[SOLVED] [Old] WD Green Power drives may kill themselves !!

[carl]

For you who wonder how to make the script run on boot, here is an short howto:

Execute following as root:

Code: Select all

cat <<EOF > /usr/local/bin/prevent-unload
#!/bin/sh
while [ 1 ]; do
    ls -l /home  > /home/.wd-prevent-unload
    sleep 5
done
EOF
chmod 755 /usr/local/bin/prevent-unload

Add following to /etc/rc.local (before the exit 0)

Code: Select all

/usr/local/bin/prevent-unload &

you can issue "/etc/init.d/rc.local start"

/Carl

[RobV]

Thanks, it works great!
RobV

[carl]

Thanks, it works great!
RobV

I made a minor error in my post, forgot to change $HOME to /home ($HOME won't be set on init, or it will be /root).

[RandomUsername]

Everything I've read on this subject suggests my disk should not be affected by this problem. It was only bought towards the end of last year and is a WD10EADS model number (three platters - this problem only affected WDEACS, four platter models apparently).

Yet after three months my load cycle count is over 110000 and that's with the drive being accessed most of the time (I hate to think how high it would be if the disk had been idle).

Does anyone know if this could be indicative of another problem? I haven't seen any other problems from the disk at all.

Cheers.

[RandomUsername]

For anyone who stumbles across this thread in the future.

I believe it was suggested earlier on in the thread that WD wouldn't honour a warranty on a drive with a high load cycle count. With that in mind I sent the above query to WD.

Here's the reply:

WD:

We do understand you concern. The drive will increase the load/unload count when it parks itself during idle periods. It does not wait for the hard drive to initiate an idle command since this allows the drive to secure itself against any potential or vibration damage. This allows us to rate the drive at higher shock tolerances; however, this should not affect the drive's performance or longevity.

Me:

Can you then confirm then that WD would still honour any warranty on the drive regardless of the load cycle count?

I was under the impression that if it was especially high you would void the warranty.

WD:

Of course we (WD) will honor the warranty on your drive regardless of the Load Cycle Count. In fact, the drive is only inspected for physical damage when sent in for warranty replacement.

Sorted.

[johannes]

Thanks, nice piece of information. Also, We still don't have a single report of any drives crashing that had a high LCC, so it might not even be a real problem.

Anyways, as I stated earlier we are willing to replace drives with a high LCC, to calm worried souls.

[nitram]

Oeps... forgot about that script after last reboot, it's running again and it should start now after a reboot as well.
In the mean time cycles are sky high:

Device Model: WDC WD1000FYPS-01ZKB0
9 Power_On_Hours 0x0032 086 086 000 Old_age Always - 10264
193 Load_Cycle_Count 0x0032 001 001 000 Old_age Always - 616326
194 Temperature_Celsius 0x0022 114 104 000 Old_age Always - 38

It's still running without problems. It is also in raid so if he crashes it should be no problem with dataloss after replacing the disk, is it?

Martin

[lelle]

I've successfully used the WDIDLE3 tool to disable the heads from parking (that is, set the timeout to 62 minutes).

I've tried this on two disks, a 500GB WDC WD5000AACS-00ZUB0 and a 2TB WDC WD20EARS.

You can download this utility from here: http://support.wdc.com/product/download ... 13&lang=en. Note that it should only be used for some specific drive models (see under Caution) but it actually works on other WDC disks. Use it at your own risk.

Since WDIDLE3 is a Windows executable, you'll need to temporary move the Bubba drive to a windows computer and run the utilty there. I've used the instructions from http://www.xlr8yourmac.com/tips/Disable ... arking.htm to create a boot cd containing the WDIDLE3. Before starting WDIDLE3, disconnect all drives except the Bubba drive.

[willem2]

Above thread is very confusing - after some months of use on my BubbaTwo:

Code: Select all

  3 Spin_Up_Time          0x0027   131   126   021    Pre-fail    Always - 6441
  9 Power_On_Hours      0x0032   093   093   000    Old_age   Always - 5164
 12 Power_Cycle_Count 0x0032   100   100   000    Old_age   Always - 30
193 Load_Cycle_Count  0x0032   001   001   000    Old_age   Always - 665250

Built in HD:

Code: Select all

Device Model:     WDC WD10EARS-00Y5B1
Serial Number:    WD-WCAV55192277
Firmware Version: 80.00A80
User Capacity:    1.000.204.886.016 bytes

IMHO the LCC (665250) is extremely high:

• do I have to worry about it?
• proposed solution?

[RandomUsername]

Above thread is very confusing - after some months of use on my BubbaTwo:

Code: Select all

  3 Spin_Up_Time          0x0027   131   126   021    Pre-fail    Always - 6441
  9 Power_On_Hours      0x0032   093   093   000    Old_age   Always - 5164
 12 Power_Cycle_Count 0x0032   100   100   000    Old_age   Always - 30
193 Load_Cycle_Count  0x0032   001   001   000    Old_age   Always - 665250

Built in HD:

Code: Select all

Device Model:     WDC WD10EARS-00Y5B1
Serial Number:    WD-WCAV55192277
Firmware Version: 80.00A80
User Capacity:    1.000.204.886.016 bytes

IMHO the LCC (665250) is extremely high:

• do I have to worry about it?
• proposed solution?

All the info you need is here. To answer your questions; you don't need to worry about it as WD don't see it as a problem and would honour the warrranty on your drive regardless of LCC. Further, there have been no reported problems arising from this issue. If you are really concerned there is a script that someone has written that keeps your drive from spinning down.

[misiu_mp]

This not at all "solved" or "old". My brand new bubba3 (january 2012) has already 40800 LoadCycles at 318 hours of PowerOn. Thats 128 cycles / hour!


While the typical load cycle limit for HDs is normally 300000-600000, I found somewhere on that forum
http://community.wdc.com/t5/Desktop/Gre ... 731/page/7,
that the WD greens are actually rated to at least 1000000 load cycles. With 128cycl/hour I would have reached that amount after only 325 days!
Surly, even if there are no reports of failing disks, the risk of such high LoadCycle count is not worth taking?

Apparently this feature is optimized for usage patters characteristic for certain OSs and file systems, none of them being those used by the Bubba server.

The write-sleep loop script seems to fix the problem, but that's just a nasty work-around.
A better solution would be to try to make the system use the disk less often (laptop_mode, less logging).
A even better solution would be not to use the WD Green drives, or any other drive with a weird and difficult to control power-saving behavior.
I expect the Excito team to solve this problem.

ps. The disk is WDC WD20EARS-00MVWB0.

[johannes]

You are absolutely right. After a long series of events on this matter, where we thought WD fixed the issue, it keeps popping up again on their new revision of drive firmware, and I know that Load Cycle Count is racing again in some drives shipped now, in B3's.

After almost a year of putting pressure on them they sent us a tool compiled for ARM now in december, so with the 2.4 release for B3, this tool will automatically run and disable the unneccessary load/unload behaviour (change the timer from 8 seconds to 5 minutes, which in practice makes it never trigger).

The 2.4 is just around the corner, but if you can't wait you can upgrade against our "testing" repo right now (hugo). Then do a power-cycle and voila, LCC won't move.

[oliver]

The 2.4 is just around the corner, b

Johannes, when will this release be available ?

Thanks

Oliver

[johannes]

Cant say for sure, hopefully a few days just.

[misiu_mp]

That's great news.
Can you spare more info about that utility for WD disks?
Is it a proprietary linux arm binary version of the dos wdidle tool? If it will find its way to our servers, what freedoms will we have with it (i.e. what license will it come with)?