Is there a problem with head parks on WD Green HDDs?

zds · Post by **zds** » Sat Dec 13, 2008 8:59 am

As per SPCR recommendation and nice-sounding specs I and multiple of my friends have bought Western Digital Green Power drives in past half a year. Because the drive is particularly suited for media server applications, this is where the drives have seen their use.

But, alas, within past month news began to hit in that the feature in WD GP drives that helps to save energy, Intelli-Park, is not that intelligent. In fact, with default settings it parks the heads after 8 seconds of inactivity. What this means depends of your OS and usage pattern, but in these low-usage media server machines running Linux it means the drive will reach it's designed lifetime total or parkings in 200-300 days.

What makes things worse is that WD has been completely silent about the issue, and the tool that can disable the Intelli-Park feature is not publicly available. Ie. WD is doing their best to just let the drives die instead of warning the customers about the issue.

What to do? If you are using Green Power drive, check the SMART data to see if your drive has alarmingly high Load_Cycle_Count value. For my WD10EACS drive the count is 87k after two months of usage - almost one third of the designed lifespan spent. If you drive has a high value, contact WD and ask for the wdidle3.exe. It's a DOS program that can disable the Intelli-Park or configure it.

It might also be worth putting a mention about this to the SPCR review recommending GP drives so that people can prepare from the very beginning and thus prevent untimely death of their drive.

Title of this thread changed by mods on 4-30-09. Existing title was basically FUD

xan_user · Post by **xan_user** » Sat Dec 13, 2008 9:13 am

198 on 5000aacs in 3 months of ~12hrs/ a day. xp pro

So linux is harder on the GP than windows? Interesting.

zds · Post by **zds** » Sat Dec 13, 2008 9:20 am

xan_user wrote:So linux is harder on the GP than windows?

So it seems. Apparently Linux tries to save the discs by buffering data and syncing it to discs every 10-15 seconds. Unfortunately this is enough time for the WD to park it's heads..

Here is pointer to discussion about the issue on Linux Kernel Mailing List: http://kerneltrap.org/mailarchive/linux ... /9/1386304

Disclaimer: I am not expert in Linux kernel, so my understanding about how the Linux kernel works with discs is pretty shallow. The only thing I know for sure is all the 4-6 GPs in my or my friends possession have alarmingly high head load/unload counts.

xan_user · Post by **xan_user** » Sat Dec 13, 2008 9:27 am

I found this ftp server with the file

...

ftp://76.17.197.241/public/software/dos/ (warning unkown ftp server)

I guess with 198 parks value i should think about using it?

zds · Post by **zds** » Sat Dec 13, 2008 9:43 am

xan_user wrote:I guess with 198 parks value i should think about using it?

I think with 198 in 3 months you should be pretty safe, so I'd just monitor the situation. My current count is 88090, that's why I'm a bit worried:

Situation November 8th:

Code: Select all

193 Load_Cycle_Count        0x0032   189   189   000    Old_age   Always       -       35865

Situation December 12th:

Code: Select all

193 Load_Cycle_Count        0x0032   171   171   000    Old_age   Always       -       88090

Comparing those and doing some math I figured that per this information my drive is designed for 600k load/unload cycles (from 200 to 171 in 88k occurrences) so after 2 months of usage I still have some 520k to spare.

Naturally I am only speculating when it comes to the life expectancy - without WD saying anything it's hard to be 100% sure.

tehcrazybob · Post by **tehcrazybob** » Sat Dec 13, 2008 1:08 pm

Can anyone confirm that this is only a problem under Linux? I've got a friend with four or five GP drives on Vista, and want to know whether I need to warn him about the problem.

porkchop · Post by **porkchop** » Sat Dec 13, 2008 3:30 pm

this is quite interesting. i remember reading some thread a while back, someone complaining about a constant clacking- i can only assume it was intellipark.

my 6400aacs has been running for about 2 months now(xp), i'm always running something that uses the hdd- but i'll definitely have a look at it with speedfan tonight.

