A word of advice on Western Digital "Green Power" Disks

Shapeshifter · 2009-06-07 10:33:15

I was looking for a new hard disk as I'm probably going to buy 2x 1TB disks and run them mirrored to store data. Among others, I looked at the Western Digital "Green Power" hard disks, because they run at 5400-7200 rpm and consume about half the power of a regular drive. There are lots of reviews on the net (at anandtech for example).

In any case, these drives have a serious problem when used with Linux:
The "Intellitpark" feature, that parks the seeker when idle has a much lower time out than the Kernel flush timer (which is 30 seconds by default). This causes the seeker to be parked like twice a minute, which will cause the drive to exceed it's specified 300'000 parks in less than a year!

There are a couple of workarounds for this problem:
- There's a WD DOS tool called wdidle3.exe floating around the net which you can use to increase the parking timer to 25.5 seconds
- Decreasing the dirty_expire_centisecs and dirty_writeback_centisecs of your kernel to timings of 3000/500 would also solve the problem, though you usually don't want to use values as low.

If you already have such a drive, you can check the drives state using smartctl (it's in extra/smartmontools). Load_Cycle_Count is the value you're interested in. If you don't have such a drive, I'd recommend not buying one until a firmware update has been released that fixes the problem. Samsung and Seagate now also offer such low power hard disks, and it seems like for Linux users, those are the better option at the moment.

(I'm probably going to buy some Samsung HD103SI, F2 harddisk. They run at 5400 rpm, but consist of only 2x 500GB platters which means increased density and as such increased speed. It's actually almost as fast as the 4x250GB 7200rpm harddisk from samsung. This is the "F2" model, the earlier "F1" has 3x333gb and as such is slower than the F2 at the same 5400 rpm. Seeking time will always be slower with 5400 rpm btw.)

edit: The problem was first discovered on the kerneltrap mailing list: http://kerneltrap.org/mailarchive/linux … 10/1396844

Last edited by Shapeshifter (2009-06-07 12:04:52)

Bogart · 2009-06-07 11:39:48

Thanks for the research and the info. Is always good to have this kind of info in the forum for anyone seeking for it.

I heard about the problem before and also that when someone contacted WD about it they answered that they don't support Linux. Perfect answer for losing customers.

Seagate probably cares much more about Linux (especially for the server market). Samsung probably doesn't care, but if the drives themselves have no problem with Linux, Samsung usually makes drives that are quiet and perform well for desktop tasks, so they should be a good option.

.:B:. · 2009-06-07 11:43:54

As for your drives' rotation speed, you know they do not switch between 5400 and 7200 RPM right? A lot of people seem to think that (in error).

Since I have three such drives (at 5400 RPM) I will check your suggestions, this is very interesting. I have this as default values though on my server (didn't play with both values yet, not now, not before):

[root@amalthea stijn]# cat /proc/sys/vm/dirty_expire_centisecs /proc/sys/vm/dirty_writeback_centisecs 
2999
1499

This is the load cycle count for my drives:

[root@amalthea stijn]# smartctl -a /dev/sda|grep Load_Cycle_Count
193 Load_Cycle_Count        0x0032   200   200   000    Old_age   Always       -       242
[root@amalthea stijn]# smartctl -a /dev/sdb|grep Load_Cycle_Count
193 Load_Cycle_Count        0x0032   200   200   000    Old_age   Always       -       245
[root@amalthea stijn]# smartctl -a /dev/sdc|grep Load_Cycle_Count
193 Load_Cycle_Count        0x0032   200   200   000    Old_age   Always       -       242

Shapeshifter · 2009-06-07 12:09:13

Well I added the link to the kernel trap mailing list, where I think the problem was discovered. In any case, there have been various workarounds presented in this thread on the Synology forums, one if which might be preventing your drives from parking the seeker a lot. All of them involve creating traffic to and from the drive so that it wont park the seeker. Or maybe they already fixed the firmare...

Yeah, the "5400-7200" got me fooled as well They just sell drives with both speeds.

Last edited by Shapeshifter (2009-06-07 12:11:59)

Robertek · 2009-06-07 12:20:57

WD5000AACS

Still no problem here ( 2y old disk )

robert robert-server:~ [2013]% sudo smartctl --all /dev/sda | grep Load_Cycle
193 Load_Cycle_Count        0x0032   001   001   000    Old_age   Always       -       881422

I also increased the vm settings, some time ago when I found this info.

