Re: cache-digest errors.

From: Karl Ferguson <karl@dont-contact.us>
Date: Tue, 06 Oct 1998 02:40:29 +0800

At 09:57 AM 5/10/98 -0700, Alex Rousskov wrote:
>Apart from enabling miss_access, you may want to check digest utilization
>using cache manager interface. Ideal utilization (percent of bits "on") is
>50%. Values far off may indicate a problem. False hit ratio will increase
>if utilization is very high. Very low utilization means you are wasting
>bandwidth (likely with a filling-up caches). Use cache manager (peer and
>store digest stats) to get the numbers.

Nod. As I just said in my last reply to Dancer, miss_access is global, so
while I'm prepared to trust "some" neighbors, "some" might take advantage
of the fact . It seems that both our cache's (by looking at cachemgr) are
between 50 and 60 % hit - I'm gald that's considered optimal I would've
been happy with 20% or so :-)

>Eventually, Squid will handle miss access problems transparently to the
>client (regardless of the protocol).

Excellent. Roll on 2.1beta and the occasional coredump <grin> :-)

Regards

--
  This message is Copyright (c) 1998 by Karl Ferguson
Tower Networking Pty Ltd		Tel: +61 8 9355 0000
t/a STAR Online Services		Fax: +61 8 9355 0033
Received on Mon Oct 05 1998 - 11:44:00 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:42:20 MST