Hello NG, I noticed quite different values on some monitoring
tools. I was installing VMWare server on my CentOs 5.3 (64
bit) and wanted to check the CPU usage. With the built-in
systemmonitor the usage was quite moderate. Then I started
HTOP and noticed peaks up to 100%. I ran them parallel and
noticed that in the same time the CPU2 had up to 100% in HTOP
the graphical Systemmonitor kept below 30%. Is this difference
normal? Do they have different points where they get their
values from? Does anyone have experience with monitoring tools
and their reliability? It is not so important, but it makes me
curious.

THX, Jacob

On comp.os.linux.setup, Jacob <spamtome at (no spam) ewetel.net> wrote:

Hello NG, I noticed quite different values on some monitoring
tools. I was installing VMWare server on my CentOs 5.3 (64
bit) and wanted to check the CPU usage. With the built-in
systemmonitor the usage was quite moderate. Then I started
HTOP and noticed peaks up to 100%. I ran them parallel and
noticed that in the same time the CPU2 had up to 100% in HTOP
the graphical Systemmonitor kept below 30%. Is this difference
normal? Do they have different points where they get their
values from? Does anyone have experience with monitoring tools
and their reliability? It is not so important, but it makes me
curious.

THX, Jacob



All the CPU-usage monitoring tools I've looked at the source
code of use /proc/stat as their source of basic data, so I
use this script to see what's going on. The more complex
monitoring tools, as you have learned, are sometimes difficult
to deal with.

(Called by /usr/bin/watch to run every few seconds:
watch -t -n 3 cpu.sh)

#!/bin/bash

# cpu.sh

# this is all one line

set - $(sed -n '/^cpu */s/\(^cpu *\)\([0-9][0-9]*\)\(
*\)\([0-9][0-9]*\)\( *\) \([0-9][0-9]*\)\( *\)\([0-9][0-9]*\)\(
*.*$\)/\2 \4 \6 \8/p' /proc/stat)

let tot=$1+$2+$3+$4

let utot=$1+$2+$3

use=$(echo "scale=6; $utot / $tot" | bc)

echo

echo "CPU usage is: $use"

echo


Sid