This message was deleted.

I'm not sure about the issue you had adding 2 nodes to a 1 node cluster, but I can tell you that I've got a 3 node cluster here, and I just shut one node down, and while the volume backing a VM shows as degraded in the Longhorn UI, the VM kept running just fine.

Rebuilding the volumes could saturate the "pipes" between your Harvester nodes, causing the VMs' writes to timeout and when that happens, the domain pauses, and KubeVirt tries to unpause, and then they cycle into the same thing again. If your disks are too slow, or your NICs and switches are too slow, the thundering herd can easily cause this.

Thank you!

Make sure to check your switchports for proper configuration and packet drops on your management network.

I agree that this is a best pracitce! Thank you!