[squid-users] Reverse-proxy-caching objects with query string

From: Jakob Borg <jakob@dont-contact.us>
Date: Fri, 22 Aug 2003 09:52:49 +0200

Hi all,

We have a website that is behind a reverse cache (http accellerator). It
works fine for caching images, but we also want to cache dynamically
generated pages with query strings (each different query string denoting
a different object, so page.asp?p1=foo and page.asp?p2=bar should be
cached as two separate objects). In principle, we would like Squid to
completely cease to understand the phenomenon query strings and simply
see that as a part of the filename. I have set:

acl QUERY urlpath_regex \?
acl CGIBIN urlpath_regex cgi-bin
no_cache allow QUERY
no_cache deny CGIBIN

and the pages that are to be cached return headers like:

HTTP/1.1 200 OK
Server: Microsoft-IIS/5.0
Date: Fri, 22 Aug 2003 07:43:09 GMT
Connection: keep-alive
Connection: Keep-Alive
Content-Length: 96750
Content-Type: text/html
Expires: Fri, 22 Aug 2003 07:44:07 GMT
Cache-control: Public

Which should make the object be cached for one minute, as far as I can
understands. However, it is not.

I realize this is very close to the usual "How come my certain objects
don't get cached" question but I _have_ read those documents and they
dont really seem to shed much light on this situation.

Right now I'm guessing there is some deep-rooted recognition of query
strings in Squid and that modification might be necessary... Any help
appreciated!

-- 
J a k o b  B o r g  /  MOROTSmedia AB  /  http://www.morotsmedia.se/
jakob@morotsmedia.se  /  +46 (0)46 2863373  /  +46 (0)70 6466677
Received on Fri Aug 22 2003 - 01:52:57 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:19:03 MST