Summary: WD Green Power Disks & High Load Cycle Problem

[tw_switzerland]

Hello

Information about the problem of fast increasing Load_Cycle_Count with some disks is spread in the web. Here in this post I try to summarize the facts I have. I will update this post as new information becomes available.

If you think that anything should be mentioned here then please write a private message to me.

Summary of the facts:

• At least some versions of the Green Power disks from Western Digital seem to be switching between energy saving state (which starts after 8 seconds without access) and normal load operation when installed in a Unix/Linux based system.
  
• Affected disks show a fast growing load cycle count (LCC), which is a metric for how many times the drive unloaded its heads.
  
• When there is no access to the NAS (idle), the LCC increases between +50 and +120 per hour.
  
• As WD specified their disks for 300'000 LCC, the threshold will be reached within a few months. Maximum lifespan of affected disks might be shortened.
  
• If the NAS is under permanent load, LCC is not increased. So it is a problem in the idle state.
  
• It may be that disks mounted in an external case and attached via esata are not affected. But if these disks are used as internal disks with a logical volume on them, then they will show the high LCC as well.
  
• Not only Synology NAS are affected. Other Unix based NAS and Linux distributions seem to have the same problem.
  
• When the disks are installed in a windows system, there seems to be no problem.
  
• Not all combinations of disk version and NAS version seem to be affected by the problem. Up to now it is not known which combinations do have the problem and which do not.
  
• 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.
  
• Western Digital has the position, that they do not support operation of their disks with other OS than Windows and Mac OS. Maybe you could add some pressure or take back the recommendation of theirs disks.
  
• The problem is discussed in many threads throughout the web. Just google for "high load cycle count" or similar terms.
  
• According to WD, setting the timeout of IntelliPark to 300 seconds should reduce the problem (in a Linux box with sync all 30 seconds that results in disabling the main feature of GreenPower technology). This workaround seems not to work for everybody (for me it does not).

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)

Reproduction of the problem:

• 1. Install the WD disk with green power technology into the NAS as disk1 or disk2 (internal!)
  
• 2. Write down the following SMART values: Power On Hours (POHstart) and Load Cycle Count (LCCstart)
  
• 3. Stop all activity on the NAS so the disks reach idle states often
  
• 4. After several hours write down again the following SMART values: Power On Hours (POHstop) and Load Cycle Count (LCCstop)
  
• 5. Calculate the LCC increase: LCCdeltaPerHour = ( LCCstop - LCCstart ) / ( POHstop - POHstart )
  
• 6. If your disk is affected, you will see LCCdeltaPerHour between +50 and +120 (--> you should post your values in the Experiment Thread by Franklin)
  
• 7. If your disk is not affected, you will see LCCdeltaPerHour between +0 and +10

Resources:

• Good explanation of the head unloading/loading technique by Hitachi. I am not sure if WD uses exactly the same "ramp" technology.
  
• Potential explanation in another forum of what could be the reason for the high load cylce counts.
  
• Specification for WD Caviar GP disks
  
• Explanation of the WD Green Power Functions, the problem is probably due to "IntelliPark" feature

Many thanks for all the valuable help!

Thomas

[Pearl]

Western Digital's Technical Bulletin on the matter follows:

Symptom: Attribute 193
WD drives are designed to reduce power consumption, in part by positioning the heads into a park position (unloading the heads) and turning off unnecessary electronics, resulting in substantial power savings. WD defines this mode as Idle 3. Some utilities, OS’s, and applications, such as some implementations of Linux, for example, are not optimized for low power storage devices and can cause our drives to wake up at a higher rate than normal. This effectively negates the power-saving advantages of low-power drives such as WD’s RE2GP and artificially increases the number of load-unload cycles.

The number of systems using such applications and utilities is limited and customers can resolve this symptom by optimizing their systems to not wake up the drives unnecessarily every 10-30 seconds or so, thereby gaining substantial power savings and eliminating superfluous activity.

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

Solution
Most customers, when made aware of the unnecessary activity caused by their systems, have modified their utility, OS, or application to take advantage of WD’s advanced power-saving mode. Other customers have requested a utility to modify the behavior of the drive to wait longer before invoking Idle 3 mode. Although such a change eliminates significant power savings during periods of inactivity, WD has created an update for those customers wishing to maintain their existing utilities, OS’s, and applications. This update is described in WD's Process Change Notice PCN 2579-701324-A02 and is available through WD's technical support organization, at http://support.wdc.com/re2gp.

