atom feed20 messages in ru.sysoev.nginxRe: Nginx + php-fpm "504 Gateway Time...
FromSent OnAttachments
rahul286Sep 6, 2010 10:38 pm 
Rahul BansalSep 6, 2010 11:27 pm 
Rahul BansalSep 6, 2010 11:27 pm 
Igor SysoevSep 6, 2010 11:31 pm 
Rahul BansalSep 6, 2010 11:40 pm 
Igor SysoevSep 6, 2010 11:46 pm 
Reinis RozitisSep 7, 2010 3:33 am 
Rahul BansalSep 7, 2010 9:03 am 
Reinis RozitisSep 7, 2010 9:56 am 
minaevSep 8, 2010 2:47 am 
Jérôme LoyetSep 8, 2010 2:54 am 
rahul286Sep 8, 2010 2:58 am 
minaevSep 8, 2010 3:29 am 
rahul286Sep 8, 2010 3:32 am 
Rahul BansalSep 8, 2010 4:01 am 
minaevSep 8, 2010 4:15 am 
Reinis RozitisSep 8, 2010 4:18 am 
Jérôme LoyetSep 8, 2010 5:16 am 
Michael ShadleSep 8, 2010 10:20 am 
net999Apr 14, 2011 1:13 pm 
Subject:Re: Nginx + php-fpm "504 Gateway Time-out" error with almost zeroload(on a test-server)
From:Michael Shadle (mike@gmail.com)
Date:Sep 8, 2010 10:20:58 am
List:ru.sysoev.nginx

On Wed, Sep 8, 2010 at 4:02 AM, Rahul Bansal <rahu@rtcamp.com> wrote:

-PHP-FPM debug really produced very less info which was not helpful at all. On sidenote, why php error shows up in nginx error.log and not in php-fpm?

502 will be when nginx's connection is broken to the upstream (it was forcibly killed by request_terminate_timeout, etc.) - it's bad gateway.

504 is when nginx is waiting around and hits fastcgi_read_timeout or other upstream related settings - it's a gateway timeout.

This isn't a "php error" per-se. nginx is logging the proxy related issue, php-fpm just isn't logging "backlog too full" type of messages (which it may or may not be able to do. No clue about that)

I've found the newer php-fpm to have plenty of error log output to play with. This may be a lower level kernel related thing that php-fpm is not aware of because the requests have been queued up and fill up/timeout/drop before php-fpm even knows about them. (Just a hunch)

_______________________________________________ nginx mailing list ngi@nginx.org http://nginx.org/mailman/listinfo/nginx