nginx 1.4.1 - slow transfers / connection resets

[Maxim Dounin]

Hello!

On Mon, Aug 19, 2013 at 02:07:09PM +0200, Philip Hofstetter wrote:

> Hi,
> 
> I have a nginx (stock ubuntu config) as a reverse proxy in front of  a
> haproxy in front of 5 more nginx machines which use fastcgi to talk to
> php-fpm.
> 
> My issue is with the frontend proxy and long-running, veeeeery
> slowwwww requests.
> 
> The clients are very underpowered mobile barcode scanners using 2G GSM
> connections. When they try to download 2.1 MB of data dynamically
> generated by PHP on the back, backend, the Frontend will close the
> connection after ~1MB has been downloaded (at ~2 KBytes/s).
> 
> I can reproduce the same behavior using curl (with --rate-limit 2K):
> 
>   % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                  Dload  Upload   Total   Spent    Left  Speed
> 100 1196k    0 1196k    0   343   2047      0 --:--:--  0:09:58 --:--:--  1888
> curl: (56) Recv failure: Connection reset by peer
> 
> The access log on the frontend server lists a 200 status code but too
> few transmitted bytes.
> 
> The error log (on info) shows
> 
> 2013/08/19 14:03:36 [info] 32469#0: *1166 client timed out (110:
> Connection timed out) while sending to client, client: xxx.xxx.xxx.xxx
> 
> Which is not true - it's showing that while curl (--rate-limit 2K) ist
> still running.
> 
> Can you give me any pointers in how to debug/fix this?

Debug log should be helpful, see 
http://nginx.org/en/docs/debugging_log.html.

-- 
Maxim Dounin
http://nginx.org/en/donation.html