Jump to content


Photo

WD GreenPower drives head parking issue


  • You cannot start a new topic
  • Please log in to reply
10 replies to this topic

#1 matt_garman

matt_garman

    Member

  • Member
  • 4 posts

Posted 22 October 2008 - 07:55 AM

I have four of the "RE2" version of the 1 TB WD GreenPower drives (model WD1000FYPS). I'm using them in a software RAID5 NAS.

Until yesterday, I was quite happy with their cool and quiet operation. And then I found this thread. It's from the Linux Kernel Mailing List; you can find many archived versions by searching for "Western Digital GreenPower drives and Linux".

The gist is that the read/write head on these drives parks itself too often. While this does result in power savings while parked, the drives are only designed to support 300k parks (consumer drives) or 600k parks (enterprise/RE2 edition).

I don't know if this problem exists on Windows or not. I would assume it does, at least under some scenarios.

On Linux, I use a program called smartctl to read SMART data from the drive. This program is also available on Windows (but I'd wager there are also GUI-based tools that do the same).

Either way, the smart data will show how many times the head has been parked via the "Load_Cycle_Count" parameter.

An (edited) example from one of my drives:
smartctl -a /dev/sda
Power_On_Hours   3377
Load_Cycle_Count 97972

That averages out to 29 loads/hour... every two minutes!

The "solution" so far has been to use the "wdidle3.exe" DOS utility provided by Western Digital to change the head-park idle timer to a different value (or disable it all together). I just found out about this issue yesterday, so I haven't had a chance to actually do try this yet. I found this FTP site that has a copy of the wdidle3.exe program.

Anyway, if anyone else has any info on this issue, I'm eager to hear it.

#2 Phoenix1997

Phoenix1997

    Member

  • Member
  • 58 posts

Posted 24 October 2008 - 02:52 AM

I'm running Windows Server 2003, and I have the results for 2 10's and a 5000:
http://bravo.made2ow...isol/drives.txt
It looks like the two WD10EACS are not parking their heads, while the 5000 is. However, at the rate the 5000 is, the drive should last about 7 years.

#3 gopack

gopack

    Member

  • Member
  • 1 posts

Posted 30 October 2008 - 09:47 PM

Some references:

https://wiki.ubuntu....PowerManagement

https://wiki.ubuntu....ePowerManagment

https://bugs.launchp...port/ bug/59695

#4 brickman

brickman

    Member

  • Member
  • 5 posts

Posted 22 December 2008 - 08:27 PM

I came here specifically to find information about this "load cycle count" issue, and am surprised with how little discussion there is. I came across this problem tonight on silentpcreview.com. I have a 4 month old WD10EACS 1TB drive that has a load cycle count of over 120,000.

The WD drive specs suggests a maximum load cycle count of 300,000 before it dies. I've had the drive for 3 months and at the current pace I'm gonna have a dead drive in 5 months.

I've spent today doing searches across the net for anything related to Western Digital and "load cycle count" and only came up with other sites discussing the problem without any known acknowledged fix by WD. NewEgg has over 500 review for this drive and not a single person mentions this problem.

#5 TRACKER_MAN

TRACKER_MAN

    Member

  • Member
  • 30 posts

Posted 23 December 2008 - 05:55 AM

I came here specifically to find information about this "load cycle count" issue, and am surprised with how little discussion there is. I came across this problem tonight on silentpcreview.com. I have a 4 month old WD10EACS 1TB drive that has a load cycle count of over 120,000.

The WD drive specs suggests a maximum load cycle count of 300,000 before it dies. I've had the drive for 3 months and at the current pace I'm gonna have a dead drive in 5 months.

I've spent today doing searches across the net for anything related to Western Digital and "load cycle count" and only came up with other sites discussing the problem without any known acknowledged fix by WD. NewEgg has over 500 review for this drive and not a single person mentions this problem.

Hello Brickman,
you can use tool that our friend matt_garman gave, i have used it and disabled ramp load/unload mechanism and i don't have this problem anymore!
wd tools
good luck!
Main PC: CPU E8400@3Ghz, 2x2GB DDR2 ADATA 1066+, 2xWD5000AAKS@RAID1 on Jmicron 363,
MSI P45 Platinum, VGA ASUS EAH4670 DVDRW NEC ND-3540A
NAS PC: CPU Intel E5200@2.5GHz, 2x2GB DDR2, 4.5TB with Opensolaris 06.2009 ZFS, Chieftec 550W, Thermaltake Matrix.

#6 brickman

brickman

    Member

  • Member
  • 5 posts

Posted 23 December 2008 - 05:50 PM

I came here specifically to find information about this "load cycle count" issue, and am surprised with how little discussion there is. I came across this problem tonight on silentpcreview.com. I have a 4 month old WD10EACS 1TB drive that has a load cycle count of over 120,000.

