Disk clicking and in increasing Load_Cycle_Count in laptop with debian/testing
 

Hi,

A few weeks ago I've noticed, that my half year old laptop started to
behave strange.

When left unused, after the screen gets black to save power, the disks
starts to click every few minutes, and after each click the
Load_Cycle_Count (as reported by smartctl -a /dev/sda ) increases by 1.


The laptop is equipped with Western Digital Scorpio WDC
WD7500BPKT-75PK4T0 disk, however similar problem (well, the disk is
much more silent, but Load_Cycle_Count is quickly increasing) occures
also my wife's laptop equipped with Seagate ST9500420AS.


I have found the website:
http://www.thinkwiki.org/wiki/Problem_with_hard_drive_clicking , which
describes the same or similar problem, but it seems, that solutions
described there do not help.


What's interesting - when I tried to run the following command (from
remote console via ssh, so that the screen remains black):


# ( LANG=C; while true ; do date ; smartctl -a /dev/sda | grep
Load_Cycle ; sleep 10 ; done ) > /tmp/rep


to find how often the head get unloaded, I have completely prevented
this effect to occur.


When I increased the delay between calls od smartctl, I stated, that
problem occures ca every 3 minutes.


I've tried to set:
hdparm -B 254 /dev/sda
and
hdparm -B 255 /dev/sda

Below are the results:
root@WZLap:~# hdparm -B 255 /dev/sda
/dev/sda: setting Advanced Power Management level to disabled
APM_level = off
root@WZLap:~# ( LANG=C; while true ; do date ; smartctl -a /dev/sda |
grep Load_Cycle ; sleep 10m ; done )

Sat Apr 14 01:07:06 CEST 2012
193 Load_Cycle_Count 0x0032 168 168 000 Old_age Always
- 97297

Sat Apr 14 01:17:07 CEST 2012
193 Load_Cycle_Count 0x0032 168 168 000 Old_age Always
- 97300

^C
root@WZLap:~# hdparm -B 254 /dev/sda
/dev/sda: setting Advanced Power Management level to 0xfe (254)
APM_level = 254
root@WZLap:~# ( LANG=C; while true ; do date ; smartctl -a /dev/sda |
grep Load_Cycle ; sleep 10m ; done )

Sat Apr 14 01:17:43 CEST 2012
193 Load_Cycle_Count 0x0032 168 168 000 Old_age Always
- 97300

Sat Apr 14 01:27:44 CEST 2012
193 Load_Cycle_Count 0x0032 168 168 000 Old_age Always
- 97303

^C

It is really allarming, as with warranted durability of 600 000
load/unload cycles, at this rate my drive will survive only ca. 3 years!


I've verified, that the set APM level remains untouched (e.g. when set
to 254, and after 20 minutes reported by "hdparm -I /dev/sda": Advanced
power management level: 254 )


The problem occurs even when the laptop is working on AC power.

Does anybody know any good solution?

Of course, I can run "smartctl -a /dev/sda >/dev/null" every 60 seconds
in the background, as it seems to prevent the problem, but it is rather
not the optimal solution ;-)

--
TIA & Regards,
Wojtek


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Archive: 4F88B9DA.2000906@ise.pw.edu.pl">http://lists.debian.org/4F88B9DA.2000906@ise.pw.edu.pl

Disk clicking and in increasing Load_Cycle_Count in laptop with debian/testing
 

On Sat, 14 Apr 2012, wzab wrote:
> When left unused, after the screen gets black to save power, the disks
> starts to click every few minutes, and after each click the
> Load_Cycle_Count (as reported by smartctl -a /dev/sda ) increases by
> 1.

...

> I have found the website:
> http://www.thinkwiki.org/wiki/Problem_with_hard_drive_clicking ,
> which describes the same or similar problem, but it seems, that
> solutions described there do not help.
>
> What's interesting - when I tried to run the following command (from
> remote console via ssh, so that the screen remains black):
>
> # ( LANG=C; while true ; do date ; smartctl -a /dev/sda | grep
> Load_Cycle ; sleep 10 ; done ) > /tmp/rep
>
> to find how often the head get unloaded, I have completely prevented
> this effect to occur.

Well, you'd be keeping the disk _and the SATA link_ busy.

Now, usually hdparm -B 254 can convince the disk to not be so silly as to
destroy itself too fast in an attempt to save power, and you've already
covered that one. So the next probable cause is SATA link power management
causing the disk to decide that, since its PHY is down, it might as well
unload the heads...

> It is really allarming, as with warranted durability of 600 000
> load/unload cycles, at this rate my drive will survive only ca. 3
> years!
>
> I've verified, that the set APM level remains untouched (e.g. when
> set to 254, and after 20 minutes reported by "hdparm -I /dev/sda":
> Advanced power management level: 254 )

Disable SATA link power saving. Does that fix the problem? If it does,
maybe a less aggressive level of SATA link power saving might also do the
trick.

