I do not use nginx but in general, this kind of problem comes when upstream terminates the request before it has been handled correctly. I see that you use worker-reload-mercy which will trigger such events if you reload while a request is being processed and processing time >= 60s.
I hope this helps, Mathieu Le lun. 27 avr. 2020 à 15:18, Sandeep Balagopal < sandeepbalagopa...@gmail.com> a écrit : > I have a Django project configured with uwsgi and Nginx. The issue is that > I am getting lot of 502 Bad gateway. The nginx error log says (104: > Connection reset by peer) while reading response header from upstream. > Looks like uwsgi is not able to give response for some of the requests. > Below is my uwsgi configuration. Could the issue be due to the bad uwsgi > configuration ? I have an ec2 t3.medium instance it has two vCPUs. These > 502 gateways increase when the site load increases. Sometimes the processor > is 100%, that is when the we get more of this error. I am thinking of > upgrading the server to t3.xlarge which has 4 vCPUs. But how to make sure > the issue is with the server performance ? Please check the uwsgi > configuration and see if there is any issue with the configuration. > > [uwsgi] > master = true > socket = /tmp/uwsgi.sock > chmod-socket = 666 > chdir = <dir_path> > wsgi-file = <wsgi.py path> > processes = 16 > threads = 8 > cpu-affinity = 1 > max-worker-lifetime = 3600 > max-requests = 1000 > reload-on-rss = 2048 > worker-reload-mercy = 60 > virtualenv = <ven_path> > vacuum = true > enable-threads = true > daemonize= <log_path> > stats= <stats_path> > buffer-size = 65535 > > _______________________________________________ > uWSGI mailing list > uWSGI@lists.unbit.it > http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi > -- Mathieu Lacage <mathieu.lac...@alcmeon.com>
_______________________________________________ uWSGI mailing list uWSGI@lists.unbit.it http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
