I am sorry for interrupting. My thoughts: i can be - nfs freeze because of temporary network falure (or maybe nfs locking or something) - something bad about subprocess.call() (btw how child exists? does it fluhes any buffers? does yields? )
if "uWSGI listen queue of socket 3 full !!!" - means no room left in kernel queue for socket to accept connections.. this means main process accepting slower than incoming connections goes in. (btw does it accepts only when there are workers avalable? or it just accepts and then waits untill any worker is ready to feed a job? ) I know it is a bad idea to debug on production server under load.. but maybe "ps axjf" and attaching to deadlocked process with "strace -p PID" will help? 2011/9/14 Szymon Kosok <[email protected]>: > 2011/9/14 Szymon Kosok <[email protected]>: >> Ok. Decreasing to 32... we will see. :) > > No luck. Again. Same problem on both servers. Restart, everything is fine. > > -- > Szymon > _______________________________________________ > uWSGI mailing list > [email protected] > http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi > -- -------------------------------------------- Турнаев Евгений Викторович +7 906 875 09 43 -------------------------------------------- _______________________________________________ uWSGI mailing list [email protected] http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
