[SQU] Problem with cache digest

Hi!

We're having a problem with squid 2.3STABLE4 (we have also tried 2.4).

The problem is:

We're having 18 machines running squid and we would like them to peer
via cache digests, the machines are running linux.

Squid is compiled with options "--enable-cache-digests", and is also
building it's digest.

# ./client -p 8080 mgr:store_digest
HTTP/1.0 200 OK
Server: Squid/2.3.STABLE4
Mime-Version: 1.0
Date: Thu, 09 Nov 2000 18:01:02 GMT
Content-Type: text/plain
Expires: Thu, 09 Nov 2000 18:01:02 GMT
Last-Modified: Thu, 09 Nov 2000 18:01:02 GMT
X-Cache: MISS from fe040.worldonline.dk
X-Cache-Lookup: MISS from fe040.worldonline.dk:8080
Proxy-Connection: close

store digest: size: 67717 bytes
          entries: count: 102658 capacity: 108347 util: 95%
          deletion attempts: 0
          bits: per entry: 5 on: 288195 capacity: 541736 util: 53%
          bit-seq: count: 269921 avg.len: 2.01
          added: 102658 rejected: 13136 ( 11.34 %) del-ed: 0
          collisions: on add: 2.02 % on rej: 2.16 %

But exchanging the digest doesn't seem to work.

On one of the peers it looks like this:

peer digest from fe040.worldonline.dk
no guess stats for fe040.worldonline.dk available

event timestamp secs from now secs from init
initialized 973683802 -109230 +0
needed 973683806 -109226 +4
requested 973761746 -31286 +77944
received 973761746 -31286 +77944
next_check 973838546 +45514 +154744
peer digest state:
        needed: yes, usable: no, requested: no

        last retry delay: 76800 secs
        last request response time: 0 secs
        last request result: Not Found
                                                                                
peer digest traffic:
        requests sent: 9, volume: 0 KB
        replies recv: 9, volume: 3 KB
                                                                                
peer digest structure:
        no in-memory copy

The cache_peer is configured like this on the peering cache:

cache_peer fe040.worldonline.dk sibling 8080 3130 proxy-only no-query

When one of the other peers tries to retrieve the the digest the line
in the access.log looks like this:

TCP_MISS/404 364 GET http://fe040.worldonline.dk:8080/squid-internal-periodic/store_digest - NONE/- text/plain

And on the peer it looks like this in the cache.log:

temporary disabling (Not Found) digest from fe040.worldonline.dk

We've have also tried to run it with ICP requests, but it's taking up to
much bandwith (we have aprox. 30-40GB of proxy-traffic pr. day), but
while running with ICP, we could see that the proxy's where peering
correct, and saving us a lot of traffic, but unfortunatly also using
about 1MB/s for ICP request, and it was only enabled on one of the
machines.

I've looked in the FAQ and in the mailinglist, but haven't found the
answer to our problem.

Please help :-)

If you need more details, please don't hesitate to ask.

-- 
Med venlig hilsen / Best Regards
  Rene G. Jørgensen (System Administrator) / World Online Denmark A/S
  Peter Bangs Vej 26, 2000 Frederiksberg, Denmark
  Tlf. (+45) 38 14 70 00 - Fax (+45) 38 14 70 07
--
To unsubscribe, see http://www.squid-cache.org/mailing-lists.html