Performance Monitor as a service?

NTDEV
1

Essentially, I want to use the Performance Monitor logging facility on a
machine as a service (so it doesn’t matter who is logged in, among other
issues). I want to save the Memory, Process, and Processor objects every 10
seconds or so. Except for the service part, I do this all the time using
Performance Monitor. (This is to track down a problem where something sucks
CPU for a few minutes on a particular machine every few days.)

So far, all I’ve found is the tool in the NT Resource Kit called PerfLog.
It does exactly what I want, but since it uses a text file for output
(instead of the compact PerfMon format), I’m concerned at how quickly the
log will grow.

PerfLog is good enough, but I’m wondering if anyone has come across anything
better?

2

>Performance Monitor. (This is to track down a problem where something sucks

CPU for a few minutes on a particular machine every few days.)

Unrelated to the topic of discussion here, I would ask that you tell this
list what the CPU-sucking culprit is - when you eventually discover it.

3

Almost sounds like the trinoo trojan horse DDOS attack. I had 2 linux
servers doing exactly the same thing, and theres supposedly an NT
version floating around: http://www.fbi.gov/nipc/trinoo.html

Trinoo is typically instructed to ping flood for 2-3 minutes, and for me
attackes were every two days. Just a thought :wink:

More on topic, I did a perfmon tool as a console app which you could
darn near just recompile as a service. Youre welcome to it if youre interested
and if I find it. Unfortunately I have the flu bad and Im outta here for the
night.
Ping me tomorrow if you want some source and docs.

-bob

Taed Nelson wrote:

Essentially, I want to use the Performance Monitor logging facility on a
machine as a service (so it doesn’t matter who is logged in, among other
issues). I want to save the Memory, Process, and Processor objects every 10
seconds or so. Except for the service part, I do this all the time using
Performance Monitor. (This is to track down a problem where something sucks
CPU for a few minutes on a particular machine every few days.)

So far, all I’ve found is the tool in the NT Resource Kit called PerfLog.
It does exactly what I want, but since it uses a text file for output
(instead of the compact PerfMon format), I’m concerned at how quickly the
log will grow.

PerfLog is good enough, but I’m wondering if anyone has come across anything
better?

You are currently subscribed to ntdev as: xxxxx@sequest.com
To unsubscribe send a blank email to $subst(‘Email.Unsub’)