Can anyone confirm: can I update k3s and restart the daemon without draining the entire node? I see things in the docs like "To allow high availability during upgrades, the K3s containers continue running when the K3s service is stopped." But it never explicitly says that's an acceptable way to upgrade. I mean, to be fair, there is https://docs.k3s.io/upgrades/manual#upgrade-k3s-using-the-binary that says: 1. Download k3s binary 2. Copy it to the right place 3. Restart the process So maybe I just want someone to explicitly confirm that a drain is unnecessary just for my peace of mind... ?

Yes, like the docs say - just replace the binary and restart the service.

Yeah, that lines up with what I have experienced elsewhere

if your pods can’t tolerate a short apiserver outage then you might want to drain. but usually things don’t care.

Good point, I could imagine a particular class of application caring, but we don't have anything quite so complex

containers keep running, container networking stays as it was last configured, and so on.

Excellent

sure if you think thats necessary.

I think it would be helpful for folks like me coming from the various hosted k8s offerings that don't really have this level of insight