Cluster destruction

We use the Alpine k3s package, rather than curlbash’ing the installer. This means that we don’t get a nice k3s-uninstall.sh script for us. We do have a k3s-killall.sh which we have adapted from upstream.
See the uninstall script code for more context.

Make an uninstall script like this and run it:

# Stop all containers and the k3s service itself
# Note: This does unmount /var/lib/rancher/k3s; we have to remount it below
k3s-killall.sh

rc-service containerd stop
rc-service iscsid stop

# This is also necessary for some reason?
killall containerd-shim-runc-v2 traefik longhorn longhorn-manager kube-vip runsvdir pause sleep tini livenessprobe csi-node-driver-registrar cephcsi entry

# Make sure all relevant longhorn/ceph processes are killed too - might be some you have to clean up manually
ps aux | grep longhorn
ps aux | grep ceph

# Unmount a bunch of bullshit that Docker mounts
umount /run/netns/* /var/lib/kubelet/pods/*/volumes/*/*

# Remove old cluster config/state.
# Do NOT remove /var/lib/rancher/k3s because that's an overlay mountpoint on psyopsOS
rm -rf /etc/rancher/k3s /psyopsos-data/overlays/var-lib-rancher-k3s/*

# Recommended in the kube-vip k3s documentation https://kube-vip.io/docs/usage/k3s/
ip addr flush dev lo
ip addr add 127.0.0.1/8 dev lo

# Make sure that /var/lib/rancher/k3s is mounted
mount -o bind /psyopsos-data/overlays/var-lib-rancher-k3s /var/lib/rancher/k3s