(edges away from petemcevoy) I hope I don't get trashed like that! Hah!
Before I proceed with some of my more glorious screw-ups, my mentor for unix told me once:
"There are two kinds of sysadmins: those that have completely ****ED a system up, and those that will."
That said, onward we go!
1: deleted the font file for solaris 2.51. If you're booting into anything other than single-user mode, it won't boot. Period.
2: destroyed the swap file by accidentally dd'ing it to /dev/null (this is what happens when you don't pay attention to what you're typing while talking to someone)
3: set the first tape device as the swap device...mainly because I wanted to see how long it would take for a box to boot...a loooooooong time.
4: did 'cleanup' on the DNS server and unfortunately blew out all the zone files because bind hadn't updated in over a month due to a previous change BY AN NT ADMIN WHO HAD ROOT ACCESS FOR SOME REASON (don't get me started on what I was yelling that day) that didn't work, go figure.
5: powered off an N-class HP 9000 server without knowing that Oracle was in hot-backup mode with Omniback...the DBAs didn't like me very well that night.
6: accidentally zero-lengthed the kernel for an HP box (I still don't know how I did this?)
I think those are some of my more glorious screwups that halted business to some level. Windows problems, while I've had many, don't really count because everyone knows there's two methods to fixing Windows: reboot or reinstall.