bug in ubuntu "kill" hardrive

[Derek Broughton]

Nils Kassube wrote:

> Derek Broughton wrote:
>>
>> I don't think that's true either.  Dapper would have the bug, if it had
>> strigi, beagle, and who knows how many other apps continuously
>> hammering the drives.
> 
> IMHO, the bug has nothing to do with applications running but with a
> setting for the hard disk to unload the head after a short time of no
> activity. 

It has to do with applications _periodically_ writing to the disc - with a
period just a little larger than the HD settings.

> Unfortunately, not all hard disks behave the same. The 
> S.M.A.R.T. setting is interpreted different by different manufacturers
> because it is no real standard. Therefore the delay is different for every
> hard disk type.

The delay is whatever the manufacturer considers to be reasonable for the
hardware - it _should_ be different for every hd type.  It could be set to
any value, but Gutsy would tend to wake up the drive in any case.  If you
set the parameter high enough to not be a problem, Gutsy will simply
_never_ let it sleep.  That's the same as disabling it, but it's plain
wrong.  There's a good reason for the setting to exist.

> I have tested the existance of the bug with Dapper, Gutsy and Windows XP
> Pro when it was mentioned on Slashdot. The short version of the test
> result: I could only verify it with Gutsy. But considering the date of
> the bug report, it probably existed since Feisty. The long version can be
> found in the archive of ubuntu-users and/or kubuntu-users (if you want to
> read some numbers which apply only for my disk).

It's not conceivable that with the same amount and type of I/O that the same
situation wouldn't happen in Dapper - ie, it's not a kernel issue, it's an
I/O issue.
-- 
derek