[mrtg] Weird results using MRTG on multi-CPU Windows 2000 servers

[Michael W. Lucas]

Hello,

I'm having poor results using MRTG to monitor CPU usage on Windows
2000 servers.  The setup includes:

* mrtg-2.9.18.p11 on FreeBSD (installed via ports)

* Multiple Windows 2000 server with the SNMP4tPC "expansion" -- this
whole technique is unsupported by Microsoft, but there doesn't seem to
be another choice.  Sigh.

When MRTG runs, it frequently reports one of the two CPUs at being at
100% user capacity.  This does not match either what an individual
snmpwalk says, or what Performance Manager says.  For example, here's
a log snippet from a very new install:

1070896963 100 0
1070896963 100 0 100 0
1070896831 0 0 0 0
1070896800 0 0 0 0
1070896500 40 0 100 0
1070896200 100 0 100 0
1070895900 100 0 100 0
1070895600 100 0 100 0
1070895300 100 0 100 0
1070895000 38 0 100 0
1070894700 100 0 100 0
....


But snmpwalk gives me:

# snmpwalk server community .1.3.6.1.4.1.311.1.1.3.1.1.2.1.4.1.49
enterprises.311.1.1.3.1.1.2.1.4.1.49 = 0
#

Odd things I've noticed:

This affect only happens on one CPU, not both.

When I removed the log of the obviously-erroneous CPU and let MRTG
start over, the graph for the other CPU started climbing and quickly
reached 100% user CPU.

I'm seeing this on several different multi-CPU Windows 2000 machines.

Here's a MRTG config for a single CPU; the CPU second is the same, with
(of course) the next MIB.

Target[webserver3.cpu2]: 1.3.6.1.4.1.311.1.1.3.1.1.2.1.4.1.49&1.3.6.1.4.1.311.1.1.3.1.1.2.1.4.5.49:community at webserver3
Title[webserver3.cpu2]: webserver3 cpu 2 usage
Options[webserver3.cpu2]: growright, gauge
MaxBytes[webserver3.cpu2]: 100
YLegend[webserver3.cpu2]: Percent (%)
WithPeak[webserver3.cpu2]: wmy
LegendI[webserver3.cpu2]: User Time:&nbsp
LegendO[webserver3.cpu2]: Privileged Time:&nbsp
Legend1[webserver3.cpu2]: % cpu 2 User Time
Legend2[webserver3.cpu2]: % cpu 2 Privileged Time
Legend3[webserver3.cpu2]: Maximal 5 Minute % cpu 2 User Time
Legend4[webserver3.cpu2]: Maximal 5 Minute % cpu 2 Privileged Time
ShortLegend[webserver3.cpu2]: %
PageTop[webserver3.cpu2]: <H1>cpu 2 usage for webserver3 </H1>

Any thoughts?  If I can get this on multiple machines, surely someone
else has seen it before?

On a related note, this machine now seems to recognize the host mib
for generic processor usage, as shown below.  Is this a better choice
to monitor CPU usage, or is the difference irrelevant?

# snmpwalk webserver3 community .1.3.6.1.2.1.25.3.3.1.2
host.hrDevice.hrProcessorTable.hrProcessorEntry.hrProcessorLoad.1 = 1
host.hrDevice.hrProcessorTable.hrProcessorEntry.hrProcessorLoad.2 = 0
#

Any help appreciated.

Thanks,

==ml


-- 
Michael Lucas		mwlucas at FreeBSD.org, mwlucas at BlackHelicopters.org
Today's chance of throwing it all away to start a goat farm: 41.8%
		http://www.BlackHelicopters.org/~mwlucas/
           Absolute OpenBSD:   http://www.AbsoluteOpenBSD.com/





-- 
Michael Lucas		mwlucas at FreeBSD.org, mwlucas at BlackHelicopters.org
Today's chance of throwing it all away to start a goat farm: 41.8%
		http://www.BlackHelicopters.org/~mwlucas/
           Absolute OpenBSD:   http://www.AbsoluteOpenBSD.com/

--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Archive     http://www.ee.ethz.ch/~slist/mrtg
FAQ         http://faq.mrtg.org    Homepage     http://www.mrtg.org
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi