This message was deleted.

That’s weird - usually that would mean a bad network disconnect between nodes. Is there a firewall or different VLAN port connectivity on the ports? Maybe one of your bond interface ports doesn’t have the same config ? Regardless, I’ve never seen this and have never needed a reset either. Can you ensure their time server is reachable, the network between them is correctly configured and power off and back on the entire cluster?

Very weird considering the home loopback port failures. Powercycle would be my go-to move too. I've seen the kernel get into really weird states before (not just in Harvester) where basic modules stop functioning. Ceph, then encryption failures, etc., and a power cycle on the box normally fixes it up.