The update also supports a low-power spin-up feature optimized for highly energy efficient, large scale storage applications, which is enabled using a utility also available at the support link above.

[DS508_customer]

Thank you for sharing all this information! I have added this paragraph on this subject to Synology's wiki.

... This workaround seems not to work for everybody (for me it does not) ...

Does this mean that the workaround does not solve the problem or that it may be an unacceptable solution because of certain (which?) drawbacks/ negative side effects of the workaround?

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)

According to the topic title, this topic deals with any WD Green Power model. Is what was mentioned in the start post regarding only the WD10EACS model (and not the WD1000FYPS for example)?

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

To which extend do above posts contradict each other, regarding which models are affected?

[tw_switzerland]

Hi DS508_customer

I have set all my affected WD10EACS drives (old 4 platter version) to an idle timer of 300 seconds as described here. It seems that the disks have an internal maximum timeout value of 25 seconds, even if something different (here 300 seconds) is set using wdidle3 tool. As the strange accesses of the Syno NAS seem to happen every 30 seconds, this results in the about 2 load cycles per minute we see. So setting the head park timeout on affected desktop class disks seems to be no solution. Maybe disabling the GP feature using wdidle3 tool could help. I will give that a try.

WD seems to have released a new firmware for the affected enterprise level disks that allows the timeout to be set to larger values. Unfortunately I do not know of such an update for the desktop class drives.

Newer 3 platter versions of the 1TB disks work fine. Older 4 platter 1TB disks are affected. Enterprise level disks can be updated to a firmware where you can set the wdidle timeout to 300 seconds. There is no such firmware update for desktop level disks.

Thomas

[AndyF]

I recently applied the WD firmware patch to my WD1000FYPS drives and the result has been very worthwhile. Previously, they racked up around 1500 LCC per day. after the patch this has dropped to around 10 per day.

Andy

[tw_switzerland]

Unfortunately there is no such patch for the desktop drives (EACS).

[HaDel]

In my opinion WD is fully responsible and must solve the issue.

In their product description they have absolutely no disclaimer that their green drives do not work well with *nix.
The argument they gave that "Linux is not using well our power saving design" is crap. Moreover they gave it in a obscure small-printed footnote and NOT in their product spec.

The argument that GP is not for intensive applications is also crap. HDD sitting in any device doing nothing is not intensive operation. This device could be a NAS, but also could be a simple media player. Or a PC with Linux installed.

All in all the IntelliPark is quite dumb. It is so simple to write and adaptive algorithm instead of the brute force they have now.
Parking after 8 sec idle time is lame. They could easily check if the parking was negated by the OS 3 times in a row, to set the timeout to 20 min, 1h or whatever.

So basically my take is that WD failed to test properly and in order to minimize immediate costs they hide the problem and when confronted put the blame elsewhere. In short term this might help but I certainly hope that in long term the damage over the brand name will fire back.
My first step - I will never ever buy any product from WD.
Second - I will spread the word on their approach as far as I can.

[vkanak]

Hi, i recently bought 2TB HDD WD20EADS-32S2B0, FW 01.00A01 and i passed this test "LCCdeltaPerHour = ( LCCstop - LCCstart ) / ( POHstop - POHstart )" with cca 2 LCCeltaPerHour result, i mean that is ok, I have one disk in synology DS210j (no raid yet). My NAS runs approximately 90 hours and perform cca 500 LCC.

[jehellevang]

Does this mean that WD have fixed the problem in the latest versions of their WD Green drives?

/Jan

[achilles.a]

Declaration : I'm not a Synology user and not remotely connected with NAS

I've been following the WD Green park/unpark/high LCC issue very keenly around the Web because I'm interested in buying the WD10EARS drive.

I had emailed WD support a few days back and today I got their reply which reads as :

Subject
Head parking/unparking + Load Cycle Count Issues

Discussion Thread
Response (Paul B.)02/16/2010 07:11 AM
Dear xxxxxx,

Thank you for contacting Western Digital Customer Service and Support. My name is Paul B.

