I work for a company that does network monitoring. Most of the devs run only monitoring a handful of elements, just enough to test what they're working on at the time. I decided that I was going to make my system run with more elements in an attempt to catch scale problems before they start. If I've been working on the backup code, I'll make sure I've got a job running even after I've tested and checked in to make sure that it is actually working as designed over a longer haul.

I've already found a few bugs this way. Keep testing after you're done testing is my point, I guess. :)