edit:
mine- WDC WD6400AACS-00G8B0 with firmware 05.04C05.
power on hours count 1140
load cycle count 130

NyteOwl · Post by **NyteOwl** » Sat Dec 13, 2008 9:45 pm

Apparently it is a problem independant of OS. A thread at the forums over at Ars Technica indicated the wdidle.exe wasn't that effective.

http://episteme.arstechnica.com/eve/for ... 1009715931

bgiddins · Post by **bgiddins** » Sun Dec 14, 2008 1:05 am

You can check this with

Code: Select all

smartctl -A /dev/sda

I've just checked mine (2 x WD10EADS in mdadm RAID 1) - drives have 410 hours of use, but only a load cycle count of 211. That's under Ubuntu Server 8.10.

dukla2000 · Post by **dukla2000** » Sun Dec 14, 2008 3:11 am

My GP has been running about 10 months 24/7. It is not root or swap so should have immune from this issue but

Code: Select all

smartctl -a /dev/sdb
...
Device Model:     WDC WD5000AAVS-00ZTB0
Serial Number:    WD-WCASU0621311
Firmware Version: 01.01B01
...
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  9 Power_On_Hours          0x0032   091   091   000    Old_age   Always       -       7053
...
193 Load_Cycle_Count        0x0032   007   007   000    Old_age   Always       -       581689
...

What really cheeses me off with this drive is

Code: Select all

197 Current_Pending_Sector  0x0012   200   200   000    Old_age   Always       -       1
...
200 Multi_Zone_Error_Rate   0x0008   200   198   051    Old_age   Offline      -       1

which I cant seem to clear and results in "Device: /dev/sdb, 1 Currently unreadable (pending) sectors" every 20 minutes in /var/log/warn

Sounds like I really must exchange the thing!

ps SuSE 10.3 initially, 11.0 more recently.

wojtek · Post by **wojtek** » Sun Dec 14, 2008 7:11 am

Windows XP user here. I just checked one of my GP's - I'm using this drive about 2-3h per day (external) since 3 months to play music.

WDC WD10EACS-00ZJB0 (WD-WCASJ1834260)
Load/Unload Cycle Count 5176

During playing just 2 songs it's increased by 5 - from 5171
Another song - +3

Strid · Post by **Strid** » Sun Dec 14, 2008 8:17 am

My GP doesn't seem to suffer from said problem. It's been running for about three months now. I've been running Ubuntu 8.04 until I updated to Ubuntu 8.10 64-Bit this week. I wonder why spin-up times are roughly 7K after three months. Seems like 60 times a day is a bit more than would expect, although not extremely high. I don't think it goes into idle mode ever, because it seems to keep writing every 5 seconds. So it never reaches a long enough idle time to configure with hdparm. I can't really find the cause of this, though.

Code: Select all

 sudo smartctl -a /dev/sdb
...
=== START OF INFORMATION SECTION ===
Device Model:     WDC WD10EACS-00D6B0
Serial Number:    WD-WCAU41667190
Firmware Version: 01.01A01
User Capacity:    1.000.204.886.016 bytes
...
=== START OF READ SMART DATA SECTION ===
...
SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  1 Raw_Read_Error_Rate     0x002f   200   200   051    Pre-fail  Always       -       0
  3 Spin_Up_Time            0x0027   163   162   021    Pre-fail  Always       -       6808
  4 Start_Stop_Count        0x0032   100   100   000    Old_age   Always       -       274
  5 Reallocated_Sector_Ct   0x0033   200   200   140    Pre-fail  Always       -       0
  7 Seek_Error_Rate         0x002e   100   253   051    Old_age   Always       -       0
  9 Power_On_Hours          0x0032   099   099   000    Old_age   Always       -       967
 10 Spin_Retry_Count        0x0032   100   100   051    Old_age   Always       -       0
 11 Calibration_Retry_Count 0x0032   100   100   051    Old_age   Always       -       0
 12 Power_Cycle_Count       0x0032   100   100   000    Old_age   Always       -       230
192 Power-Off_Retract_Count 0x0032   200   200   000    Old_age   Always       -       18
193 Load_Cycle_Count        0x0032   200   200   000    Old_age   Always       -       267
194 Temperature_Celsius     0x0022   115   105   000    Old_age   Always       -       35
196 Reallocated_Event_Count 0x0032   200   200   000    Old_age   Always       -       0
197 Current_Pending_Sector  0x0032   200   200   000    Old_age   Always       -       0
198 Offline_Uncorrectable   0x0030   200   200   000    Old_age   Offline      -       0
199 UDMA_CRC_Error_Count    0x0032   200   200   000    Old_age   Always       -       0
200 Multi_Zone_Error_Rate   0x0008   200   200   051    Old_age   Offline      -       0

