[squid-users] Accelerating multiple servers on different ports

We're currently looking at using Squid to accelerate two different web
applications, one in Zope and the other in a J2EE container. These
apps currently run on the same box, listening on different ports.
We're using Apache and mod_rewrite to map name-based virtual hosts to
the appropriate backend.

I read through the Squid documentation and have been able to get
caching working for one of these applications (either one). I saw
some discussion in the mailing list archive about setting up explicit
IP address mappings in the hosts file, and I've been able to do that
in tests to map to two different back-end servers, so long as the
content is running on the same port.

My question is how to send requests to different ports. If I've read
the documentation correctly, it seems that a redirector program should
be able to handle this. I've installed Squirm and set the
redirect_program setting in squid.conf, and have tested it
successfully in interactive mode. However, from my observations, it
appears that Squid is still using the hosts file/DNS to map the
requested name to the backend.

Any suggestions? Something I'm missing? Thanks,

Nathan R. Yergler
Software Engineer
Creative Commons
Received on Mon Jun 20 2005 - 08:23:56 MDT