Hello!

On Tue, May 24, 2016 at 01:15:54PM -0400, hheiko wrote:

> We are runnging nginx reverse proxy on windows, the upstream consists of two
> lunixes based Apache/PHP Backend Servers. There is only one PHP application
> to be proxied. After starting nginx everything works fine, but then the
> backends become partly unresponsive and nginx is logging
>
> 2016/05/24 18:42:27 [warn] 5116#4660: *8863 upstream server temporarily
> disabled while reading response header from upstream, client: 77.23.138.88,
> server: wahl2.hannover-stadt.de, request: "GET / HTTP/1.1", upstream:
> "http://192.168.57.14:80/wrs/", host: "wahl2.hannover-stadt.de"
> 2016/05/24 18:42:27 [error] 5116#4660: *8863 upstream timed out (10060: A
> connection attempt failed because the connected party did not properly
> respond after a period of time, or established connection failed because
> connected host has failed to respond) while reading response header from
> upstream, client: 77.23.138.88, server: wahl2.hannover-stadt.de, request:
> "GET / HTTP/1.1", upstream: "http://192.168.57.14:80/wrs/", host:
> "wahl2.hannover-stadt.de"

[...]

> Googeling for " upstream server temporarily disabled while reading response
> header from upstream" brings up a few russian links but no clue whats the
> reason for this. There is no load on the system, both backend servers and
> nginx are idle.

The "upstream server temporarily disabled" warning means exactly
this: the server was disabled due to failures, and will not be
considered for balancing for some time.

This warning is logged in addition to normal logging of errors
("upstream timed out" in your case), and means that number of
errors observed by nginx crossed the max_fails threshold.

More details on max_fails / fail_timeout can be found here:

http://nginx.org/en/docs/http/ngx_http_upstream_module.html#max_fails

--
Maxim Dounin
http://nginx.org/

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