[wellylug] ubuntu load average

[Daniel Pittman]

Pete Black <pete at marchingcubes.com> writes:
>> Atom Smasher <atom at smasher.org> writes:
>>
>>> my wife's laptop is running ubuntu 8.10. with top showing the system
>>> at close to 100% idle (>96%) i haven't seen the load averages drop
>>> below 1.0.  i'd expect it to be closer to 0.0 when it's >95% idle. any
>>> ideas why the load averages seem skewed?
>>
>> Something is either running full time, or waiting on disk I/O.

[...]

> Some machines have irq issues with the linux kernel, this can manifest
> as a high load average, clock drift etc.

Really?  Your description sounds very much like a screaming interrupt of
some sort, which is a very serious issue, but your description here
doesn't quite match up.

The only citations I can find for high load average on idle or clock
drift don't suggest interrupt issues, but rather kernel or hardware
faults in other areas.

Can you point me to some more information about these IRQ issues?

> check /proc/interrupts and if the rtc or some other thing is spewing
> thousands of interrupts per second, that will raise you load average
> on idle significantly.

It shouldn't: hard and soft IRQ time isn't counted against the load
average, but is rather a moving average of the number of tasks found in
the scheduler run queue(s).

In fact, one of the common problems with performance analysis for Linux
routers that act primarily outside of process context is that they don't
show up in the load, or in the standard CPU accounting information,
precisely because hand and soft IRQ processing is hidden...

> Not sure how to fix this - on a desktop machine you could try
> disabling apic/acpi with kernel args but a lappy kind of needs that
> stuff for power management.

More and more equipment will function incorrectly, at best, with ACPI
disabled, and that includes desktop hardware, not just laptop hardware.

Generally, disabling ACPI is very wrong on anything recent.

Regards,
        Daniel