[uWSGI] restarting uwsgi with HUP occasionally gives bad fd

Tim Tisdall tisdall at gmail.com
Wed Apr 16 17:42:16 CEST 2014


It is used in the master too, so I guess that must be it.  I'll see what I
can do about it, thanks!


On Wed, Apr 16, 2014 at 10:56 AM, Roberto De Ioris <roberto at unbit.it> wrote:

>
> > Which config?  The INI config file I use to start up uwsgi?  Here's that
> > config:
> >
> > ------------------
> > [uwsgi]
> > # %d - the absolute path of the directory containing the config file
> > socket = /var/run/uwsgi/metrics_dev.sock
> > master = true
> > processes = 2
> > threads = 20
> > harakiri = 60
> > harakiri-verbose = true
> > limit-post = 65536
> > post-buffering = 8192
> > daemonize = /var/log/uwsgi/metrics_dev.log
> > pidfile = /var/run/uwsgi/metrics_dev.pid
> > max-requests = 1000
> > log-slow = true
> > virtualenv = %denv/
> > uid=33
> > gid=33
> > vacuum=False
> > enable-threads=True
> > attach-daemon = %denv/bin/python %dmainserver/msg_server.py
> > %ddevelopment.ini
> > attach-daemon = %denv/bin/python %dmainserver/alert_generator.py
> > %ddevelopment.ini
> > -----------------
> >
> > I am using zeromq, but not as the interface between uwsgi and nginx, just
> > for internal process use.  If it were that, I'd think that uwsgi would
> > still be able to recover and restart as expected.
> >
> >
>
> if zeromq is initialized in the master (and it crashes), the master will
> die too. This happens because zeromq generates a posix thread.
>
>
> Check where you initialize zeromq
>
> --
> Roberto De Ioris
> http://unbit.it
> _______________________________________________
> uWSGI mailing list
> uWSGI at lists.unbit.it
> http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.unbit.it/pipermail/uwsgi/attachments/20140416/ccc0899f/attachment.html>


More information about the uWSGI mailing list