is proxy_send_timeout too small ? I newer 2012/3/3 赵帅 <rostin...@gmail.com>
> I suggest you do a pressure test on the url that have triggered the > problem to see whether it will show up again.If yes, to see the system > status such as the number of sockets being used and the number of > connections from nginx to fastcgi service in the status of waiting to be > accepted. > > > 2012/3/3 Daniele Procida <dani...@vurt.org> > >> We're having a devil of a time with our new server, which went live >> yesterday after two months of testing. >> >> Every so often, nginx will display a 502 gateway error page, and >> something like this will appear in the logs: >> >> 2012/03/02 18:05:38 [error] 29743#0: *1479 upstream prematurely closed >> connection while reading response header from upstream, client: >> nn.nn.nn.nn, server: nn.nn.nn.nn, request: "GET /some/path/ HTTP/1.1", >> upstream: "passenger:unix:/passenger_helper_server:", host: "example.com" >> >> This only occurs on some pages, but once it has happened on a page, it >> seems to keep happening. >> >> It doesn't happen straight away - the server can run happily for some >> time before the errors start occurring. >> >> The pages that trigger the error don't seem to be particularly hard work >> for the server. However, the same pages can be expected to trigger it, >> while some never do. >> >> Merely restarting nginx doesn't get rid of the errors - they come back >> again. >> >> As soon as DEBUG = True is applied in settings and nginx restarted, the >> errors disappear. >> >> Some things that seem to have improved matters: >> >> * increasing somaxconn to 1024 >> * turning off cacheing >> >> In fact after turning off cacheing it was some hours before the errors >> returned. >> >> We are running nginx, with a passenger_wsgi.py file setting up the path >> and pointing to the settings.py file. >> >> I'd appeciate any tips on what the problem might be, or how to go about >> isolating it. >> >> Obviously an error that occurs under heavy load is not unusual, though I >> don't think the load is particualrly heavy (and this server has taken over >> from one rather less powerful that has been doing the same job for two >> years). >> >> What is puzzling is the error that occurs only on particular apparently >> arbitrary pages, and then persists, even after the load conditions (we were >> using ApacheBench to test) have disappeared. >> >> Daniele >> >> -- >> You received this message because you are subscribed to the Google Groups >> "Django users" group. >> To post to this group, send email to django-users@googlegroups.com. >> To unsubscribe from this group, send email to >> django-users+unsubscr...@googlegroups.com. >> For more options, visit this group at >> http://groups.google.com/group/django-users?hl=en. >> >> > -- > You received this message because you are subscribed to the Google Groups > "Django users" group. > To post to this group, send email to django-users@googlegroups.com. > To unsubscribe from this group, send email to > django-users+unsubscr...@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/django-users?hl=en. > -- as I learning english, any help about english grammar is welcome -- You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-users@googlegroups.com. To unsubscribe from this group, send email to django-users+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/django-users?hl=en.