I apologize for the inconvenience this has caused you; because this is a Green Power Drive it will do more load and unloads than a normal drive for power savings. Although the counter of the drive will increase incrementally with each load and unload, this is only a counter and does not affect the drive. To date, we have had no reported failures due to high cycle and load times with any of our drives.

[anprieto]

Hi all,

I just installed my new DS110j + Caviar green WD8000AARS

During the first night without activity the average LLC was 9, so I assume that this combination is ok.

Thank you
anprieto

[maxxfi]

Hello,

also my DS-210j with a harddisk WD10EARS-00Y5B1 (Firmware 80.00A80) seemed to suffer
from the LCC problem. Already shortly after initial installation, e.g. as soon as the drive
was idle, I started to notice that LCC counter was going up, and I could even hear that
'tock'-sound of the HDD heads every 8 seconds, after which the SMART tool was reporting
an increase in the parameter 193.

So I ran the procedure described with the WDIDLE3 in order to change idle time to 300s.
It definitely seems to work in my case as the LCC counter is not increasing any further.

Thanks to tw_switzerland for the clear summary of the problem.

[shadow2000]

I have my DS-201J with a WD10EADS-00M2B0.

My request to WD support was smashed down with "We do not support linux". Also my hint that this fact is not mentioned anywhere was ignored.
I'm so what disappointed from WD. Is this customer orientated?

My LCC is now at 340.000 so I will order new Seagate disks.

[congo]

"We do not support linux"..

What an arrogance - not supporting linux from harddrive manufacturer, is like if a motherboard vendor wouldnt support running an OS on their board. Its rubbish - i never heard anything like it. I have been truly WD customer for decades - if I hear similar stories again from WD consumers i will certainly go to another vendor of my future drives <-- as a kind of small crusade against this nonsense.

[OAG]

WD brings more light to LCC issue for Linux environment: see their KB article "The S.M.A.R.T Attribute 193 Load/Unload counter keeps increasing on a SATA 2 hard drive" (http://wdc.custhelp.com/cgi-bin/wdc.cfg/php/enduser/std_adp.php?p_faqid=5357&p_created=1266947046&p_sid=BXxBGP5k&p_accessibility=0&p_redirect=&p_srch=1&p_lva=&p_sp=cF9zcmNoPTEmcF9zb3J0X2J5PSZwX2dyaWRzb3J0PSZwX3Jvd19jbnQ9MTQsMTQmcF9wcm9kcz0yMjcsMjk0JnBfY2F0cz0mcF9wdj0yLjI5NCZwX2N2PSZwX3BhZ2U9MSZwX3NlYXJjaF90ZXh0PWxvYWQgY3ljbGUgY291bnQ!&p_li=&p_topview=1).

Affected Models are WD20EADS, WD20EARS, WD15EADS, WD15EARS, WD10EADS, WD10EARS, WD8000AARS, WD7500AADS, WD7500AARS, WD6400AADS, WD6400AARS, WD5000AADS, WD5000AARS.

Along with recommendation to use "RE2GP Idle Mode Update Utility", they suggest more options (see quote below).
Can any of you, experienced Linux users, do the test and post your comments?

Linux users: Decrease the logging message

i. Examine your /etc/syslog.conf file for unnecessary logging activity and to optimize its performance. If you don't want to log any system activity, consider disabling syslogd and klogd entirely; or, at the very least, minimize the amount of logging your system performs. You can also prefix each entry with the minus sign (-) to omit syncing the file after each log entry. This will log anything with a priority of info or higher, but lower than warning, to /var/log/messages or /var/log/mail without needing to sync to disk after each write. Since we want to keep all messages with a priority of warning, this will be logged to a different file without disabling disk syncing (to prevent data loss in the event of a system crash).
*.warning /var/log/syslog
*.info;*.!warning;mail.none -/var/log/messages
mail.info;mail.!warning -/var/log/mail

ii. Another item to be aware of is the -- MARK -- messages that syslogd(8) writes. This will affect your hard drive inactivity settings. You can simply disable this by running syslogd(8) with:

if [ -x /usr/sbin/syslogd -a -x /usr/sbin/klogd ]; then
# '-m 0' disabled 'MARK' messages
/usr/sbin/syslogd -m 0
sleep 1
# '-c 3' displays errors on console
# '-x' turns off broken EIP translation
/usr/sbin/klogd -c 3 -x
fi