A couple of years ago a friend of mine asked me for an advice. He has just started to work as a sysadmin at some middle-sized bank. They have had a lot of *NIX servers that were working just fine. But there was also one Windows NT-based server there with a *very* special application which was absolutely crucial for bank’s business processes. The very special “feature” of this application was that it crashed quite often and took the whole server to the realm of BSODs with every crash. The old sysadmin was working in the server room and has had no problem resetting the server manually if needed. My friend, however, preferred remote administration from his sunny office (yea, i know, how weird it sounds

:) ) so walking down to server room in order to restart a server was hardly an option for him.
Fixing a program was not possible – no source code was available.

After a weirdest brainstorming i have ever participated at, he finally found a solution – he has built together a crappy PC with linux 2.4 on board and connected it to the server via a crossover cable.
The sole task of this PC was to ping the alleged server and if it wasn’t responding – eject /dev/hdc.
The cd-rom drive opened itself and pressed the Reset button on the server. Fool-proofed system and the funniest linux-based solution to solve windows proglems I’ve seen in my life.

from /.