atom feed13 messages in ru.sysoev.nginxRe: Client closed keepalive connection
FromSent OnAttachments
Wilson BilkovichApr 27, 2007 10:39 am 
Igor SysoevApr 27, 2007 10:52 am 
Wilson BilkovichApr 27, 2007 11:11 am 
Igor SysoevApr 27, 2007 11:17 am 
Wilson BilkovichApr 27, 2007 11:22 am 
Igor SysoevApr 27, 2007 11:33 am 
Wilson BilkovichApr 27, 2007 11:46 am 
Cliff WellsApr 27, 2007 9:10 pm 
Igor SysoevApr 27, 2007 11:47 pm 
Wilson BilkovichApr 28, 2007 10:01 am 
Igor SysoevApr 28, 2007 11:46 am 
Wilson BilkovichApr 28, 2007 1:35 pm 
Igor SysoevApr 29, 2007 7:03 am 
Subject:Re: Client closed keepalive connection
From:Wilson Bilkovich (wils@public.gmane.org)
Date:Apr 28, 2007 1:35:55 pm
List:ru.sysoev.nginx

On 4/28/07, Igor Sysoev <is-G@public.gmane.org> wrote:

On Sat, Apr 28, 2007 at 01:02:16PM -0400, Wilson Bilkovich wrote:

On 4/28/07, Igor Sysoev <is-G@public.gmane.org> wrote:

On Fri, Apr 27, 2007 at 09:10:47PM -0700, Cliff Wells wrote:

On Fri, 2007-04-27 at 21:53 +0400, Igor Sysoev wrote:

Does the client do POST and pass "Expect: continue" ? nginx does not support "Expect: continue/100 Continue" dialog because current browsers still do not support it (at least I never see).

I'm not aware of any browsers that support "Expect: continue", however there are some tools and libraries that do (curl/libcurl for one).

As for me, the main use of "Expect: continue/100 Continue" dialog is the request body limitation. I had even written article (sorry, in Russian) "Why it's impossible to limit correctly the uploaded file size": http://sysoev.ru/web/upload.html

The problem is that browsers do not read server response until they will send whole request body. If server will close connection before they have sent body, the browsers issue some network or common error, that has no relation to the "413 Request Entity Too Large" error.

OK. I've disabled the "100 Continue" piece of the response, and I am still seeing these error messages.

Do you mean these messages ?

With "use kqueue" 2007/04/26 22:26:15 [info] 4133#0: *3048 kevent() reported that client 192.168.0.100 closed keepalive connection

With "use poll" 2007/04/26 22:59:48 [info] 10189#0: *375 client 192.168.0.100 closed keepalive connection

They are not actually errors, they are the info messages and this is normal. If client is MSIE you will see

"... [info] ... client ... closed keepalive connection (54: Connection reset by peer)"

It's also normal: MSIE always closes keep-alive connection with RST packet.

I guess my question is this.. Why does nginx think that there is a keepalive connection when, to my knowledge, I have not requested one? Is this a message I can receive even if there is no keepalive connection?

I am not using a web browser, but rather an HTTP library, so I have a good amount of control over the request itself.