I don't know if this helps anyone, but if you want to check your Green Power drive under Ubuntu/Debian do this:

Strid@UbersilentPC:~$ sudo apt-get install smartmontools

This will install what you need to check the SMART data off of your drive by doing this:

Strid@UbersilentPC:~$ sudo smartctl -a /dev/sda

You might want to replace "sda" with the correct handle for your drives.

ronnylov · Post by **ronnylov** » Sun Dec 14, 2008 11:13 pm

If you are running ubuntu, then this may help:
http://ubuntuforums.org/showpost.php?p= ... stcount=26

AuraAllan · Post by **AuraAllan** » Mon Dec 15, 2008 8:31 am

Here's SMART info for the WD10EADS in my Synology DS107+ NAS.

Strid · Post by **Strid** » Mon Dec 15, 2008 9:53 am

Does anyone know what Spin_Up_Time stands for? Seems like most if not all of us have a high number in this parameter.

xan_user · Post by **xan_user** » Mon Dec 15, 2008 12:50 pm

update of email to WD asking for the wdidle3.exe file.

There are lots of GP drives dying prematurely, apparently the file "wdidle3.exe" can prolong the life span, but it s not on your site.

Why?

Where can I get the "wdidle3.exe" file needed to fix my drive(s)?

Auto-Response 12/13/2008 09:23 AM
Question: How can I get support for my WD product in LINUX or UNIX?
Answer: Western Digital technical support only provides jumper configuration (for EIDE hard drive) and physical installation support for hard drives used in systems running the Linux/Unix operating systems. For setup questions beyond physical installation of your Western Digital hard drive, you would have to contact the vendor of your Linux/Unix operating system.

Not a good answer for a company looking to expand its customer base.

m0002a · Post by **m0002a** » Mon Dec 15, 2008 2:37 pm

xan_user wrote:Not a good answer for a company looking to expand its customer base.

I got the same "auto-response" because I selected UNIX/Linux as the OS in the questionaire. They give you a chance to reply by email if the auto-response email is not satisfactory.

CoccoBill · Post by **CoccoBill** » Tue Dec 16, 2008 7:02 am

Hm, running 4 WD10EACS drives on an Adaptec 5805 under Ubuntu 8.10 server for a couple months 24/7, no sign of problems here. Power_On_hours around 1750 and Load_Cycle_Counts between 8 and 12.

Nick Geraedts · Post by **Nick Geraedts** » Tue Dec 16, 2008 12:49 pm

I've got 8 WD drives connected to a 3ware controller. After a few months of running, I've got the following data from smartmontools:

