Re: [gentoo-user] Re: load too high

[Alan McKinnon]

On Tuesday 12 February 2008, James wrote:
> Alan McKinnon <alan.mckinnon <at> gmail.com> writes:
> > > One of the workstations (amd64 2gig ram) has a load that never
> > > drops below 1.0, as seen by top. Looking at a ps nothing stands
> > > out. I did notice that 'X' is at the top of the list, even when
> > > the machine is quiescent (nobody doing anything). Suspiciaous.
> > > Clearly I have a run away or hidden process using resources.
> > > Although all my system run kde 3.5.8 only one shows this problem.
> >
> > vmstat is your friend here. It's all in the man page, so use it and
> > narrow down the process that's blocking. Maybe you have a threading
> > race condition or similar.
>
> # vmstat
> procs -----------memory---------- ---swap-- -----io---- -system--
> ----cpu---- r  b   swpd   free   buff  cache   si   so    bi    bo
> in   cs us sy id wa 0  0      0 847368 224736 403404    0    0    26
>   12  172  251  1  0 98  1

According to this, that machine is sitting there doing nothing. So I see 
two maybe three possibilities:

uptime and top are talking shit (happens way more often than one might 
assume)

your machine is indeed hacked and trojaned, but the script kiddie forgot 
to upload a modified top and uptime (highly unlikely - someone who 
knows to replace vmstat will certainly replace top and uptime)

your kernel scheduler has a bizarre view of life. This is most likely, 
I'd say you have a collection of settings that cause the kernel to 
collect it's utilization stats at precisely the moment when it really 
does do something useful. I saw someone rag poor Ingo two months back 
on lkml with a similar thing. Turns out the user was right.

What are your relevant settings of things like:
cpu scheduler (not i/o scheduler)
timer freq
tickless kernel?



-- 
Alan McKinnon
alan dot mckinnon at gmail dot com

-- 
gentoo-user@l.g.o mailing list