Why is idle machine reported as 1.0 load average?

[Andre Costa]

On Wed, 2 Feb 2005 21:43:52 -0500
B Wooster <bwooster47 at gmail.com> wrote:

> On Wed, 02 Feb 2005 13:31:20 -0700, Robin Laing
> <Robin.Laing at drdc-rddc.gc.ca> wrote:
> > B Wooster wrote:
> > > Using uptime, w, and top - I noticed that the current load average
> > > is listed as 1.0 - never falls below 1.0 - even though there is
> > > nothing going on on the machine, and top reports 99.8%+ idle
> > > time..
> 
> > How much memory?
> > You have 228 tasks listed?
> 
> 104 of those processes had "hotplug" in their name. I think there is
> some USB strangeness going on, every time I plugin and remove my SD
> Card Reader, I think 5-10 processes are created, most of which end up
> as zombies (I posted a message about this last week, but no reply on
> that yet).
> 
> Anyway, I rebooted my box, and now I see uptime is back close to 0.01
> when it reports 99% idle.
> 
> I guess I'll just have reboot the machine every week or so !

Usually, when there's no CPU usage and load is high, it's because
there's lots of I/O happening (or pending). I usually see this on our
machines at work when there's a lot of demand for file I/O through NFS,
and the servers are busy. When this happens, client machines' load tend
to increase significantly.

You could try 'sar' and 'iostat' commands (from the 'sysstat' package),
they will give you valuable info about I/O subsystem, queues etc.

HTH

Andre


-- 
Andre Oliveira da Costa