ProLiant Servers (ML,DL,SL)
cancel
Showing results for 
Search instead for 
Did you mean: 

enterprise MIB for CPU util correct?

 
Ayman Altounji
Valued Contributor

enterprise MIB for CPU util correct?

Have installed softpaq 5.14A dated 17 Jan 02
This system has 8 CPUs, 700Mhz each.
I'm interested in measuring cpu performance (usr, sys, idle).

Compaq enterprises mib, "enterprises.232" includes the

cpqHoCpuUtilTable (232.11.2.3.1). In this subtree are defined (in the

MIB included with the softpaq) OIDs that return - for each cpu - the

avg utilization over the past minute, 5 minutes, etc.
I've tracked these values (using rrdtool/mrtg) for more than 24 hours

and though the reported values have sometimes been as high as 40,

most of the values average to zero. I'm taking samples every 20s (and

keeping thousands of samples (4,320) - so there are plenty of

samples.

The reported values just don't seem correct. Similar software running

on E4500 Solaris 7 servers with 4 x 400Mhz processors report

utilization that hovers between 5-10% even at idle.

Are these results similar to what others are seeing?
Are there any configuration steps I may have overlooked?

I have queried on the corresponding microsoft (311) OIDs and they

*always* report zero or 100 (very arbitrarily).

I will double-check my collection of data samples ( rrdtool dump ) to

make sure rrdtool isn't averaging them down to zero. The reported

values just seem suspect.

Is there any documentation that describes how the snmp agents are

implemented?

Many questions, I admit. Cheers,

Lyle