[bclavie]

Sorry about that, I'll dig into the logs -- we had a lot more traffic than we expected overnight so it could be that the backend didn't scale quickly enough and ended up timing out some requests.