Inaccurate process time accounting

[Pavel Machek]

Hi!

> I'm working on monitoring system for Linux clusters and I need to get
> dynamicaly CPU time consumed by process (no matter in which mode) for
> performance prediction. It seems that time accounting in my kernel
> (2.2.14) does not count time slices shorter than one jiffie. I made
> process that runs for less than jiffie and then sleeps one jiffie (in
> cycle). Although benchmarking showed that test takes around 30% of CPU
> time, accounting said that it had consumed less than 1%.
> 
> I tested patch from Pavel Machek
> http://www.tux.org/hypermail/linux-kernel/1999week20/0232.html
> 
> Results are more accurate but only if there is no other CPU bound
> process. 

Yes, that's expected. Main thing I wanted to avoid machine saying 99%
idle if it is in fact 90% busy.
								Pavel
-- 
I'm pavel en ucw.cz. "In my country we have almost anarchy and I don't care."
Panos Katsaloulis describing me w.r.t. patents me at discuss en linmodems.org

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo en vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/