Il giorno 28/dic/2012, alle ore 11:27, André Cruz <[email protected]> ha scritto:
> On Dec 28, 2012, at 10:05 AM, Łukasz Mierzwa <[email protected]> wrote: > >> Fri Dec 28 10:02:16 2012 - ...The work of process 12260 is done. Seeya! >> Fri Dec 28 10:02:16 2012 - Respawned uWSGI worker 1 (new pid: 14695) > > Ah, yes, I have found those messages. > > Dec 28 10:11:05 disco-block01 discoapi: ...The work of process 13222 is done. > Seeya! > Dec 28 10:11:05 disco-block01 discoapi: Gracefully killing worker 17 (pid: > 13222)... > Dec 28 10:11:05 disco-block01 discoapi: stopping gevent signals watchers for > worker 17 (pid: 13222)... > Dec 28 10:11:05 disco-block01 discoapi: stopping gevent sockets watchers for > worker 17 (pid: 13222)... > Dec 28 10:11:05 disco-block01 discoapi: main gevent watchers stopped for > worker 17 (pid: 13222)... > > > But it seems to me, looking at preceding log messages, that some requests > were not completed. Does anyone know if that's the case? > If you have created additional greenlets in your requests they will be brutally destroyed as the system has no way to know how to manage them. You can try checking if atexit hooks (i mean the python atexit module) are called on worker destroy. If It is the case (sorry no way to test it now), you may easily wait for "still running" greenlets (even if it could continue to increase memory usage...) -- Roberto De Ioris http://unbit.it JID: [email protected] _______________________________________________ uWSGI mailing list [email protected] http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
