UNIX for Advanced & Expert Users

Hi Unix Gurus,

We're trying to verify few alerts to be generated by our application when the threshold values are crossed. As part of this we need to generate an alert when the %CPU utilization of the Kernel goes beyond a certain value. We monitor the %CPU of kernel using the mpstat command (the field is highlighted in bold and red color below).

Code:

06:19:36 AM  CPU   %user   %nice    %sys %iowait    %irq   %soft  %steal   %idle    intr/s
06:19:36 AM  all    0.39    0.00    0.34    0.66    0.00    0.03    0.00   98.58   1211.49

And right now i'm finding it difficult to increase this kernel %CPU. It takes a lot of time to increment the value of the %CPU of Kernel. I'm using the following command to do this :

Code:

fulload() { dd if=/dev/zero of=/dev/null | dd if=/dev/zero of=/dev/null | 
           dd if=/dev/zero of=/dev/null | dd if=/dev/zero of=/dev/null & }; 
           fulload; read; killall dd

But the cpu utlilisation increments very slowly with the above command.

I'm working on RHEL 5.6
Pls. advise if you guys have any ideas.

Thanks

None of this makes a lot of sense.

Is there a reason you cannot simply abuse the your application to make it use more kernel?

If you simply have to increase kernel time another way then put together something that creates lots of threads with short duration. Or lots of processes. Where lots==100+.

Somewhat like the idea of a fork bomb. DON'T use a fork bomb. Not a good idea. All it does is create endless processes. Limit the total number of processes.

Code:

while true
do
   cnt=1
   while [ $cnt -le 100 ]
   do
         ls /tmp > /dev/null &
         cnt=$(( $cnt + 1 ))
   done
done

None of this is a great idea. There is code out there - google for 'stress test unix'

Try setting dd's block size to 1. Most of the time will be spent in useless system calls.