Im also looking for some samsung 1T

Last edited by Robertek (2009-06-07 12:22:17)

Procyon · 2009-06-07 12:23:22

Does this apply to the WD MyBook 1TB too? They also park though I don't know how fast (smartctl isn't working on it)

.:B:. · 2009-06-07 12:26:32

Thanks for the link to that Synology topic. It seems not all revisions exhibit this problem (I have three WDC WD10EACS-65D6B0 1 TB HDs). I got my HDs on the 5th of november 2008.

So you think those values look fine? (I have honestly no idea what they should be).

Also, it seems the Start_Stop_Count is identical (as is mentioned in the Synology thread):

[root@amalthea stijn]# smartctl -a /dev/sda|grep Start_Stop_Count
  4 Start_Stop_Count        0x0032   100   100   000    Old_age   Always       -       242
[root@amalthea stijn]# smartctl -a /dev/sdb|grep Start_Stop_Count
  4 Start_Stop_Count        0x0032   100   100   000    Old_age   Always       -       245
[root@amalthea stijn]# smartctl -a /dev/sda|grep Start_Stop_Count
  4 Start_Stop_Count        0x0032   100   100   000    Old_age   Always       -       242

I just found in this thread in the Synology forum a nice list of the most common symptoms of this problem, among which

When there is no access to the NAS (idle), the LCC increases between +50 and +120 per hour.

Given I have my system up daily for at least a few hours (not 24/7) I think a value of ~240 is not a bad one. Looks like my model is not affected...

Same topic also says:

# It seems that disks with version WD10EACS-00D6B0 (Nov 08) or later no longer show high LCC. Instead they show the START_STOP_COUNTER value as LCC. Question remains, if WD did really fix the problem or just mask it away by no longer showing the high LCCvalues to us.

* B knocks on wood.

A final quote from that summary thread:

Only 4 platter versions of the disk affected:
It seems that only 4 platter versions of the WD10EACS seem to be affected. All newer 3 platter versions of the WD10EACS are not affected. It guess that all WD10EADS (larger disk cache) are 3 platter versions and should not suffer from high LCC.
3 platter: WD10EACS-65D6B0, WD10EACS-00D6B0, WD10EACS-00D6B1 (show no high LCC but LCC equals SSC)
4 platter: WD10EACS-00ZJB0, WD10EACS-32ZJB0, WD10EACS-00C7B0 (show high LCC)

Shapeshifter · 2009-06-07 12:38:34

Robertek wrote:

WD5000AACS
Still no problem here ( 2y old disk )
robert robert-server:~ [2013]% sudo smartctl --all /dev/sda | grep Load_Cycle
193 Load_Cycle_Count        0x0032   001   001   000    Old_age   Always       -       881422
I also increased the vm settings, some time ago when I found this info.
Im also looking for some samsung 1T

881'422 is very high. AFAIK the desktop market drives are rated for 300k and the server segment drives for 600k. Broken down on two years, this means your drive parked 36 times per hour. But I could just be wrong, I'm not very experienced with this kind of thing. I just thought I'd let you know that there is apparently some sort of problem with the Green Power drives from WD ^^

B: Your values look fine to me, It's like 1 cycle per day...

.:B:. · 2009-06-07 12:41:45

Thanks for confirming - the more I read up the more they seem to be okay indeed. I also added some more relevant info from the Synology forum to my previous post.

Also, it seems in its wisdom Western Digital stated that they only support their drives on Windows and Mac OS X. I think these are the last drives I purchase from their part, if that's true. Certainly perfect publicity towards a public that has a comparatively high share of people building their own systems - and knowing what's in it.

Also, for people wondering how their LLC/SSC relates to the age and use of their drives, it might be useful to check the total uptime of their drives:

[root@amalthea stijn]# smartctl -a /dev/sda|grep Power_On_Hours
  9 Power_On_Hours          0x0032   099   099   000    Old_age   Always       -       1293
[root@amalthea stijn]# smartctl -a /dev/sdb|grep Power_On_Hours
  9 Power_On_Hours          0x0032   099   099   000    Old_age   Always       -       1293
[root@amalthea stijn]# smartctl -a /dev/sdc|grep Power_On_Hours
  9 Power_On_Hours          0x0032   099   099   000    Old_age   Always       -       1295

Procyon wrote:

Does this apply to the WD MyBook 1TB too? They also park though I don't know how fast (smartctl isn't working on it)

I think those run Linux too (not sure) so Western Digital better make sure there's no problem with that... If you really want to know you'll have to open it up and connect the HD through (e)S-ATA to read out SMART values.

