Got the multiple-master-node (k8s terminology) problem solved. Now that etcd is running on all nodes, I can lose any of them and the cluster keeps running. Really I simply had to change a few parameters in the startup script to enable etcd and to start the other nodes in server mode rather than agent mode. (Before I had one node as master, and if it failed the cluster was in bad shape).

#k8s #k3s #highavailability

$ kubectl get nodes -o wide
NAME          STATUS   ROLES                       AGE    VERSION        INTERNAL-IP   EXTERNAL-IP   OS-IMAGE                         KERNEL-VERSION           CONTAINER-RUNTIME
k3s-home-01   Ready    control-plane,etcd,master   195d   v1.29.4+k3s1   10.0.1.234    <none>        Armbian 24.2.1 bookworm          5.10.160-legacy-rk35xx   containerd://1.7.15-k3s1
k3s-home-02   Ready    control-plane,etcd,master   44d    v1.29.4+k3s1   10.0.1.235    <none>        Debian GNU/Linux 12 (bookworm)   6.1.0-18-amd64           containerd://1.7.15-k3s1
k3s-home-03   Ready    control-plane,etcd,master   28m    v1.29.4+k3s1   10.0.1.236    <none>        Debian GNU/Linux 12 (bookworm)   6.1.0-21-amd64           containerd://1.7.15-k3s1