Cluster of squids

I'm playing with some network configurations right now, and am setting up a
number of small machines to "cluster" together. What I would like is a
single parent squid with lots of disk space, ram, possessors, etc, and all
of the smaller machines to be siblings. These machines would all have very
small caches, and are somewhat limited in resources.

If all the squids are compiled with the digest option, and defined as
peers...will they all automatically share the digests? I've read the FAQ,
but I don't understand how they just decide to do this.

The first machine would define it's peers as:

cache_peer squid-parent parent 3128 3130 no-query proxy-only
cache_peer squid-sib-2 sibling 3128 3130 no-query proxy-only
cache_peer squid-sib-3 sibling 3128 3130 no-query proxy-only
cache_peer squid-sib-4 sibling 3128 3130 no-query proxy-only
cache_peer squid-sib-5 sibling 3128 3130 no-query proxy-only

The second as:

cache_peer squid-parent parent 3128 3130 no-query proxy-only
cache_peer squid-sib-1 sibling 3128 3130 no-query proxy-only
cache_peer squid-sib-3 sibling 3128 3130 no-query proxy-only
cache_peer squid-sib-4 sibling 3128 3130 no-query proxy-only
cache_peer squid-sib-5 sibling 3128 3130 no-query proxy-only

etc....

How can I make sure that the siblings all check out each before checking
the parent?

Is this configuration going to work how I think it will?

Is this stupid? :)

Thanks, this really is a great resource. I've feel that I've learned a lot
from being on this lists for the past month...

-jimi
Received on Thu Jan 06 2000 - 19:17:27 MST