hierarchy_stoplist and direct requests

Hi,

I am trying to tune my Squid up somewhat by avoiding the use of ICP on
queries which are clearly not cacheable - by forwarding uncacheable
requests directly to origin servers. In particular, URL's containing
strings such as cgi-bin and the ? character. Obviously these requests can
be forwarded directly to origin servers as they are not cached in the peers
I am working with.

I was under the impression that the hierarchy_stoplist statement was the
one that I needed to use to achieve this. At this stage, I have this
uncommented out in my squid.conf: hierarchy_stoplist cgi-bin ? but I'll
add to it a bit after I've figured out how to get Squid to work how I want.

In my web browser, I tried to go to a bogus URL with the cgi-bin statement
in it: www.dogpile.com/cgi-bin/ I didn't expect to get anything back,
but used it to observer Squid's behaviour.

In my access.log file I note this:

929891506.127 962 192.168.168.18 TCP_MISS/200 17 GET
http://www.dogpile.com/c
gi-bin/ - TIMEOUT_DEFAULT_PARENT/whale.telstra.net -

Why did the request go through this peer?

Config:
Squid 2.2Stable3 (with Henrik's 6/6 snapshot...I can afford to take risks ;> )
Redhat Linux 6.0
cache_peer whale.telstra.net parent 3128 3130 weight=5 default
prefer_direct off
always_direct allow (some other unrelated ACL's)
never_direct allow all
^^^^^^^^^^^^
...maybe this is the offending statement? If so, how do I go never_direct
on everything else _except_ what I have specified in my
hierarchi_stoplist? [should I set up another ACL and put it in
always_direct to cover this situation?]

Reuben

-------------------------------------------------------------
Reuben Farrelly Sunbury, VIC 3429, Australia
Received on Sun Jun 20 1999 - 09:13:14 MDT