pm-utils for F8 and Advanced power management
 

I am trying to fix the Load_Cycle_Count
bug that is increasing at a very fast pace each day
on my laptop running* F8. Suddenly, I realized that apparently the
version of pm-utils for F9 has a kook to deal with it:

99hd-apm-restore.hook and
/etc/pm-utils-hd-apm-restore.conf


Why was not this ported to F8?

The problem, as I understand, is quiet serious,
and can kill the drive in one or two years.


F8 is still maintained, right?

--
Paulo Roma Cavalcanti
LCG - UFRJ
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

pm-utils for F8 and Advanced power management
 

2008/7/6 Paulo Cavalcanti <promac@gmail.com>:
> I am trying to fix the Load_Cycle_Count
> bug that is increasing at a very fast pace each day
> on my laptop running F8. Suddenly, I realized that apparently the
> version of pm-utils for F9 has a kook to deal with it:
>
> 99hd-apm-restore.hook and
> /etc/pm-utils-hd-apm-restore.conf
>
>
> Why was not this ported to F8?
>
> The problem, as I understand, is quiet serious,
> and can kill the drive in one or two years.

Model Family: Western Digital Scorpio family
Device Model: WDC WD600VE-11KWT0
Firmware Version: 01.03K01

Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE
UPDATED WHEN_FAILED RAW_VALUE
4 Start_Stop_Count 0x0032 100 100 000 Old_age
Always - 357
5 Reallocated_Sector_Ct 0x0033 199 199 140 Pre-fail
Always - 8
9 Power_On_Hours 0x0032 094 094 000 Old_age
Always - 4414
10 Spin_Retry_Count 0x0012 100 100 051 Old_age
Always - 0
12 Power_Cycle_Count 0x0032 100 100 000 Old_age
Always - 199
192 Power-Off_Retract_Count 0x0032 200 200 000 Old_age
Always - 192
193 Load_Cycle_Count 0x0032 065 065 000 Old_age
Always - 406672
194 Temperature_Celsius 0x0022 110 077 000 Old_age
Always - 33
196 Reallocated_Event_Count 0x0032 199 199 000 Old_age
Always - 1

F-cking great. I don't think this drive is even a year old, and it's
up to 400000 already! And has reallocated sectors? This is a plain,
un-tweaked F9 on an eMachines m6805, which is run entirely off AC
since the battery is no good. Shouldn't we do something about this? It
would be easy enough to automatically detect a rapidly increasing
Load_Cycle_Count and alert the user. Or just fix it automagically.

I just put '/sbin/hdparm -B 254 /dev/sda' into my rc.local to control
the damage. I shouldn't have to do this. No finger pointing, the plain
fact is Fedora should Just Work, no matter what retarded things the
BIOS vendor or HD manufacturer does.

We really do need some kind of thing set up to monitor SMART and alert
the user via notification-daemon. I already had to replace the drive
in this thing because the previous one got toasted due to a failed CPU
heatsink roasting the whole system. It sure would have been nice to
get a "Warning! Your hard drive appears to be overheating!" alert.

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

pm-utils for F8 and Advanced power management
 

On Tuesday 08 July 2008 18:44:17 Callum Lerwick wrote:

> Vendor Specific SMART Attributes with Thresholds:
> ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE
> UPDATED WHEN_FAILED RAW_VALUE

> 193 Load_Cycle_Count 0x0032 065 065 000 Old_age
> Always - 406672

> F-cking great. I don't think this drive is even a year old, and it's
> up to 400000 already! And has reallocated sectors? This is a plain,

Are you sure that you know what the value means? My drive reports this:

193 Load_Cycle_Count 0x0032 071 071 000 Old_age
Always - 2990488599305

I do not believe that the value is exactly the number of load cycles that my
drive already had.

Regards,
Till
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

pm-utils for F8 and Advanced power management
 

2008/7/8 Till Maas <opensource@till.name>:

On Tuesday 08 July 2008 18:44:17 Callum Lerwick wrote:



> Vendor Specific SMART Attributes with Thresholds:

> ID# ATTRIBUTE_NAME * * * * *FLAG * * VALUE WORST THRESH TYPE

> UPDATED *WHEN_FAILED RAW_VALUE



> 193 Load_Cycle_Count * * * *0x0032 * 065 * 065 * 000 * *Old_age

> Always * * * - * * * 406672



> F-cking great. I don't think this drive is even a year old, and it's

> up to 400000 already! And has reallocated sectors? This is a plain,



Are you sure that you know what the value means? My drive reports this:



193 Load_Cycle_Count * * * *0x0032 * 071 * 071 * 000 * *Old_age

Always * * * - * * * 2990488599305



I do not believe that the value is exactly the number of load cycles that my

drive already had.



My laptop is less than a month old, and it is running F8.

