[squid-users] Cached objects dependent on browser?

From: Vicente Aguilar <vjaguilar@dont-contact.us>
Date: Tue, 20 Jun 2006 18:29:43 +0200

Hi

I'm setting up a typical httpd-accelerator configuration, with a squid
(v2.5.9) reverse-cache in front of the actual httpd serving and caching
all the requests.

It's working great so far, except for one somewhat minor detail: the
object's cache seems to be browser-dependent.

I mean, I see this behaviour when monitoring side by side both squid's
and Apache's access logs:

- Computer A access page.html with FireFox: MISS
- Computer A access page.html with Konqueror: MISS (why? same page!)

Then:

- Computer A access page.html again with FireFox: HIT (ok)
- Computer A access page.html again with Konqueror: HIT (ok)

And then:

- Computer B access page.html with FireFox: HIT (it's not IP-dependent!)
- Computer B access page.html with Konqueror: HIT

So it seems to me that squid is using both the URL and the User_Agent
(or part of it, maybe other headers too) to index the cached objects. Is
that right? Have searched through the archives and the docs but couldn't
confirm that point.

I've tried to "hide" the User_Agent from the actual server by using
header_access and header_reply, to no avail. Is it possible to force
squid into sharing the same cached objects between all browsers? Not
that there are so many out there, but hey...

Thanks in advance.

-- 
 Vicente Aguilar <vjaguilar@renr.es>
 Departamento de Sistemas
 Tlf.: 965 98 71 92
 Recursos en la Red, S.L.U.
 http://www.renr.es
Received on Tue Jun 20 2006 - 10:31:28 MDT

This archive was generated by hypermail pre-2.1.9 : Sat Jul 01 2006 - 12:00:02 MDT