Jan Christoph Ebersbach

@milosmns thank you for raising the issue. I'm also experiencing these issues and haven't found a solution, yet. I'll get back to you in the next days. A workaround I'm...

I noticed in the k3s' logs the presence of this line: ``` Mar 25 10:00:35 muses-dev-system-2 k3s[100135]: {"level":"warn","ts":"2025-03-25T10:00:35.917959+0100","caller":"rafthttp/probing_status.go:68","msg":"prober detected unhealthy status","round-tripper-name":"ROUND_TRIPPER_RAFT_MESSAGE","remote-peer-id":"23ff04bc7d082ddd","rtt":"13.233934ms","error":"dial tcp 10.0.1.4:2380: i/o timeout"} ``` From this error onward,...

@mysticaltech @vitobotta have you experienced anything like this before with k3s on Hetzner servers? One specialty of this k3s setup is that it only uses Hetzner's internal network.

@vitobotta thank you for your quick reply!

@milosmns no, SSH won't be affected by this. If SSH isn't reachable, I recommend you reach out to the support before you make any additional changes to the systems.

I just ran into the issue again and found that the default route disappeared from the affected server. Weirdly, the IP address remains in place: 

The issue might be related to https://github.com/systemd/systemd/issues/28358 since we're also using systemd-networkd and I guess Hetzner migrates virtual machines at will from one system to the next. See also the...

As a workaround, I'll add and install this script (https://github.com/systemd/systemd/issues/28358#issuecomment-1909985912) on my servers that will run `networkctl reconfigure enp7s0` after a system resumes after a suspend. Let's see if it...

Brief status update: apparently, Hetzner runs dhcpd automatically on all nodes. This might be another source for errors. To disable dhcpd these steps need to be performed. It can be...

Status update: so far, the cluster nodes have been stable. The suspend script hasn't been triggered, yet. It looks like the deactivation of dhcpd and the sole use of systemd-networkd...