Assuming a worker node dies and needs to be replaced, what’s the correct procedure to follow?
I’ve tried to rebuild a worker node and re-run microk8s join afterwards (without touching anything on the master), and things seems to work automagically.
Is this all there is to it or am I missing some important details?
I’ve only run single node microk8s, but usually on multi-node clusters there’s no action required on a control plane node, the worker simple registers itself in to the kube API and the scheduler will start assigning new pods to it.
Yes, that is when the worker is first created and joined. Here I’m specifically asking for the case of an existing worker node being physically replaced (due to hw failure etc).