We have a clustered DB here. If you run the shutdown script on a node which is down, it helpfully fails over to another node and shuts that down instead.