Code: Select all

Drive 0
  9 Power_On_Hours          0x0032   091   091   000    Old_age   Always       -       6759
193 Load_Cycle_Count        0x0032   093   093   000    Old_age   Always       -       321155
Drive 1
  9 Power_On_Hours          0x0032   091   091   000    Old_age   Always       -       6756
193 Load_Cycle_Count        0x0032   100   100   000    Old_age   Always       -       301268
Drive 2
  9 Power_On_Hours          0x0032   094   094   000    Old_age   Always       -       4436
193 Load_Cycle_Count        0x0032   078   078   000    Old_age   Always       -       367302
Drive 3
  9 Power_On_Hours          0x0032   094   094   000    Old_age   Always       -       4437
193 Load_Cycle_Count        0x0032   078   078   000    Old_age   Always       -       368137
Drive 4
  9 Power_On_Hours          0x0032   094   094   000    Old_age   Always       -       4437
193 Load_Cycle_Count        0x0032   078   078   000    Old_age   Always       -       367378
Drive 5
  9 Power_On_Hours          0x0032   094   094   000    Old_age   Always       -       4437
193 Load_Cycle_Count        0x0032   075   075   000    Old_age   Always       -       376041
Drive 6
  9 Power_On_Hours          0x0032   094   094   000    Old_age   Always       -       4437
193 Load_Cycle_Count        0x0032   023   023   000    Old_age   Always       -       531946
Drive 7
  9 Power_On_Hours          0x0032   094   094   000    Old_age   Always       -       4437
193 Load_Cycle_Count        0x0032   076   076   000    Old_age   Always       -       372869
Press any key to continue . . .

However, there's no odd clicking or signs of imminent death from the drives. SMART data is actually quite DUMB in some ways, and it definitely shouldn't be taken as an absolute measure of life expectancy of the drives. Google did a lengthy analysis of drive life expectancy, and in a large number of the failures, SMART gave no warning of failure. Furthermore, there were several drives that had worked far beyond their "rated specs" according to the SMART data.

EDIT - There's also the fact that head "parking" has several stages involved. There are soft parks, hard parks, and stages inbetween. My guess is that the rated values on WD's website are for hard parks (the kind that happens when the drive is turned off), while Intellipark does a soft park - just moving it away from the disk.

protellect · Post by **protellect** » Tue Dec 16, 2008 1:43 pm

Mine seems ok. XP64 bit.

bgiddins · Post by **bgiddins** » Tue Dec 16, 2008 3:32 pm

There's some extremes here - either counts in the low hundreds, or counts in the tens or hundreds of thousands.

Is it a disk firmware issue, or a controller issue?

Derek Baker · Post by **Derek Baker** » Wed Dec 17, 2008 3:22 am

zds wrote:I think with 198 in 3 months you should be pretty safe, so I'd just monitor the situation.

My new WD10EACS-00ZJB0 has reached 210 in 17 hours - Vista 64 Home Premium SP1.

_MarcoM_ · Post by **_MarcoM_** » Wed Dec 17, 2008 6:52 am

Ok, now i have my new 640GB GP, date 29 oct 2008, firmware 05.04C05 (same as porkchop).

I will use it as a system boot disk under Windows Xp Pro SP3, what should I check? The load/unload cycle count?

KlaymenDK · Post by **KlaymenDK** » Wed Dec 17, 2008 8:21 am

Strid wrote:Does anyone know what Spin_Up_Time stands for? Seems like most if not all of us have a high number in this parameter.

FYI: "Spin-Up Time":
Average time of spindle spin up (from zero RPM to fully operational [millisecs]). (source)

In other words, it's perfectly normal to have a value of a couple of thousand, as it represents the drive's boot-up time -- only, it should not fluctuate too much over time.

Derek Baker · Post by **Derek Baker** » Wed Dec 17, 2008 8:51 am

The Data for my GP's Spin Up Time is 8116, though the Current and Worst are both 177 and the Status is OK (all from HDTune).

zds · Post by **zds** » Wed Dec 17, 2008 9:16 am

xan_user wrote:
Auto-Response 12/13/2008 09:23 AM
Question: How can I get support for my WD product in LINUX or UNIX?
Answer: Western Digital technical support only provides jumper configuration (for EIDE hard drive) and physical installation support for hard drives used in systems running the Linux/Unix operating systems. For setup questions beyond physical installation of your Western Digital hard drive, you would have to contact the vendor of your Linux/Unix operating system.
Not a good answer for a company looking to expand its customer base.

Indeed. It's one thing to make mistakes, and another thing to make mistakes and then refuse to help your customers to fix it.. If that's the occifial stance of WD, I will not be buying their hardware for machines running any OS for next couple of years..

