Re: [squid-users] problems with HTTP acceleration

Thank you so much Henrik!

> > ==> /var/log/squid/access.log <==
> > 1097694562.077 326 192.168.2.1 TCP_MISS/200 4586 GET
> > http://HOST/webnative/getimage?-small+/mnt/IDERAID_01/tsi_1/current/covers/Jun2003.eps - DIRECT/X.X.X.X image/jpeg
>
> http://www.mnot.net/cacheability/
>
> Also note that Squid won't cache requests requiring authentication unless
> the server says the content is public (Cache-Control: public). This is
> mandated by the HTTP RFC 2616.

Is there some way to override this behavior in squid outside of me
hacking the code?

> > root # curl --head
> > http://LOGIN:PASSWORD@HOST/webnative/getimage?-small+/mnt/IDERAID_01/tsi_1/current/covers/Jun2003.eps
> > HTTP/1.1 200 OK
> > Date: Wed, 13 Oct 2004 19:08:33 GMT
> > Server: Apache/1.3.14 (Unix)
> > Content-length: 4301
> > Content-Type: image/jpeg ; name="Jun2003.eps"
>
>
> This normally won't be cached as there is no information to base freshness
> upon. No Expires, not Last-Modified, not Cache-Control: max-age. Nothing.
>
> On both, see refresh_pattern, but you still need to solve the "public"
> criteria.

Excellent: refresh_pattern looks perfect. Assuming I get the public
issue resolved and the refresh pattern acceptable, will squid be able to
handle these dynamic pages give than the actual unique file name is
AFTER the "?" in the URL? By default the logs truncate everything after
the "?" and I worry that it might also truncate the rest of the URL when
doing the hash algorithm for cached objects....

        Thanks,
        Octo
Received on Thu Oct 14 2004 - 11:36:24 MDT