On Sat, 17 Apr 2004, Andreas Pettersson wrote:

> I've been monitoring squid 2.5 on FreeBSD 5.0 with mrtg for several months.
> The MIB is dated 2003/05/02 and defines cacheCpuUsage as "cacheSysPerf 5".
> However the retrieved integer nearly always shows up as 0 or 1, even when
> top or the cachemgr says squid is consuming 15-20% of the cpu.

The problem with this is that it is the average CPU usage since startup,
meaning that if your Squid has been idle for a long time this has a
significant impact on this gauge.

Use the derive of the cacheCpuTime counter instead with the Y scale

The "5 min average" from the general runtime information is also useable.
This is not represented in the SNMP mib however.