Ok - another edit. The same topic contains a post which defines the following models as affected:

WD RE2 GP WD1000FYPS-01ZKB0, WD7500AYPS-01ZKB0, WD7501AYPS-01ZKB0

Which of course contradicts with what someone else said earlier on the topic...

R00KIE · 2009-06-07 15:30:37

Mind you that the Power_On_Hours counter does not count the whole power on hours, it flips around after some time.
As for WD not supporting linux .... >_> thats a bad move. Also I guess they offer 3 years warranty if the disk starts acting strange just nuke it with zeros (or use the smart secure erase) and fill an rma and exchange it

Anyway if the hard disk does behave like that then I guess they have ensured it can do that for quite a long time, it's not the same thing as one setting the idle timeout too short on a disk that wasn't meant to work like that.

As for external hard disks, I have one WD Essentials 1TB and it sometimes stops when there is no activity but most of the times it just seems to be doing some kind of self test like a full smart test so I guess thats not a problem, I guess most if not all external hard disks will work just fine.

Bogart · 2009-06-08 01:53:24

B wrote:

Same topic also says:
# It seems that disks with version WD10EACS-00D6B0 (Nov 08) or later no longer show high LCC. Instead they show the START_STOP_COUNTER value as LCC. Question remains, if WD did really fix the problem or just mask it away by no longer showing the high LCCvalues to us.
* B knocks on wood.

The load/unload cycles can be easily heard, so it would be stupid for WD to hide it. If you don't hear your drive doing some click-click every minute then the displayed values should be true.

.:B:. · 2009-06-08 05:47:47

The whole box is silent as a whisper, I've never heard any clicks. I'd have to put my ear to it.

R00KIE · 2009-06-08 09:30:23

My notebook has a WD too and I think it does like to park its heads too often too, the clicking sound does happen when torrents are not in use, issuing "hdparm -B 255 /dev/sda" seems to keep that in check (still checking what happens to disk temperature, it might increase a little, I'll keep an eye on the logs), the disk in question is: WD3200BEVT-22ZCT0 fw:11.01A11

As for the external hd I have no way of knowing, the interface is usb and I don't feel like voiding the warranty to find out

I have an Hitachi disk that came with my previous notebook that also did that clicking sound and it worked for three years (and is still working) without any problems whatsoever (no way of reading the LCC now as the old notebook died and I don't have any adapter to connect the disk to a "normal" size ide connector).

Shapeshifter · 2009-06-08 10:23:02

Afaik, the only disks in questions are the Green Power ones, not WD disks in general.

R00KIE · 2009-06-08 13:42:36

Maybe not, maybe those are the ones people buy more and have noticed that take as an example the disk in my notebook (which is 7 months old).
I have never used the laptop scripts to save power so the issue with load cycle count due to some problem with those doesn't apply. This is normal disk usage with the defaults stored on the firmware.

193 Load_Cycle_Count        0x0032   133   133   000    Old_age   Always       -       201960

Edit:
Ooops seems like my external drive is from the green series WD10EACS but no clicks hope it holds itself together for a long time

Last edited by R00KIE (2009-06-08 13:49:45)

Shapeshifter · 2009-06-08 13:52:50

Uhm, 201960 is a lot! As said 300k is the life time spec.

brendanz · 2009-07-12 07:59:23

Hiyas.

I have picked up a few WD Caviar "Green" disks.
They are 1tb SATA with 16mb cache.
They show in the system as WD10EACS-22D with revision 01.0

Below sdb is the WD10EACS-22D, sda is an older seagate:

smartctl -a /dev/sdb|grep Power_On_Hours
9 Power_On_Hours 0x0032 100 100 000 Old_age Always - 34
smartctl -a /dev/sdb|grep Load_Cycle_Count
193 Load_Cycle_Count 0x0032 200 200 000 Old_age Always - 24
smartctl -a /dev/sdb|grep Start_Stop_Count
4 Start_Stop_Count 0x0032 100 100 000 Old_age Always - 24

smartctl -a /dev/sda|grep Power_On_Hours
9 Power_On_Hours 0x0032 091 091 000 Old_age Always - 8230
smartctl -a /dev/sda|grep Start_Stop_Count
4 Start_Stop_Count 0x0032 100 100 020 Old_age Always - 694
smartctl -a /dev/sda|grep Load_Cycle_Count does not return anything for the seagate

As i've just set up the box and had multiple powerup/downs, they seem pretty fine.
No change in running sound over several minutes, i have the components running on my desk without a case.
I'll keep an eye on them and post again if anything interesting happens.

