[squid-users] Determing # threads for async io?

We are migrating from squid-nt to squid on a linux box, using
2.5 stable 7 compiled with async io. Unfortunately, due to the
holidays, we've been unable to subject the new linux proxy to
any significant load since re-compiling with async io.

We have experienced one apparent lockup of squid so far, with
only 30 clients online, and minimal use. The squid process was
using 200Mb RAM, and 99.9% CPU, and refusing connections.
It did not respond to a shutdown, and processes had to be
killed off.

We have also seen a few "queue congestion" messages in
cache.log. I read one thread which suggested too many
threads may result in these messages. Are there other
methods to determine the optiumum number of threads?

Compile options are as follows:
Squid Cache: Version 2.5.STABLE7
configure options: --quiet --sysconfdir=/etc/squid --
localstatedir=/var --enable-delay-pools --enable-snmp --enable-
async-io=30 --disable-wccp --disable-ident-lookups --enable-
auth=ntlm,basic --enable-removal-policies

Cache is on three 9Gb 10k RPM scsi drives, using aufs.
System is a dual P3/733 with 768Mb running linux 2.4.22-37
(Mandrake 9.2)

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Shawn Wright, I.T. Manager
Shawnigan Lake School
http://www.sls.bc.ca
swright@sls.bc.ca
Received on Thu Dec 23 2004 - 16:50:18 MST