BTW, I have two WD drives in the same server and I used wdidle3 to set head parking time to maximum allowed amount (not disable, however) for both of them. In 5.5 days Green Power one has made 2600 more head unloads/loads, the older WD drive - tadaa - one (1).

Naturally usage pattern is not 1:1 for those drives, but this starts to smell like a real firmware bug, not just a feature. Ie. it looks to me the GP just does not have too short time, the timer itself is not working properly. Bug, not a feature.

EDIT:

Firmware version of my drive is 01.01B01.

zds · Post by **zds** » Wed Dec 17, 2008 9:25 am

Nick Geraedts wrote:Furthermore, there were several drives that had worked far beyond their "rated specs" according to the SMART data.

There's also the fact that head "parking" has several stages involved. There are soft parks, hard parks, and stages inbetween. My guess is that the rated values on WD's website are for hard parks (the kind that happens when the drive is turned off), while Intellipark does a soft park - just moving it away from the disk.

If this is the case, then WD should come up, say "your drives are fine, we just goofed up and track soft parks in hard park category in SMART data". They are actively not doing that, instead trying to hide the problem.

BTW, another minus to WD: GP drives do not follow the standard commands to disable head parking (via hdparm in Linux). Ie. you indeed need the special program from WD to do it.

I understand every company makes mistakes, bug what bugs me here is that WD is not confirming anything, instead trying to cover the tracks and behave like there was no issue.

Nick Geraedts · Post by **Nick Geraedts** » Wed Dec 17, 2008 4:04 pm

zds wrote:If this is the case, then WD should come up, say "your drives are fine, we just goofed up and track soft parks in hard park category in SMART data". They are actively not doing that, instead trying to hide the problem.

BTW, another minus to WD: GP drives do not follow the standard commands to disable head parking (via hdparm in Linux). Ie. you indeed need the special program from WD to do it.

I understand every company makes mistakes, bug what bugs me here is that WD is not confirming anything, instead trying to cover the tracks and behave like there was no issue.

So... because a company doesn't allow Linux software to park the heads, that's a bad thing? Perhaps the Linux command might interfere with the normal operation of the drive - something that in turn could cause more headaches for WD.It sounds to me like they're preventing people from meddling with a custom piece of firmware. It's not like we have direct access to NCQ ordering operations in modern drives either, but nobody complains about that since it works and doesn't throw any erroneous errors in SMART data.

I haven't seen any evidence of WD trying to pull the wool over anyone's eyes here. I think they're just not letting the issue grow out of proportion. I aside from the horrible JB series, I've seen one WD drive die within it's warranty period. This includes the 16 drives in the computers I own, along with those that I've used for client builds in the past two years (approximately 30).

zds · Post by **zds** » Wed Dec 17, 2008 4:50 pm

Nick Geraedts wrote:So... because a company doesn't allow Linux software to park the heads, that's a bad thing? Perhaps the Linux command might interfere with the normal operation of the drive - something that in turn could cause more headaches for WD.

Negative. The command in question was instruction for the drive to turn off head parking as an energy saving measure, ie. sets a new value to configuration option (that we know to exist in WD GP drives as well). It does not command drive to do anything at that very point. Also, many other manufacturers have been able to implement it just fine.

cmthomson · Post by **cmthomson** » Thu Dec 18, 2008 5:59 pm

There was a similar problem with IBM/Hitachi 2.5" drives about 6 years ago. The default firmware settings parked the heads after 20 seconds, and the OS of the particular appliance that my company built touched the disks every 30 seconds. After about a year of this, the soft-landing area started to shred and spew bits of gunk all over the platter, resulting in data errors.

All predictable after the fact, and really no one was to blame (right, try explaining that to our customers). The disks started to fail in alarming numbers, with park counts over a million.

The lesson here that engineers will appreciate (but not consumers or executives) is that full and open communication about a product's attributes and limitations benefits everyone.