Couple of quick queries regarding cache digest build

Hi,

We're running Squid version 2.2.1PATCH2, and after some initial tests, I
found that it's possible to create a redirector that returns a different
digest from the internal one. As a result, I want to use this feature to
return a digest that represents the contents of our cluster rather than
the just a clients entry point into the cluster.

Questions:
  * Squid keeps in memory 16 byte keys for URLs, and appears to utilise
    these in order to build the digest - yes?
  * It appears that cacheDigestAdd is effectively called only during
    digest build time, effectively like this; (excuse the perlism)
      foreach $key (@keys)
         cacheDigestAdd(digest, key).
    At least in terms of logic is this right?

If so then _one_ way (not necessarily the "best" :-) of building a cluster
digest would be :
  * modify cacheDigestAdd to dump the keys to a file as it builds its
    normal digest.
  * allow an auxillary machine to come along and pick up these files from
    all the caches.
  * The auxillary machine would then be able to make a digest very simply
    from the keys.

Whilst not necessarily a "reasonable" way of doing it - is this
essentially correct? If so then I'll probably knock up the small code
required to do this since it would mesh very well with some other work on
clustering I'm doing at the moment.

Michael.

--
National & Local Web Cache Support        R: G117
Manchester Computing                      T: 0161 275 7195
University of Manchester                  F: 0161 275 6040
Manchester UK M13 9PL                     M: Michael.Sparks@wwwcache.ja.net