Issue with coredns pods after initial cluster setup

gskornowicz

Hi. I'm stuck at below error on coredns (ContainerCreating) pods after initial cluster setup:

Warning  FailedCreatePodSandBox  6m52s                   kubelet            Failed to create pod sandbox: rpc error: code = Unknown desc = failed to setup network for sandbox "7d0accc9987b4005f2188857ae1a1d5953a73073e823d05f5c0e41480c7069c7": plugin type="cilium-cni" failed (add): failed to find plugin "cilium-cni" in path [/usr/lib/cni]

I've followed the advice in the course but deleting the coredns pods didn't helped

root@k8s-control-plane-1:~# kubeadm version
kubeadm version: &version.Info{Major:"1", Minor:"34", EmulationMajor:"", EmulationMinor:"", MinCompatibilityMajor:"", MinCompatibilityMinor:"", GitVersion:"v1.34.1", GitCommit:"93248f9ae092f571eb870b7664c534bfc7d00f03", GitTreeState:"clean", BuildDate:"2025-09-09T19:43:15Z", GoVersion:"go1.24.6", Compiler:"gc", Platform:"linux/amd64"}
root@k8s-control-plane-1:~# kubectl version
Client Version: v1.34.1
Kustomize Version: v5.7.1
Server Version: v1.34.0

chrispokorni

Hi @gskornowicz,

I would recommend following the lab guide, especially the versions presented in the early exercises to bootstrap the cluster.

If the issues persist, please provide details about your cluster VMs:

• cloud (which provide) or local (which hypervisor, which arch: intel/amd or arm),
• size (CPU, RAM, vDisk (dynamic allocation or full pre-allocation) ),
• count of network interfaces, type of private network (host, nat, bridged),
• guest OS,
• private IP address,
• firewalls.

Besides the installed versions, have you deviated from the installation steps in any other way?

Regards,
-Chris

gskornowicz

Hi Chris! Nope, only derivations are the newest possible versions. I’ll give it a try with the versions mentioned in the lab steps. Thanks!

Best regards,
Greg

gskornowicz

@chrispokorni unfortunately still the same issue.

  Normal   Scheduled               2m29s                  default-scheduler  Successfully assigned kube-system/coredns-668d6bf9bc-8zc7d to k8s-worker-1
  Warning  FailedCreatePodSandBox  2m29s                  kubelet            Failed to create pod sandbox: rpc error: code = Unknown desc = failed to setup network for sandbox "988ebdf4cd5b8b6f965d927869c6484c5c5d394b922ba421c0d8af8b481c4e4d": plugin type="cilium-cni" failed (add): failed to find plugin "cilium-cni" in path [/usr/lib/cni]

root@k8s-control-plane-1:~# kubeadm version
kubeadm version: &version.Info{Major:"1", Minor:"32", GitVersion:"v1.32.1", GitCommit:"e9c9be4007d1664e68796af02b8978640d2c1b26", GitTreeState:"clean", BuildDate:"2025-01-15T14:39:14Z", GoVersion:"go1.23.4", Compiler:"gc", Platform:"linux/amd64"}
root@k8s-control-plane-1:~# kubectl version
Client Version: v1.32.1
Kustomize Version: v5.5.0
Server Version: v1.32.1

I'm running it on Debian 13 Trixie.

Indeed the path /usr/lib/cni is empty. I will debug this further. Maybe I've missed something

chrispokorni

Hi @gskornowicz,

The lab guide recommends Ubuntu 24.04 LTS (amd64) as the guest OS. Not sure though if it was tested on other distributions.

Regards,
-Chris

gskornowicz

Thanks Chris. I was able to resolve the issue by adding --set cni.binPath=/usr/lib/cni to the installation. Works with cilcium-cli install and helm installation. Moving forward!

chrispokorni

Hi @gskornowicz,

We will revisit the installation method from the lab guide to see if the behavior can be reproduced and if such a flag (or perhaps a manifest edit) is necessary as a fix.

Regards,
-Chris