Hope this helps someone.

Have fun all.

brendanz · 2009-07-12 08:36:07

Hi again.

Out of interest, i did the same on this laptop which is a Dell Inspiron 1520.
The disk is a Western Digital 2.5 inch 320gb SATA WD3200BEVT-7 revision 11.0.

During use, the disk sounds are similar to the ticking of a clock.
When idle the descending ping+clunk of the park can clearly be heard more than once per minute.

smartctl -a /dev/sda|grep Power_On_Hours
9 Power_On_Hours 0x0032 095 095 000 Old_age Always - 3899
smartctl -a /dev/sda|grep Load_Cycle
193 Load_Cycle_Count 0x0032 152 152 000 Old_age Always - 146929

Seems like the problem may not be confined to a particular series/type of disk?

I'm going to look at the suggested workarounds and consider paging the disk at regular intervals.
I'm also going to consider getting solid flash for my laptop's main hd.

Have fun.

Bogart · 2009-07-12 10:25:29

brendanz wrote:

Seems like the problem may not be confined to a particular series/type of disk?

Yes, the WD Green Power drives are an exception because they are _desktop_ hard drives. Any other normal desktop HD does not suffer from this problem.

However, it's a complete different story with laptop HDs: All of them have this "problem".

If you use laptop-mode-tools (and have the laptop-mode daemon started at boot), usually the best thing is to edit /etc/laptop-mode/laptop-mode.conf and change BATT_HD_POWERMGMT to 254 (assuming you have CONTROL_HD_POWERMGMT set to 1).

Dieter@be · 2009-07-12 15:27:34

I have one of these WD green power 1TB disks. but I do not do much writing/reading to it. I sometimes put some files on it and read some, but quite infrequent.
The OP makes it sound like you'll have problems "by definition" but I think in use cases like mine, there is no problem at all

[dieter@dieter-p4sci-arch ~]$ cat /proc/sys/vm/dirty_expire_centisecs /proc/sys/vm/dirty_writeback_centisecs 
3000
500
[dieter@dieter-p4sci-arch ~]$ sudo smartctl -a /dev/sdb |grep Load_Cycle_Count # GP 1 TB, few months old
193 Load_Cycle_Count        0x0032   200   200   000    Old_age   Always       -       32
[dieter@dieter-p4sci-arch ~]$ sudo smartctl -a /dev/sda |grep Load_Cycle_Count # system disk, a few years old
193 Load_Cycle_Count        0x0032   253   253   000    Old_age   Always       -       2229

Arch Linux

#1 2009-06-07 10:33:15

A word of advice on Western Digital "Green Power" Disks

#2 2009-06-07 11:39:48

Re: A word of advice on Western Digital "Green Power" Disks

#3 2009-06-07 11:43:54

Re: A word of advice on Western Digital "Green Power" Disks

#4 2009-06-07 12:09:13

Re: A word of advice on Western Digital "Green Power" Disks

#5 2009-06-07 12:20:57

Re: A word of advice on Western Digital "Green Power" Disks

#6 2009-06-07 12:23:22

Re: A word of advice on Western Digital "Green Power" Disks

#7 2009-06-07 12:26:32

Re: A word of advice on Western Digital "Green Power" Disks

#8 2009-06-07 12:38:34

Re: A word of advice on Western Digital "Green Power" Disks

#9 2009-06-07 12:41:45

Re: A word of advice on Western Digital "Green Power" Disks

#10 2009-06-07 15:30:37

Re: A word of advice on Western Digital "Green Power" Disks

#11 2009-06-08 01:53:24

Re: A word of advice on Western Digital "Green Power" Disks

#12 2009-06-08 05:47:47

Re: A word of advice on Western Digital "Green Power" Disks

#13 2009-06-08 09:30:23

Re: A word of advice on Western Digital "Green Power" Disks

#14 2009-06-08 10:23:02

Re: A word of advice on Western Digital "Green Power" Disks

#15 2009-06-08 13:42:36

Re: A word of advice on Western Digital "Green Power" Disks

#16 2009-06-08 13:52:50

Re: A word of advice on Western Digital "Green Power" Disks

#17 2009-07-12 07:59:23

Re: A word of advice on Western Digital "Green Power" Disks

#18 2009-07-12 08:36:07

Re: A word of advice on Western Digital "Green Power" Disks

#19 2009-07-12 10:25:29

Re: A word of advice on Western Digital "Green Power" Disks

#20 2009-07-12 15:27:34

Re: A word of advice on Western Digital "Green Power" Disks

Board footer