> Old message: > http://lists.unbit.it/pipermail/uwsgi/2012-October/004890.html > > But the configs are a bit different - in my case, it's a pretty simple one > without fastrouter, just serving a single site on a dedicated tcp socket, > started via emperor. The error message is the same tho, we get tons of > "upstream > prematurely closed connection while reading response header from > upstream", > as if the socket is improperly closed / processed killed, and the OS sends > a RST packet to nginx. > > > I haven't had the time yet to properly reproduce it and create a simple > test case and try to hack uwsgi myself a bit, so that's why I haven't > asked > for anything else. > > I've already found the relevant section, and at the first glance, it looks > fine - serve request, check signal, if should stop, die. We don't even run > with threads so that section is super simple. My first guess would be that > the "die" part may need either a sleep or a tiny flush and close socket > block added, but that experiment has to wait until my todo list is a bit > shorter. > > > Kind regards, > > Chi Ho Kwok >
Are you sure you are not using some old uWSGI version ? What you are trying to accomplish works by about 3 years without emperor, and from 1 year with the Emperor. -- Roberto De Ioris http://unbit.it _______________________________________________ uWSGI mailing list [email protected] http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
