On Mon, 13 Aug 2007, Edwards, David (JTS) wrote:
> Hi,
>
> On a multiprocessor box, I'm seeing the following loads.
Please include dmesg.
>
> top:
> load averages: 1.18, 1.17, 1.16
> 15:48:11
> 49 processes: 48 idle, 1 on processor
> CPU0 states: 2.8% user, 0.0% nice, 0.6% system, 0.0% interrupt,
> 96.6% idle
> CPU1 states: 0.0% user, 0.0% nice, 0.0% system, 0.0% interrupt,
> 100% idle
> CPU2 states: 0.0% user, 0.0% nice, 0.0% system, 0.0% interrupt,
> 100% idle
> CPU3 states: 0.2% user, 0.0% nice, 0.0% system, 0.0% interrupt,
> 99.8% idle
>
> uptime:
> 3:48PM up 15 days, 19:59, 2 users, load averages: 1.15, 1.17, 1.16
>
> These seem like high average loads to me but the server is idling (as
> shown by CPU usage in top).
>
> I understood that load averages were related to the number of processes
> in the run-queue (i.e. waiting to get some processor time) so average
> loads above 1 would show a processing bottleneck.
Yes, size of run-queue and load average are related. But interpreting
load average is almost impossible. Look a the top processes (including
system processes, shown with S in top) to see which ones request CPU
time.
> Did I get this wrong? If so, what do the load averages show and what
> would be a high load for a dual CPU (4 core) system?
Load average is a one-dimensional number and very much unusable to
describe the behaviour of the most complex systems mankind has
invented.
-Otto
