This message was deleted.

There's a known bug that the hosts config is lost on docker host restarts as K3s rewrites the Core DNS config on every start. So if it's not a fresh cluster this might be it.

k3d cluster stop

and then

k3d cluster start

may help.

Sorry for resurrecting the thread; but could this be why when I create VMs out of an AMI one of my pods is in an infinite

CrashLoopBackoff

? It does not happen all the time I say 1 in 5 VMs works fine. When I do a stop/start, everything looks good. Do you have any links on this "known bug"?