Re: f9: question on load average

On Mon, 2008-09-15 at 17:09 +0200, Dario Lesca wrote:
Il giorno lun, 15/09/2008 alle 10.04 -0430, Patrick O'Callaghan ha
scritto:

"Load average" is a running average of the number of ready processes,
i.e. those that could do something if they had the cpu. Processes that
are waiting for something to happen, e.g. I/O to complete, can't use the
cpu even if it's idle. Thus it's perfectly possible to have the cpu
doing nothing while the load average is high. It often means you have
some long I/O operation on a slow disk system e.g. copying a DVD to an
external USB device, or multiple downloads waiting for data to arrive
from the net.

Thanks Patrick!

There is a command for see, in this case, what's the process that is
causing the hight Load Average?

It's not one process, it's caused by the set of processes on your system
in relation to the system's capabilities. The fact that there's a high
LA simply means that a lot of processes spend a lot of time waiting for
slow events. You can get an idea of which ones they are by looking at
the output of "ps axl". The STAT column shows process state; if it's 'D'
then the process is blocked on some event that might take a long time,
such as disk or network I/O. The WCHAN column shows the specific reason
for the wait (it's generally the address of a kernel data structure),
but you need to be familiar with kernel sources to really understand it.

Here's a pretty good explanation with some extra detail:
http://www.andymillar.co.uk/blog/index.php/2006/12/24/geekery/linux-load-average-explained/

Here's a more technical explanation of how the LA is calculated:
http://www.linuxjournal.com/article/9001

In my case I discovered that the problem was caused by a remote NFS
mounted with the remote NFS server out of office (my notebook).

That would probably do it :-)

poc

--
fedora-list mailing list
fedora-list@xxxxxxxxxx
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
Guidelines: http://fedoraproject.org/wiki/Communicate/MailingListGuidelines

Relevant Pages

  • Re: Strange high load
    ... When he imports it, it creates a high load on the machine, peaking ... about the state of the CPU. ... if there is something waiting on physical I/O does ... it's running on a processor, waiting for a processor to become ...
    (comp.unix.solaris)
  • Re: UNS: who/what uses my CPU?
    ... Amount of time the CPU has been waiting for I/O to complete. ... Ups, I've checked the man pages but I guess I missed this line. ...
    (Debian-User)
  • Re: Why is deleting (or reading) files not counted as IO-Wait in top?
    ... Simply because the only I/O that XFS does during a delete is ... CPU is.... ... 'Amount of time the CPU has been waiting for I/O to complete.' ... Async I/O means that typically your CPU does not get held up ...
    (Linux-Kernel)
  • system idle process
    ... In my experience, this means system is (CPU is waiting for) doing some I/O, ... Are there any other cases which will cause system idle process happen and consume a non-neglitible percentage of CPU? ...
    (microsoft.public.win32.programmer.kernel)
  • IO WAIT Information From IBM
    ... the I/O wait metric in AIX. ... AIX scheduler, the CPU "queues", the CPU states, and the idle or wait ... To summarize it in one sentence, 'iowait' is the percentage ...
    (AIX-L)