Re: [gentoo-user] [OT] HDD I/O and UI responsiveness

[Boyd Stephen Smith Jr.]

On Wednesday 29 November 2006 22:17, "Vladimir G. Ivanovic" 
<vgivanovic@×××××××.net> wrote about 'Re: [gentoo-user] [OT] HDD I/O and 
UI responsiveness':
> Boyd Stephen Smith Jr. wrote:
> > On Tuesday 28 November 2006 17:45, Dale <dalek@××××××××××.net> wrote
> > about 'Re: [gentoo-user] [OT] HDD I/O and UI responsiveness':
> >> I read somewhere that they are trying to 'nice' the drive usage like
> >> they do the CPU.  That may help if you can find it and enable it.  I
> >> think it is in some of the very new kernels if I read it correctly.
> >> Sorry, I didn't bookmark it.  < slaps hand >
> >
> > There's the ionice utility provided by...
> > <Running equery>
> > ...sys-process/schedutils.
>
> Isn't this just masking the problem?

Yes and no.

> I don't see the CPU pegged, so why should other applications be
> unresponsive?

Because they aren't waiting to be scheduled on the CPU.  They are waiting 
to be scheduled for I/O against a certain device.  ionice will cause a 
process to be more likely to have interruptable I/O operations 
interrupted, allowing other non-ioniced processes to access the disk(s) 
with less latency.

Dumping the contents of /dev/null to disk will slow other processes down 
even though your CPU will be used very little, because all your I/O is 
consumed.

It is "masking" the problem that you are running out of I/O 
bandwidth / "latency" / "cycles".  Of course, that's not actually that 
rare; I have a 6 disk RAID-5 array, and I can still make playing video 
from disk stutter, by performing I/O intensive tasks.

-- 
"If there's one thing we've established over the years,
it's that the vast majority of our users don't have the slightest
clue what's best for them in terms of package stability."
-- Gentoo Developer Ciaran McCreesh