Epoll error with http1.1 and keep-alive enabed

[Maxim Dounin]

Hello!

On Fri, Aug 19, 2011 at 03:02:14PM +0800, quan nexthop wrote:

> Hi all:
> 
> I use nginx 1.0.5 with Maxidom's patch. NGINX works as reverse proxy,
> backend is IIS 6.0.
> 
> we can  not list the data from server, I checked the nginx error_log, there
> is problem as following:
> there are two error, anyone can give some help or clue?
> 
> 
> 1) epoll_wait() error on fd:10 ev:001D
> 2011/08/19 06:17:20 [debug] 1780#0: epoll timer: 599985
> 2011/08/19 06:17:20 [debug] 1780#0: epoll: fd:10 ev:001D d:080E4FF0
> 2011/08/19 06:17:20 [debug] 1780#0: epoll_wait() error on fd:10 ev:001D
> 
> 
> 2) recv() not ready (11: Resource temporarily unavailable)

These are debugging messages.  The one in your log which 
isn't debugging explains what is happening:

> 2011/08/19 06:17:20 [info] 1780#0: *1 client closed prematurely connection,
> so upstream connection is closed too (104: Connection reset by peer) while
> reading response header from upstream, client: 192.168.226.200, server:
> abc.com, request: "GET
> /Server/extjs/resources/css/ext-all-embedded-css/ext.axd?v=29344 HTTP/1.1",
> upstream: "
> http://192.168.2.10:80/Server/extjs/resources/css/ext-all-embedded-css/ext.axd?v=29344",
> host: "abc.com", referrer: "http://abc.com/Server/default.aspx"

That is, client closed connection before nginx was able to get 
response from upstream server.

Maxim Dounin