On Tue, 2002-11-12 at 15:44, richard.fuser@ernstyoung.com.au wrote:
> This email is to be read subject to the disclaimer below.
>
> Hi All,
>
> I have seen many post about this error and it seems a few fixes have been
> applied to squid over time to combat this problem.
>
> I am trying to cache an internal intranet site which has been working fine
> up until a few days ago.
>
> My cache.log shows the following:
>
> 1037074988.945 69598 10.2.3.1 TCP_MISS/000 0 GET
> http://staging.sub.domain.com/ userid TIMEOUT_DIRECT/111.22.33.44
>
> 1037075097.595 92116 10.2.3.1 TCP_MISS/000 0 GET
> http://staging.sub.domain.com/ userid TIMEOUT_DIRECT/111.22.33.44
>
> I think I understand the problem in that the server is sending a content
> length of 0 and squid does not like this.
IIRC it's the following sequence:
Squid connects to the remote server.
The remote server accept()'s the connection.
Squid sends it's request to the server.
The remote server calls shutdown() on it's connection without sending
any data.
I've got to run now, but I hope that helps. Try a TCP dump of the
behaviour.
Rob
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:11:17 MST