Re: [squid-users] Cache acl result: possible?

tor 2006-08-24 klockan 11:34 +0200 skrev Stefano Fraccaro:
> Hey all,
>
> I've SquidNT 2.6 Stable3 with very large porn domains file: 600.000
> sites blocked with dstdomain acl. When i try to surf html pages with
> only 2-3 images all seems to work fine, but if they are 20-30 images the
> fetching process is very slow (20-30 seconds).

Are you sure it's dstdomain and not a regex acl?

> I think squid will
> compare every time all requests with all porn domains. But if the first
> request is for domain www.aaaaaa.it and it's OK, all following requests
> for the same domain are valid... with or without url checking.

dstdomain automatically speeds up lookups of recently queried domains,
so I don't think it's this. But maybe.. haven't really thought of how
dstdomain works on blacklists..

You could try implementing the blacklist via an external acl instead
(external_acl_type) directive. Requires a small console application to
query the blacklist, nothing complicated.. I would probably use a perl
script querying a BDB containing the blacklist, or maybe SQL..

Regards
Henrik