We've been experiencing random server crashes as of late but I'm happy to say I think we've got it under control. There is a bug somewhere in the web server code that is causing connections to timeout. The symptom seems to be that at some point in time the web server stops sending responses back to your browsers. Once it gets stuck in this loop it requires a restart to clear it.

While I don't know what is causing the problem exactly nor how to actually fix it I think we've got it isolated to the point where we can work around the problem reliably. We know how to identify this condition and have a script which monitors the server and if/when it happens, restarts it.

The net result is that you might bump into it now and then but it should clear itself in quick fashion without one of us noticing it and restarting by hand.

I guess you can call it good news