Failing that, you can either disable SATA link power saving entirely and
waste power, or start a hunt (maybe with the help of blocktrace, or the
recently released LTTng 2.0 toolkit), to get rid of every crap that pesters
the disk at regular intervals while the box is idle.

Alternatively, you could also make sure something pesters the disk often
enough for it to not unload heads.

If it is not SATA link power saving, I am out of ideas.

--
"One disk to rule them all, One disk to find them. One disk to bring
them all and in the darkness grind them. In the Land of Redmond
where the shadows lie." -- The Silicon Valley Tarot
Henrique Holschuh


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20120414021718.GA10926@khazad-dum.debian.net">http://lists.debian.org/20120414021718.GA10926@khazad-dum.debian.net

Disk clicking and in increasing Load_Cycle_Count in laptop with debian/testing
 

>> to find how often the head get unloaded, I have completely prevented
>> this effect to occur.

Notice that you have 2 problems:
1- Unloading.
2- Reloading.
>From where I stand, the unloading is normal, harmless, and even
desirable so the real problem is: why is it re-loaded every 3 minutes?
I'd recommend you use /proc/sys/vm/block_dump to try and figure it out.


Stefan


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: jwv4nsn11ns.fsf-monnier+gmane.linux.debian.user@gnu.org">http://lists.debian.org/jwv4nsn11ns.fsf-monnier+gmane.linux.debian.user@gnu.org

Disk clicking and in increasing Load_Cycle_Count in laptop with debian/testing
 

wzab wrote at 2012-04-13 18:42 -0500:
> A few weeks ago I've noticed, that my half year old laptop started
> to behave strange.
> When left unused, after the screen gets black to save power, the disks
> starts to click every few minutes, and after each click the
> Load_Cycle_Count (as reported by smartctl -a /dev/sda ) increases by
> 1.

You might want to look at the idle3-tools package and
http://idle3-tools.sourceforge.net

Disk clicking and in increasing Load_Cycle_Count in laptop with debian/testing
 

On Fri, 13 Apr 2012, green wrote:
> wzab wrote at 2012-04-13 18:42 -0500:
> > A few weeks ago I've noticed, that my half year old laptop started
> > to behave strange.
> > When left unused, after the screen gets black to save power, the disks
> > starts to click every few minutes, and after each click the
> > Load_Cycle_Count (as reported by smartctl -a /dev/sda ) increases by
> > 1.
>
> You might want to look at the idle3-tools package and
> http://idle3-tools.sourceforge.net

Added to http://www.thinkwiki.org/wiki/Problem_with_hard_drive_clicking,
thanks.

Although wzab mentioned problems with a Seagate drive as well, so
Intellipark might not be the root cause...

--
"One disk to rule them all, One disk to find them. One disk to bring
them all and in the darkness grind them. In the Land of Redmond
where the shadows lie." -- The Silicon Valley Tarot
Henrique Holschuh


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20120414034135.GA20159@khazad-dum.debian.net">http://lists.debian.org/20120414034135.GA20159@khazad-dum.debian.net

Disk clicking and in increasing Load_Cycle_Count in laptop with debian/testing
 

>> to find how often the head get unloaded, I have completely prevented
>> this effect to occur.

>Notice that you have 2 problems:
>1- Unloading.
>2- Reloading.
>From where I stand, the unloading is normal, harmless, and even
>desirable so the real problem is: why is it re-loaded every 3 minutes?
>I'd recommend you use /proc/sys/vm/block_dump to try and figure it out.

OK. Unloading is normal harmless and so on... However what's strange in my case -
unloading occures, wehen I USE my laptop remotely.
(As my laptop has much higher performance than my desktop workstation, while the
workstation has better keyboard, mouse and monitor, I very often work via
remote ssh or X session).

When working remotely, the laptops screen remains black, and when I e.g.
edit my VHDL sources, loading may be simply caused by the autosave option in the editor.

So the problem is why there is a difference between work on local keyboard
(no unloading/loading due to entering the idle state) and remote work?
I'd like to prevent the disk from entering the idle state, when working remotely...

--
Wojtek


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20120414102957.GA4770@wzab.nasz.dom">http://lists.debian.org/20120414102957.GA4770@wzab.nasz.dom

Disk clicking and in increasing Load_Cycle_Count in laptop with debian/testing
 

On Sb, 14 apr 12, 12:29:57, wzab wrote:
>
> So the problem is why there is a difference between work on local keyboard
> (no unloading/loading due to entering the idle state) and remote work?
> I'd like to prevent the disk from entering the idle state, when working remotely...

Probably the keyboard/mouse events keep the laptop active.

Kind regards,
Andrei
--
Offtopic discussions among Debian users and developers:
http://lists.alioth.debian.org/mailman/listinfo/d-community-offtopic