* 5 Reallocated_Sector_Ct** 0x0033** 100** 100** 024*** Pre-fail* Always****** -****** 8589934592000
* 9 Power_On_Minutes******* 0x0032** 100** 100** 000*** Old_age** Always****** -****** 167h+02m

*12 Power_Cycle_Count****** 0x0032** 100** 100** 000*** Old_age** Always****** -****** 89
191 G-Sense_Error_Rate***** 0x0012** 100** 100** 000*** Old_age** Always****** -****** 2
192 Power-Off_Retract_Count 0x0032** 100** 100** 000*** Old_age** Always****** -****** 7

193 Load_Cycle_Count******* 0x0032** 100** 100** 000*** Old_age** Always****** -****** 5325

The number of reallocated sectors is 000 (forget the high order bits for fujtisu).
The load cycle 5325. This number is correct. When I started to use hdparm in rc.local and in /usr/lib/pm-utils/sleep.d/ (the parameter is lost when returning from hibernation

and suspend mode), it just increments a single unit every time I shut down or suspend.
This is the way it should be in my opinion.


According to this post

http://www.theinquirer.net/gb/inquirer/news/2007/10/31/ubuntu-eats-lappy-hard-drive


(look at the end) the reason is that ext3 commits each 5 seconds, awaking the disk.


--
Paulo Roma Cavalcanti
LCG - UFRJ
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

pm-utils for F8 and Advanced power management
 

2008/7/8 Till Maas <opensource@till.name>:
> Are you sure that you know what the value means? My drive reports this:

Yes I've monitored it, it increments every few seconds when the system
is idle. If I do something to keep the disk active, it does not
increment. If I set the power save level up to 254, it stops
incrementing. I'm certain the value is accurate.

> 193 Load_Cycle_Count 0x0032 071 071 000 Old_age
> Always - 2990488599305
>
> I do not believe that the value is exactly the number of load cycles that my
> drive already had.

Google the interwebs for more information. Some drives seem to report
impossibly high, thus completely bogus numbers. It would appear yours
is one of them.

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

pm-utils for F8 and Advanced power management
 

On Tue, 2008-07-08 at 17:49 -0500, Callum Lerwick wrote:
>
> Yes I've monitored it, it increments every few seconds when the system
> is idle. If I do something to keep the disk active, it does not
> increment. If I set the power save level up to 254, it stops
> incrementing. I'm certain the value is accurate.

In some testing tonight, the wakeups appear to be from kjournald,
perhaps on the order of the atime updates. When I remount my /
and /boot filesystems with noatime the very frequent writes go away.
Now I only see writes every 20 seconds or so, caused by various parts of
software that I'm trying to track down.

I think it's really two things going on. 1) bioses are setting the
harddrive to be very aggressive about parking, and 2) we're waking up
the disk /way/ too frequently.

If anybody would like to help me track down causes of 2 and file bugs,
that would be awesome!

--
Jesse Keating
Fedora -- Freedom² is a feature!
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

pm-utils for F8 and Advanced power management
 

On Tue, Jul 08, 2008 at 09:53:10PM -0400, Jesse Keating wrote:

> If anybody would like to help me track down causes of 2 and file bugs,
> that would be awesome!

Try echoing 1 >/proc/sys/vm/block_dump and then check dmesg. Stopping
syslog before doing this might be a good plan for obvious reasons...

--
Matthew Garrett | mjg59@srcf.ucam.org

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

pm-utils for F8 and Advanced power management
 

On Wed, Jul 9, 2008 at 3:57 AM, Matthew Garrett <mjg59@srcf.ucam.org> wrote:

On Tue, Jul 08, 2008 at 09:53:10PM -0400, Jesse Keating wrote:



> If anybody would like to help me track down causes of 2 and file bugs,

> that would be awesome!



Try echoing 1 >/proc/sys/vm/block_dump and then check dmesg. Stopping

syslog before doing this might be a good plan for obvious reasons...
Putting syslog on tmpfs by default and limiting the total size to something like 5MB would be something to evaluate for the default desktop.



--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

pm-utils for F8 and Advanced power management
 

On Wed, Jul 09, 2008 at 06:32:16AM -0400, Colin Walters wrote:

> Putting syslog on tmpfs by default and limiting the total size to something
> like 5MB would be something to evaluate for the default desktop.

I looked into this, and we're not actually doing too badly - our syslogd
doesn't fsync() after every log entry. I'm broadly in favour of using
tmpfs, though. There's an argument that certain priorities probably want
to stay on disk, but otherwise we can simply sync the tmpfs to disk on
shutdown or reboot.

--
Matthew Garrett | mjg59@srcf.ucam.org

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

pm-utils for F8 and Advanced power management
 

Matthew Garrett (mjg59@srcf.ucam.org) said:
> On Wed, Jul 09, 2008 at 06:32:16AM -0400, Colin Walters wrote:
>
> > Putting syslog on tmpfs by default and limiting the total size to something
> > like 5MB would be something to evaluate for the default desktop.
>
> I looked into this, and we're not actually doing too badly - our syslogd
> doesn't fsync() after every log entry. I'm broadly in favour of using
> tmpfs, though. There's an argument that certain priorities probably want
> to stay on disk, but otherwise we can simply sync the tmpfs to disk on
> shutdown or reboot.

Maybe I'm unusual, but I rarely reboot my laptop cleanly. I'm not sure I'd
want X weeks of logs to go away.

Are there any sorts of mechanisms to do per-directory or per-file writeback
caching, so that /var/log/ could be set to 'sync only every 15 minutes',
or similar?

Bill

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list