server has 512Mb, and only runs nginx and mercurial, which should be enough.
> In addition to this i prefer to run mercurial over threads (not processes) as it is I/O bound. If I wish to change this, what do I need to change? On Sun, Apr 14, 2013 at 4:51 PM, Roberto De Ioris <[email protected]> wrote: > > > I've been using on nginx+uwsgi without issue and now its started to kill > > my > > vps container, by pushing the cpu to 100% and eventually exhausting ram, > > forcing a reboot. And it doesn't last long either between 5-30mins before > > uwsgi is pushing 90-100% CPU > > > > nginx seems fine, and the site isn't that popular (using > > mercurial+nginx/uwsgi) > > > > exec /root/uwsgi-1.9.4/uwsgi --master --enable-threads --idle 30 --socket > > /var/lib/mercurial-server/repos/hg.sock --pidfile /var/run/hgweb.pid > > --disable-logging --daemonize /var/log/uwsgi.log --chdir > > /var/lib/mercurial-server/repos --pp -M -p 4 -w hgweb > > > > Is there something I can do to improve ram/cpu constraints of uwsgi? > > > > Log: http://pastebin.com/AkehgyFe > > _______________________________________________ > > uWSGI mailing list > > [email protected] > > http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi > > > > How much memory your server has ? > > I see a lot of disconnection errors in the logs, i suppose this is because > the server is swapping and so each operation is extremely slow. > > You should add --memory-report to get the amount of memory used after each > request. > > In addition to this i prefer to run mercurial over threads (not processes) > as it is I/O bound. > > (note: --pp takes an argument, otherwise remove it) > > -- > Roberto De Ioris > http://unbit.it > _______________________________________________ > uWSGI mailing list > [email protected] > http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi >
_______________________________________________ uWSGI mailing list [email protected] http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
