8s v1.24 installation have failing calico-kube-controller pod

Summary

On a fresh microk8s-v1.24.4 installation on ubuntu-server-22.04, I see that calico-kube-controllers deployment’s pod is failing from the start, causing it to be in “CrashloopBackOff” state. Here are the list of pods in the system:

NAMESPACE     NAME                                       READY   STATUS             RESTARTS        AGE
kube-system   calico-node-knlnn                          1/1     Running            2 (35m ago)     57m
ingress       nginx-ingress-microk8s-controller-clx2q    1/1     Running            0               27m
kube-system   calico-kube-controllers-64d4cb6ccb-26krj   0/1     CrashLoopBackOff   7 (4m38s ago)   15m

Here is the log-output from the calico-kube-controllers pod:

standard_init_linux.go:228: exec user process caused: exec format error

What Should Happen Instead?

The calico, and all other kube-system pods should all be running properly

Reproduction Steps

1. Install Ubuntu-server 22.04 on VirtualBox Version 6.1.38 r153438 (Qt5.12.8)
2. While installing ubuntu-server, enable microk8s installation via cloudinit, and select stable/1.24 channel to install kubernetes v1.24.4
3. Once installed, login to the ubuntu server, and run microk8s kubectl commands to check k8s installation

Introspection Report

inspection-report-20220914_112938.tar.gz

Can you suggest a fix?

Are you interested in contributing with a fix?

[dnewsholme]

I ran into this after upgraded to 22.04.1 LTS. My microk8s snap also upgraded to 1.25, when trying to rollback to 1.24 I got this issue because snap doesn’t clear up the calico virtual network devices on removal.

[tapanhalani]

I have successfully tried installing microk8s multiple times with multiple fresh virtualbox machines. And i cannot reproduce this issue again.

Closing this issue now. Thank you @neoaggelos for attending to this.