I`m sorry folks, i dont quite recall whether i poked lkml with that,
but here it is:
2.4.13, reiserfs
i have a disk access _every_ 5 sec, unregarding the system load,
24x7x365, so i suppose while it doesnt hurts me, it hurts folks with power
bound boxes...
I must add that i `m experiencing this on -ac tree too, adn this is true
as far as my memory goes... (in the kernel-version context i mean)
cheers, Samium Gromoff
Hi!
> > > > i have a disk access _every_ 5 sec, unregarding the system load,
> > > > 24x7x365, so i suppose while it doesnt hurts me, it hurts folks with power
> > > > bound boxes...
>
> That's a kernel daemon called kupdated. Under Linux buffers are flushed
> every 5 seconds (I don't like this myself, it should be triggered by
> something dependant on free mem, dirty buffers, disk access, etc. but
> not time, this doesn't scale.
>
> Under 2.2 you can try the noflushd package - perhaps it works on 2.4, I
> haven't tried. It works more or less.
noflushd does work on 2.4
Pavel
--
Philips Velo 1: 1"x4"x8", 300gram, 60, 12MB, 40bogomips, linux, mutt,
details at http://atrey.karlin.mff.cuni.cz/~pavel/velo/index.html.
" Dominik Kubla wrote:"
>
> On Thu, Nov 08, 2001 at 01:51:05AM +0300, Samium Gromoff wrote:
> > I`m sorry folks, i dont quite recall whether i poked lkml with that,
> > but here it is:
> > 2.4.13, reiserfs
> > i have a disk access _every_ 5 sec, unregarding the system load,
> > 24x7x365, so i suppose while it doesnt hurts me, it hurts folks with power
> > bound boxes...
> > I must add that i `m experiencing this on -ac tree too, adn this is true
> > as far as my memory goes... (in the kernel-version context i mean)
> >
> > cheers, Samium Gromoff
>
> That's a FAQ: you have cron running...
hehe...
i`ve actually compiled cron only a week ago, so its not an issue ;)
(i havent had cron before on my homemmade linux)
i mean i`m not this lame, and i told *no_system_load*... :-)
>
> Dominik
> --
> ScioByte GmbH Zum Schiersteiner Grund 2 55127 Mainz (Germany)
> Phone: +49 700 724 629 83 Fax: +49 700 724 629 84
>
> GnuPG: 717F16BB / A384 F5F1 F566 5716 5485 27EF 3B00 C007 717F 16BB
>
cheers, Samium Gromoff
le jeu 08-11-2001 ? 06:02, Samium Gromoff a ?crit :
> > > i have a disk access _every_ 5 sec, unregarding the system load,
> > > 24x7x365, so i suppose while it doesnt hurts me, it hurts folks with power
> > > bound boxes...
That's a kernel daemon called kupdated. Under Linux buffers are flushed
every 5 seconds (I don't like this myself, it should be triggered by
something dependant on free mem, dirty buffers, disk access, etc. but
not time, this doesn't scale.
Under 2.2 you can try the noflushd package - perhaps it works on 2.4, I
haven't tried. It works more or less.
Xav
> > On Thu, Nov 08, 2001 at 01:51:05AM +0300, Samium Gromoff wrote:
> > > I`m sorry folks, i dont quite recall whether i poked lkml with that,
> > > but here it is:
> > > 2.4.13, reiserfs
> > > i have a disk access _every_ 5 sec, unregarding the system load,
> > > 24x7x365, so i suppose while it doesnt hurts me, it hurts folks with power
> > > bound boxes...
> > > I must add that i `m experiencing this on -ac tree too, adn this is true
> > > as far as my memory goes... (in the kernel-version context i mean)
Did you mount your partitions with "noatime"?
This (and not keeping Mozilla open) does the trick on my laptop, the
harddisk goes off after about 30 seconds and if I don't do anything local,
it stays off.
(Debian unstable, Kernel 2.4.12-ac6 currently)
c'ya
sven
--
The Internet treats censorship as a routing problem, and routes around it.
(John Gilmore on http://www.cygnus.com/~gnu/)
> i have a disk access _every_ 5 sec, unregarding the system load,
I used to have this problem sometimes when my CD-ROM drive
had no disk in it. I would get an infinite series of
"disk change detected" messages in the syslog. The writes
to the syslog kept the disk from spinning down.
Thomas
On Thu, 8 Nov 2001, Samium Gromoff wrote:
> " Dominik Kubla wrote:"
> >
> > On Thu, Nov 08, 2001 at 01:51:05AM +0300, Samium Gromoff wrote:
> > > I`m sorry folks, i dont quite recall whether i poked lkml with that,
> > > but here it is:
> > > 2.4.13, reiserfs
> > > i have a disk access _every_ 5 sec, unregarding the system load,
> > > 24x7x365, so i suppose while it doesnt hurts me, it hurts folks with power
> > > bound boxes...
> > > I must add that i `m experiencing this on -ac tree too, adn this is true
> > > as far as my memory goes... (in the kernel-version context i mean)
> > >
> > > cheers, Samium Gromoff
> >
> > That's a FAQ: you have cron running...
> hehe...
> i`ve actually compiled cron only a week ago, so its not an issue ;)
> (i havent had cron before on my homemmade linux)
>
> i mean i`m not this lame, and i told *no_system_load*... :-)
Not to mention that 5 second obviously implies it can't be cron (given
that it runs once a minute), and points to bdflush, since that is the
flushing interval.
Now, on my home box, If I run /bin/sync, the disk seeks, whether there is
data to write or not. Before you point to me that I need to mount noatime
- already done :) I think it also did this when bdflush told it to sync,
but after installing noflushd (and hacking the b0rked init scripts),
all was fine. At least, until I installed ext3 :(
--
TimC -- http://www.physics.usyd.edu.au/~tcon/
Some witty text here,
can be any number of lines
long