[SQU] URLs vs. MD5 key?

Hello fellow squid-sters,

(Running squid2.3STABLE4)

I'm running a significantly large squid cache whose
content is static. I'd like to be able to restore a lost
cache_dir (disk/partition) with the original content.
So I'm keeping a daily backup of the swap.state files
thinking they're much like a manifest of each cache_dir.

NOT so I discovered! Yes they can tell you IF they
they have something in cache not WHAT they have
in cache, there's a difference. The reason for this
is because the incoming URL I understand is hashed
through the MD5 algorithm which is only one-way.
That means unlike the old squid 1.x swap.state file I
can't map URLs to swap-filenames i.e. the object.

The alternatives I've realised is to store my "store.log"
files which contains both URL and swap filename.
Problem is this equates to a ludicrous amount of bytes
in disk space since I'm looking many millions of objects.
Similarly the other alternative to scan the header info
of the objects in cache daily is just as painful for like
reasons.

Is my understanding of the above correct and if so can
anybody suggest anything to help me out?

Many thanks,
Sam Liapis.
Melbourne, Australia.

--
To unsubscribe, see http://www.squid-cache.org/mailing-lists.html