I'm running high load jobs on k3s (embedded etcd, single-node cluster), and I get:

$ kubectl describe nodes

  Warning  ContainerGCFailed  51s (x621 over 10h)  kubelet  rpc error: code = ResourceExhausted desc = grpc: trying to send message larger than max (16782381 vs. 16777216)

And somehow, my node automatically gets placed into a

NotReady

state. Has anyone seen this before?

Check the logs. ContainerGCFailed suggests that you are low on disk space and it is failing to prune unused images.

Okay, I checked

systemctl status k3s.service

. I realize this issue is not k3s specific. I likely need better garbage collection.

I would look at the logs not the status

Logs via

journalctl -u k3s.service

?