Checking CPU Utilization

[Paul L. Allen]

jeff vier writes: 

> This is why (as I said in another email on this thread), Load is the
> correct way to judge a Unix (or Linux) server's...well...load.

Well, for some values of "correct." :) 

I have seen firewalls crippled with low loads and 80+% idle by
virus-laden machines on the secure side.  I have seen a mail server
crippled with high-ish loads and 80+% idle caused by a mail loop (my
fault) yet have higher loads and about 1% idle yet be responsive. 

Loads and idle-time are guides which may differ on machines with
different tasks or even on the same machine with different problems.
You can have a high load without significant performance hit or a low
load on a machine that is on its knees.  Your CPU can be idling because
the bottleneck is disk I/O (my VAX days come back to haunt me, but not
quite as badly as when I was looking after DEC kit). 

Similarly, your mail queue count might indicate that your machine is
falling behind or might indicate that you've had a lot of spam from
undeliverable addresses that your mailer is trying to bounce to.  All these
things are indications that a human, rather than a computer, needs to
evaluate what is really happening.  If you get constant false alerts then
you have to consider the trade-offs of setting thresholds higher (fewer
false alerts) against leaving them where they are (you don't miss real
problems). 

The day I can leave it to HAL to figure out if there really is a problem
or not is the day I get fired because my boss replaced me with HAL... 

-- 
Paul Allen
Softflare Support 




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
Nagios-users mailing list
Nagios-users at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nagios-users
::: Please include Nagios version, plugin version (-v) and OS when reporting any issue. 
::: Messages without supporting info will risk being sent to /dev/null