Article delegate-en/1721 of [1-5169] on the server localhost:119
  upper oldest olders older1 this newer1 newers latest
[Top/Up] [oldest] - [Older+chunk] - [Newer+chunk] - [newest + Check]

Newsgroups: mail-lists.delegate-en

[DeleGate-En] Antwort: Re: [DeleGate-En] broken post-response with http/1.0 keep-alive
30 Jun 2002 20:57:48 GMT "Marc Pohl" <>

Hi Yukata,

i have placed a additional sv1log() call at the beginning HTTP_putMIMEmsg()
and started delegated with -vv

This experiment shows that the function HTTP_putMIMEmsg() is not even
called if i reproduce the error. I think one special thing in this scenario
is, that the first requests in keep-alive are all 304-results without any
body and the last request is a post without content-length. Because of this
the failed request is always the last request of the keep-alive connection
"HCKA:[4] closed -- M:Method is ``POST'' cpid[0]"

I think the flush-thing is right direction, but what can i do to support
additional data?


Marc Pohl
WWF Wort + Ton GmbH
Bereich e-media

Tel. 0000 0000 00X
Fax. 0000 0000 00X

                    (Yutaka Sato)        Kopie:                                       
                                         Thema:  Re: [DeleGate-En] broken post-response with http/1.0          
                    30.06.2002           keep-alive                                                            
                    antworten an                                                                               

On 07/01/02(01:14) you "Marc Pohl" <> wrote
in <_A1717@delegate-en.ML_>
 |But this happens only with http/1.0 keep-alive. If the IE 5.5-Setting
 |HTTP/1.1 for Proxy-Connections" is activated, all works as desired. If i
 |change the Cache-Behaviour of IE, so that the 304-requests don't happen,
 |there is also no problem ...

The difference between behaviors of DeleGate for HTTP/1.0 Keep-Alive
and HTTP/1.1 is that, only in former case DeleGate does buffering
of response message to make sure the length of response message
to be put into the Content-Length header, to ensure the functionality
of Keep-Alive mechanism in HTTP/1.0, which is not necessary in HTTP/1.1
because of its "chunked" encoding.

In HTTP/1.0 Keep-Alive, a buffered message is flushed toward the client
in HTTP_putMIMEmsg().  A possible scenario to explain your situation
is that flushHead()->fflush() called from the function returns EOF
by some reason after flushed the header, thus aborting relay of body.
In this case, you see "## stop relay-3: flush()=EOF, client seems dead"
in the LOGFILE.  I wish if I could see the LOGFILE.

  @ @ Yutaka Sato <>
 ( - ) National Institute of Advanced Industrial Science and Technology
_<   >_ 1-1-4 Umezono, Tsukuba, Ibaraki, 305-8568 Japan

  admin search upper oldest olders older1 this newer1 newers latest
[Top/Up] [oldest] - [Older+chunk] - [Newer+chunk] - [newest + Check]