[QUESTION/HELP] Is there a way to have the HOSTNAME variable in k3d pods to match the node names if changed using k3s's "--node-name"

[abullock66]

Question / Where do you need Help?

I was using the following yaml config to create a simple cluster and pass in the KUBECONFIG environment variable to remove the following messages in the docker logs

E0722 16:11:57.389363  143578 memcache.go:265] couldn't get current server API group list: Get "http://localhost:8080/api?timeout=32s": dial tcp [::1]:8080: connect: connection refused
E0722 16:11:57.389756  143578 memcache.go:265] couldn't get current server API group list: Get "http://localhost:8080/api?timeout=32s": dial tcp [::1]:8080: connect: connection refused
E0722 16:11:57.391202  143578 memcache.go:265] couldn't get current server API group list: Get "http://localhost:8080/api?timeout=32s": dial tcp [::1]:8080: connect: connection refused
E0722 16:11:57.391531  143578 memcache.go:265] couldn't get current server API group list: Get "http://localhost:8080/api?timeout=32s": dial tcp [::1]:8080: connect: connection refused
The connection to the server localhost:8080 was refused - did you specify the right host or port?

This error stems from the following line in the /bin/k3d-entrypoint.sh file

until kubectl uncordon "$HOSTNAME"; do sleep 3; done

My config.yaml

kind: Simple
apiVersion: k3d.io/v1alpha5
metadata:
  name: nodenames
servers: 1
agents: 1
kubeAPI:
  hostPort: "6443"
  host: "xxx.xxx.xxx.xxx"
network: k3d-edu
env:
  - envVar: KUBECONFIG=/var/lib/rancher/k3s/agent/k3scontroller.kubeconfig
    nodeFilters:
      - agent:*
options:
  k3d:
    wait: true
    disableLoadbalancer: false
    disableImageVolume: false
    disableRollback: false
    loadbalancer:
      configOverrides:
        - settings.workerConnections=2048
  k3s:
    extraArgs:
    - arg: --node-name=server-0
      nodeFilters:
      - server:0
    - arg: --node-name=node-0
      nodeFilters:
      - agent:0
  kubeconfig:
    updateDefaultKubeconfig: true
    switchCurrentContext: true
  runtime: {}

It seems that if the k3s --node-name arg is used, the HOSTNAME environment variable is still set to the default node name in which case it is unknown

root@server:~# docker exec -it k3d-nodenames-agent-0 sh
~ # hostname
k3d-nodenames-agent-0
~ # kubectl get nodes
NAME       STATUS   ROLES                  AGE   VERSION
node-0     Ready    <none>                 72s   v1.31.5+k3s1
server-0   Ready    control-plane,master   80s   v1.31.5+k3s1
~ # kubectl uncordon k3d-nodenames-agent-0
Error from server (NotFound): nodes "k3d-nodenames-agent-0" not found
~ # hostname
k3d-nodenames-agent-0
~ # kubectl uncordon $HOSTNAME
Error from server (NotFound): nodes "k3d-nodenames-agent-0" not found

Removing the k3s arg --node-name works

root@server:~# docker exec -it k3d-nonodenames-agent-0 sh
~ # kubectl get nodes
NAME                       STATUS   ROLES                  AGE   VERSION
k3d-nonodenames-agent-0    Ready    <none>                 53m   v1.31.5+k3s1
k3d-nonodenames-server-0   Ready    control-plane,master   53m   v1.31.5+k3s1
~ # until kubectl uncordon "$HOSTNAME"; do sleep 3; done
node/k3d-nonodenames-agent-0 already uncordoned

Is there a way to have the node names configured in the k3s Args to be used as the hostname ?

Thanks!