Create cluster got pending

[seikyo-cho-lvgs]

I use k3d create a cluster but pending for a long time.

$ wget -q -O - https://raw.githubusercontent.com/k3d-io/k3d/main/install.sh | bash
$ k3d cluster create mycluster
$ kubectl get pods -A
NAMESPACE     NAME                                      READY   STATUS    RESTARTS   AGE
kube-system   coredns-576bfc4dc7-tl66t                  0/1     Pending   0          105m
kube-system   helm-install-traefik-crd-22kkl            0/1     Pending   0          105m
kube-system   helm-install-traefik-zh46f                0/1     Pending   0          105m
kube-system   local-path-provisioner-6795b5f9d8-8dzg8   0/1     Pending   0          105m
kube-system   metrics-server-557ff575fb-cmrwz           0/1     Pending   0          105m

I can't see any logs in these pods.

But nodes seem been created.

$ k3d node list
NAME                     ROLE           CLUSTER     STATUS
k3d-mycluster-server-0   server         mycluster   running
k3d-mycluster-serverlb   loadbalancer   mycluster   running

What happened?

[iwilltry42]

Can you kubectl describe one of the pods?

[iwilltry42]

Oh, maybe rather describe a node, as the pods are still pending - my mistake

[seikyo-cho-lvgs]

@iwilltry42

$ kubectl get nodes
No resources found
$ k3d node list
NAME                     ROLE           CLUSTER     STATUS
k3d-mycluster-server-0   server         mycluster   running
k3d-mycluster-serverlb   loadbalancer   mycluster   running
$ kubectl describe node k3d-mycluster-server-0
Error from server (NotFound): nodes "k3d-mycluster-server-0" not found
$ kubectl describe node k3d-mycluster-serverlb
Error from server (NotFound): nodes "k3d-mycluster-serverlb" not found

There isn't kubernetes node been created. What's k3d node?

[seikyo-cho-lvgs]

@iwilltry42
After I restarted rancher-desktop, the k3d pods had been created and the cluster works well.
Does it conflict with rancher-desktop?

[iwilltry42]

It should not conflict with rancher-desktop, as it's running completely in Docker.
However, IIRC rancher-desktop ships with kubectl, so maybe it just didn't correctly point to the k3d cluster?
Or there are some network level conflicts ..
I don't know for sure, as I never used it with RD and this is the first time something like this was reported.
However, I know that some people used k3d with RD's docker engine, so I would be surprised if there was a hard conflict 🤔