Hello, This looks weird isn't it ? Here it is the uwsgi.log : *** HARAKIRI ON WORKER 4 (pid: 3131) *** HARAKIRI: --- uWSGI worker 4 (pid: 3131) WAS managing request / since Thu Mar 3 01:42:37 2011 --- DAMN ! process 3131 died :( trying respawn ... Respawned uWSGI worker (new pid: 3216)
*** HARAKIRI ON WORKER 1 (pid: 3128) *** HARAKIRI: --- uWSGI worker 1 (pid: 3128) WAS managing request / since Thu Mar 3 01:44:30 2011 --- DAMN ! process 3128 died :( trying respawn ... Respawned uWSGI worker (new pid: 3219) Regards, --yml On Thu, Mar 3, 2011 at 7:58 AM, Roberto De Ioris <[email protected]> wrote: > > > Hello, > > > > I have an app where uWSGI cause a weird issue with the information that > is > > posted. The simple login form returns a 504 when I try to POST the > > username/password. > > I have try multiple version of uwsgi up the latest changeset in hg repo. > > > > Here it is the logs: > > > > #uwsgi.log > > writev(): Broken pipe [plugins/python/wsgi_headers.c line 171] > > write(): Broken pipe [plugins/python/wsgi_subhandler.c line 186] > > [pid: 4043|app: 0|req: 10/16] 127.0.0.1 () {48 vars in 986 bytes} [Wed > Mar > > 2 15:08:01 2011] POST /radarnation/login/ => generated 0 bytes in 60011 > > msecs (HTTP/1.1 403) 2 headers in 18446744073709551615 bytes (0 switches > > on > > core 0) > > > > #nginx.log > > 2011/03/02 22:09:01 [error] 1380#0: *106 upstream timed out (110: > > Connection > > timed out) while reading response header from upstream, client: > 127.0.0.1, > > server: , request: "POST /radarnation/login/ HTTP/1.1", upstream: > > "uwsgi://unix:///tmp/sockets/site_gamesradar.sock:", host: > > "gamesradar.local:8080", referrer: " > > http://gamesradar.local:8080/radarnation/login/"; > > > > #uwsgi xml config > > <uwsgi> > > <home>%d/ve</home> > > <chdir>%d</chdir> > > <socket>/tmp/sockets/%n.sock</socket> > > <module>%n</module> > > <processes>4</processes> > > <post-buffering>1</post-buffering> > > <master/> > > </uwsgi> > > > > I have tried with and without "<post-buffering>1</post-buffering>". Do > you > > have any idea of what could possibly go wrong here ? > > > > > You must find why it is taking more than 60 seconds. > Remove post-buffering (and update to latest tip) than add harakiri > (use something like 10 seconds) and add the new <harakiri-verbose/> > option. > > With this option turned on, you will get the syscall in which the process > is blocked (it is in numeric form, if you cannot decode it post it to the > list). > > Another way is spawning only a process and attach a strace to it: > > strace -p <pid> > > Try the login and check the strace output > > -- > 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
