No, the thread started with Ie8 being suspects, but at least from my experiments it is a problem in Rocket which can be triggered with any browser or even without a browser (e.g. wget/curl instead of a browser)
See e.g. https://github.com/explorigin/Rocket/issues/1#issuecomment-3734231 The reason it works locally for you and not in production is probably because you have lower bandwidth to the production machine, triggering timeouts. If changing SOCKET_TIMEOUT to 10 in rocket.py makes the problem go away for you (or at least much less frequent), then this is your problem too. (This is not really a solution -- it doesn't address the cause, and it interferes with timely detection of problems -- but it will let you know if you are experiencing the same problem or a different one)
