Re: [tlug] Bogus load data

[Junichi Uekawa <dancer@example.com>]

Hi,

> So, thoughts? This is extremely unscientific, but do you think it
> might give some indication of a problem on a system that is *really*
> heavily loaded? My idea was that if the system was busy, this would
> report trying to sleep for a short time, but actually sleeping for a
> much longer one, due to slow context switches. The "sleep(1);" line at
> the end of the main loop is to ensure at least one context switch
> occurs between usleep() syscalls.

#include <sched.h>
sched_yield()

might give more interesting results; although it really depends on
what you're after.

regards,
	junichi
-- 
dancer@{debian.o??,netfort.gr.jp}   Debian Project