[wplug] cpu load?

[Bryan J. Smith]

"Jonathan S. Billings" <billings at negate.org> wrote:
> Every process that is waiting on I/O can easily increase your load
> average.

These types of "real world" scenarios cannot be stressed enough.

It also means that, as Jonathan (among others) have pointed out,
average system load concept must be well understood.  I.e., system
load is a good "quick glance" to see if something is "inefficient"
with your system (like processes waiting on I/O).  It's
values/indicies are *NOT* a "good, comparable metric" between runs. 
Again, it's more of a set values that may be "red flags."

> If I recall correctly, every process that's in uninterruptible
> sleep increases the load average by one, so if you have 32
> processes in that state, your load average is increased by
> 32.

Good catch!  I'm a little rusty myself, because I largely ignore it. 
I'm typically running vmstat, iostat and a few others, if not using a
hardware PCI-X or PCIe tap directly (when I have a client with money
-- like in the financial industry ;).



-- 
Bryan J. Smith   Professional, Technical Annoyance
b.j.smith at ieee.org    http://thebs413.blogspot.com
--------------------------------------------------
     Fission Power:  An Inconvenient Solution