Il giorno 14/ott/2011, alle ore 10:11, Ryan Showalter ha scritto: > Hey everyone, > > I've been running into an issue lately where after my server has been > running for a few hours, the uWSGI master process will start killing > off workers for seemingly no reason. When this happens, spool jobs > also fail to be created, but no failures seem to occur in the logs. > What's weird is that this error seems to have only begun manifesting > itself in the last couple of weeks. I'm not sure if this is related > to increased traffic or what. My memory usage is not outrageous, I > still have plenty of physical memory and swap available. Also, CPU > usage is very minimal. Restarting uwsgi will fix the issue for a few > hours before it begins to creep up again, at first just one or two > workers will be killed off every few minutes, and eventually it gets > to the point where at least 1 worker is killed at every request. > > Below is my configuration, along with an excerpt from my log file > which shows several workers being killed and respawned within less > than a minute. I'm running uWSGI 0.9.9.2 with debug enabled (this > issue also happens with debug disabled). > > Any help is much appreciated! > > [program:uwsgi-www] > command=/opt/webapps/www.example.com/bin/uwsgi > --master > --processes 16 > --memory-report > --harakiri 7200 > --vacuum > --max-requests 500
Can you try removing --max-requests directive ? You should check in the logs if the continuous reloads starts after the max requests count has been reached > --optimize 2 remove even optimizations (just to be sure) -- Roberto De Ioris http://unbit.it JID: [email protected] _______________________________________________ uWSGI mailing list [email protected] http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
