Re: [squid-users] Bypassing Squid completely for specific domains/IPs

Mike Raath wrote:
> Hi,
>
> There's an old discussion in these archives about how squid does a DNS
> lookup on the requested hostname and directs the request to the IP
> returned from DNS, effectively bypassing the destination IP supplied
> by the squid client, and I fully understand and take on board the
> reasons given for this, viz to prevent cache poisioning by spoofing
> domains.
>
> However, there is a very valid reason for not wanting this behaviour
> in some instances, and that's to test releases of web applications in
> a development environment by spoofing the production environment with
> hosts file entries, and in this case you wouldn't want to be viewing
> cached content anyway, so my question is this - is there a way of
> telling squid to just pass the request on wholesale to the requested
> IP without doing any DNS, and without caching?
>
> I don't think that always_direct does what I want, as squid still does
> the DNS lookup here. I can't believe that I'm the only one with this
> requirement, so how have others solved it?

Since you want to affect caching as well as routing. You will need to
setup ACL in side squid anyway to match the appropriate 'dstdomain'.
With the test web application configured as a cache_peer:

  cache_peer_access allow <ACL>
  cache_peer_access deny all
  cache deny <ACL>
  http_access allow <ACL>

Will route just that domain at the test app without using any form of
DNS (dues to dstdomain not needing it), and prevent caching of that
traffic all at once.

Amos

-- 
Please use Squid 2.7.STABLE4 or 3.0.STABLE8