For the live migration both node need to be alive . :cool:
For example if you want to do a maintenance on on node, while running , you move every VM on another node. And then you can reboot the node.
If you have a crash, there is no way to save the content of the ram. So reboot is the only...