I forgot to mention, someone said - "That happens because you do not read the whole request body in your app."
I do not believe this is true.. we are handling a multipart form and the image is complete uwsgi fails silently without any error, but nginx seems to detect the socket issue. On Tue, Oct 23, 2012 at 8:15 PM, Roberto De Ioris <[email protected]> wrote: > > > Hi, > > > > Turning post-buffering off seems to solve the issue. But we seem to need > > post-buffering enabled for other scenarios > > > > For example, when we try to upload certain files to our webapp, uploads > > fail unless we have post buffering enabled. I have attached the > > problematic > > file which always triggers this behavior > > > > the error message in such case goes something like: (from the logs) > > > > NGINX: > > > > 2012/10/23 16:22:47 [error] 9758#0: *67902 readv() failed (104: > Connection > > reset by peer) while reading upstream, client: 128.111.185.24, server: , > > request: "POST /import/transfer HTTP/1.1", upstream: > > "uwsgi://unix:///tmp/bisque.sock:", host: "loup.ece.ucsb.edu:9090", > > referrer: "http://loup.ece.ucsb.edu:9090/import/upload"; > > > > > > UWSGI: > > [pid: 24263|app: 0|req: 13/79] 128.111.185.24 () {54 vars in 1193 bytes} > > [Tue Oct 23 16:22:46 2012] POST /import/transfer => generated 394 bytes > in > > 736 msecs (HTTP/1.1 200) 2 headers in 79 bytes (1 switches on core 7) > > > That happens because you do not read the whole request body in your app. > > Regarding tmpfile it looks like it does not follow TMPDIR :( > > I suppose you have non-root access on that machine ? > > -- > 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
