[markbnj]

> For a few hours, builds ran smoothly. As expected, eventually the agent died during a build. Builds sat queued, my frustration grew, and AWS notified my instance needed a doctor. After an hour I gave up hope for the unresponsive server and commanded it to reboot.

Sounds a lot like some process on the build server is leaking memory. We have a couple of older services that have this problem, and if we don't catch it early enough the server will become unresponsive and unable to even handle ssh connections.