[wellylug] ubuntu load average

Daniel Pittman daniel at rimspace.net
Wed Jan 21 12:13:20 NZDT 2009


Atom Smasher <atom at smasher.org> writes:
> On Thu, 15 Jan 2009, Jethro Carr wrote:
>
>> Have you checked to see if there are any odd processes which are
>> crashed but stuck waiting for disk?
>>
>> In theory, if a program crashed in such a way that it was always
>> waiting for I/O it would cause 1 to be added to the load.
>>
>> Run a ps aux and see if there are any processes not in sleep that
>> should be. Check for anything with a STAT of "D".
> =====================
>
> guilty!
>   4873 ?     D    0:00 hald-addon-storage: polling /dev/scd0 (every 2 sec)
>
> i can kill that process (i was surprised that it died without a fight,
> being in "uninterruptible" sleep) and the load average quickly dropped
> to <0.05. much better.

Very strange; presumably your CD[1] drive is doing something strange that
means it isn't sitting in a normal sleep.

> i can make the load average climb again by running "/etc/init.d/hal
> restart".
>
> i'm familiar with FreeBSD, and hal isn't part of my setup. i'm not
> even sure if this is a real problem, or something that's ok to
> ignore. what do ya'll think?

Well, the load average is a more or less meaningless figure alone.
There is no particular number that means anything good, bad or
indifferent.[2]

If you just want to disable the CD polling, which means no notification
when a CD is inserted, use 'hal-disable-polling'

Regards,
        Daniel

Footnotes: 
[1]  For "CD" read "CD or DVD, or equivalent optical media drive"

[2]  As an example, on some of the recent Sun hardware that goes to
     extremes for multi-core operation a load average of more than 200
     is just fine — the system has 250 cores, so a load average of 200
     indicates that it is running at a comfortable ~80 percent capacity.



More information about the wellylug mailing list