fwdAbort problem on Squid 2.3.stable2

Hello,

We have a very interesting problem:

Our institution has got two proxies, both are Sun boxes running Solaris
2.6 with patches.
One proxy (a Sun Ultra 1) is running Squid 2.2 Stable 5 (let us call it
proxystable 2.2), the other one (a Sun ultra 5000 with 6 CPU's) is
running Squid 2.3 Stable2 - let us call it proxystable2.3. They are both
configured as siblings, and their squid.conf files are roughly the same.

When accessing some sites from proxystable 2.3, such as www.compaq.com,
browsers keep indicating

connect: host xxx contacted. waiting for reply.

and they keep waiting for a long, long time, while not data is loaded...
That problem does not occur on proxystable2.2, where www.compaq.com
first page is loaded almost instantly.
An analysis of cache.log on proxystable 2.3 shows that the request is
first ALLOWED, and then fwdAbort for that request appears in a time
range that is between 0 second and 10 minutes later. But browsers still
display the above message....waiting for reply.

A solution was to configure proxystable 2.2 as a parent of proxystable
2.3, and to configure their peering as proxy-only. Then, access to
www.compaq.com was just fine, but access to secure sites was still a
problem. So SSL requests from proxystable 2.3 were instructed not to go
direct. Since then, that solution works most of the time, but it is not
elegant and response times are obviously higher when our users are using
proxystable 2.3.

Did anybody experience problems when accessing the following sites with
squid 2.3 stable2?

www.compaq.com
www.mycio.com
www.winfiles.com
www.teamcast.com

thanks

Laurent Evrard
Received on Wed Apr 12 2000 - 08:57:00 MDT