Hard drive noises

[Giles Orr]

On 7 January 2011 14:35, D. Hugh Redelmeier <hugh-pmF8o41NoarQT0dZR+AlfA at public.gmane.org> wrote:
> Here's something I sent my family almost two years ago.  It might or
> might not be relevant.
>
> I've slightly edited it.  It deserves more but I don't have time at
> the moment.
>
> Subject: notebook hard disk stress on Ubuntu
> Date: Wed, 18 Mar 2009 14:14:10 -0400 (EDT)
>
> It seems as if there is a problem with some notebooks and Ubuntu.  A
> notebook's hard drive heads can be "parked" at a furious rate causing
> the drives to wear out prematurely.  Probably other Linux distros have
> this problem, but that doesn't matter to you.
>
> I don't know why there is no problem with desktop computers.
>
> I've read about the problem in these places:
>  https://wiki.ubuntu.com/DanielHahler/Bug59695#Affected hardware
>  https://launchpad.net/ubuntu/+source/acpi-support/+bug/59695
>  https://wiki.ubuntu.com/PowerManagement
>
> Look at the output of smartctl to see if this is a problem for you.
>
> Disks these days keep statistics of various times and report them
> through an interface called SMART (an acronym).  The Linux program to
> deal with SMART is called smartctl.
>
> Here's what I would like you to do.  At your convenience.
>
> 1) make sure that you have smartctl installed
>        sudo smartctl --version
>
>   If it isn't installed, Ubuntu will say so and tell you how to
>   install it:
>        sudo apt-get install smartmontools
>
> 2) capture the output of smartctl in a file called smartctl.LOG01:
>        sudo script smartctl.LOG01
>        date
>        smartctl -a /dev/sda
>        exit
>
> 3) look at the line starting with "193 Load_Cycle_Count".  Compare it
>   with other numbers like power cycles (I don't remember the name of
>   the attribute) and power on hours.
>
> 4) It would be interesting to see how things evolve.  Could you do steps
>   2 and 3 again after you have used your notebook for a few hours more?
>   Change all "smartctl.LOG01" to "smartctl.LOG02" so that we don't lose
>   the old one.
>
> smartctl prints out a lot of information.  Here is the most
> interesting line from redact, Mom's notebook:
>
> 193 Load_Cycle_Count        0x0012   092   092   000    Old_age   Always       -       83455
>
> This says that the drive has put itself into and out of rest state
> 83455.  This is apparently likely to be wearing out the drive.
>
> To put that number in perspective, the machine has been powered on 517
> times and has been on for a total of 1243 hours.  So these load cycles
> are happening at a rate of more than one a minute (including the vast
> majority of the time when the computer is sitting there powered on but
> unused).

Thank you Hugh!

When testing yesterday, I discovered that the drive had been
power-cycled 28 times, but load-cycled 5100 times.  The Debian system
had set itself up such that it was load-cycling the drive every 20
seconds.  Load cycle life on most drives is in the neighbourhood of
600,000 and it would take a long time to get there (five to ten years
with my use-pattern), but I suspect that behaviour accelerated the
death of the previous drive.  I added this:

/dev/sda {
    apm = 254
}

to /etc/hdparm.conf and the problem was fixed.

Ubuntu appears to have addressed the problem in Lucid.  Debian doesn't
seem to have done anything about it (I got the most recent updates and
still had the problem before applying a fix myself), and that my OS
endangers my hardware (even in what they refer to as "a very small
number of cases") disgusts me.

-- 
Giles
http://www.gilesorr.com/
gilesorr-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists