if you can think of anything that might make it easier for your infrastructure to handle this occasional failure (automatic restarting, or something) please let us know, we would probably want to use it too