This message was deleted.

If you can ssh into the first controlplane node you can probably debug what is going on. https://gist.github.com/superseb/3b78f47989e0dbc1295486c186e944bf rke2-commands are you friend. I usually do the following after SSHing into a misbehaving node:

alias k="sudo /var/lib/rancher/rke2/bin/kubectl --kubeconfig /etc/rancher/rke2/rke2.yaml"

k get pod -A

Most likely it is something wrong with the cloud-provider deployment inside the kube-system namespace If /var/lib/rancher/rke2/bin/kubectl is not even installed yet it is probably a problem with the rancher-system-agent.service that downloads the rke2 stuff from rancher

sudo systemctl status rancher-system-agent.service
sudo journalctl -u rancher-system-agent.service -f

should help

Looks broken even before that service appears.