home | list info | list archive | date index | thread index

Re: [OCLUG-Tech] Interpreting system load

  • Subject: Re: [OCLUG-Tech] Interpreting system load
  • From: "Brenda J. Butler" <bjb [ at ] istop [ dot ] com>
  • Date: Fri, 7 Oct 2005 00:40:17 -0400
On Thu, Oct 06, 2005 at 08:27:22PM -0400, Alex @ Avantel Systems wrote:
> Is there some usefull way of interpreting the System Activity reports to 
> determine if the load on my system is more than it should be.  For example 
> that 1, 5, 15 minute average we get with "uptime" as well as the results of 
> the command "sar" and then of course the full report obtained with 
> "less /var/log/sa/sar05"  (all assuming it's redhat or fedora - don't know 
> about other distros)

The 1, 5, and 15 minute averages are the average number of processes
ready to run for the last 1, 5 and 15 minutes.  If this is less than 1,
you have cpu to spare.

I've heard that if you have a sparc machine, the number can be up to
10 and you're still ok...

So you have to observe your machine to see for yourself how peppy or
sluggish it is at various levels of utilization.

in /proc/loadavg, the 4th item is the current number of runnable processes
and the current total number of processes.  The last item is just the largest
pid so far, not really useful unless you're looking at boot time
and boot number of processes.  (The first 3 items are the 1, 5, and 15
minutes loadavgs.)

cheerio,
bjb

references