On Fri, 2 Oct 1998, Teemu Peltonen wrote:
> We have three Linux boxes running Squid 1.2.beta25. I decided to give
> cache digests a chance, as we have a very busy system and the ICP-queries
> were tying up resources. However, the computers don't communicate with each
> others, i.e. there are no CACHE_DIGEST_HITs.
To see what is going on, try to enable level 5 debugging for the
peer_digest.c module (#72).
> Each of the caches have an unique IP address and it resolves to an unique
> name. However a hostname proxy resolves to three different IP addresses and
> that was a visible hostname.
> On regular intervals I got:
> 907264912.903 362 195.16.192.68 TCP_IMS_HIT/200 66705 GET
> http://proxy.netti.fi/squid-internal-periodic/store_digest - NONE/-
> application/cache-digest
This line looks correct. 66705 is a reasonable size for a digest.
> Then I changed visible_hostname to unique for each proxy:
>
> 907314454.606 2 195.16.192.68 TCP_MEM_HIT/200 736 GET
> http://mnemosyne.netti.fi/squid-internal-periodic/store_digest - NONE/-
> application/cache-digest
This line suggests that instead of a digest you've got an error message (too
small object size), but the response code (200) does not support that guess.
Enable debugging to find out what happened.
> What did I do wrong?
The most common problem is access controls for internal objects. There may be
other reasons, of course.
Alex.
Received on Fri Oct 02 1998 - 10:43:49 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:42:19 MST