Test setting the maintenance mode for the node that is the current active
master manually, then reboot it and ensure that maintenance mode doesn't get
cleared automatically that way.