The WD drive specs suggests a maximum load cycle count of 300,000 before it dies. I've had the drive for 3 months and at the current pace I'm gonna have a dead drive in 5 months.

I've spent today doing searches across the net for anything related to Western Digital and "load cycle count" and only came up with other sites discussing the problem without any known acknowledged fix by WD. NewEgg has over 500 review for this drive and not a single person mentions this problem.

Hello Brickman,
you can use tool that our friend matt_garman gave, i have used it and disabled ramp load/unload mechanism and i don't have this problem anymore!
wd tools
good luck!


tracker man, I did try wdidle3. Setting it to 25 seconds did nothing. Setting it to the max of 25.5 seconds did nothing. Trying the disable flag "WDIDLE3 /D" caused it to revert to 6300 milliseconds. There's no way to disable it. This drive is incrementing 1 per minute. At this rate, I'll reach the 300,000 threshold in about 3 months time.

#7 K15

K15

    Member

  • Member
  • 389 posts

Posted 25 December 2008 - 01:09 AM

I came here specifically to find information about this "load cycle count" issue, and am surprised with how little discussion there is. I came across this problem tonight on silentpcreview.com. I have a 4 month old WD10EACS 1TB drive that has a load cycle count of over 120,000.

The WD drive specs suggests a maximum load cycle count of 300,000 before it dies. I've had the drive for 3 months and at the current pace I'm gonna have a dead drive in 5 months.

I've spent today doing searches across the net for anything related to Western Digital and "load cycle count" and only came up with other sites discussing the problem without any known acknowledged fix by WD. NewEgg has over 500 review for this drive and not a single person mentions this problem.



It's worth pointing out that, however WD words it, 300k cycles is the minimum amount before the chance of problems increases. Now I understand your concern, I wouldn't like my drive to have 120,000 load/unloads after only 4 months either. But it's not going to die at the 300,001th head unload. In all likelihood, the drive would do well over 1 million cycles without a problem. Even so, it would be a good idea to stop such frequent load/unloading, but I just wanted to ease your concern at least a bit.

Just an interesting example; I have a Bell expressvu 5100 satellite receiver. Its about 5-6 years old and has a Seagate U6, 40GB drive in it. It averages between 2-4 starts/stops each hour (to download guide info, or whatever). This works out to between 17k and 35k start/stops each year. Given that the drive specifies 50k start/stops, it should have died in its second year.
Of course it does fewer cycles, since the drive constantly runs whenever I have the receiver on, but I just figure it's an interesting example. One has to wonder why the receiver's firmware would cycle the drive so often, rather than just leaving it run. Bell's newer receiver (the 9200) does just that, the drive runs 24/7. Even more interesting with that U6 is how incredibly hot it gets during operation. Honestly it's rather impressive that the drive still works, albeit with the odd seek error every now and again. ;)
Posted Image

#8 whiic

whiic

    hauptsturmführer

  • Member
  • 921 posts

Posted 25 December 2008 - 07:50 AM

"It's worth pointing out that, however WD words it, 300k cycles is the minimum amount before the chance of problems increases. Now I understand your concern, I wouldn't like my drive to have 120,000 load/unloads after only 4 months either. But it's not going to die at the 300,001th head unload. In all likelihood, the drive would do well over 1 million cycles without a problem."

But at this rate, it will reach 2.6 million cycles within 5 year service life. However, it may survive over 2 million cycles... and 5 years is a long time anyway even without unload cycles. I have the same problem with my GreenPowers and would rather not have 1440 unloads a day. The cause is SpeedFan utility that monitors temperatures of mobo, CPU and HDDs (via SMART). Other temperatures are polled every 10 seconds or more frequently but even SMART polling is done once a minute. WD unloads faster than that, even at most conservative setting in wdidle tool and result is always 1440 cycles.

I have "solved" the problem by stopping using SpeedFan, but I'd rather monitor my temperatures. CoreTemp can do it, but CoreTemp cannot control fan speeds. So I would need some tool to hide a HDD that is visible and enabled in device manager, to be hidden from certain utilities (i.e SpeedFan) so they can't read SMART data.

"I have a Bell expressvu 5100 satellite receiver. Its about 5-6 years old and has a Seagate U6, 40GB drive in it. ..."

Seagate U-series. ewww. Those are pretty bulletproof suckers. Especially U5 was pretty horrible performer even when compared to other drives of it's era. I have one. It warms up decently for 2-platter 5400rpm but it doesn't get hot. Not bad sectors, no reported errors. After cold-starts, during seeks, it sometimes makes the most odd and load screeching noises. It might also corrupt data sometimes...
Antec 1200 | HX520W | Commando | Q6600 G0 @ 3.15GHz | Noctua NH-U12F | 8GB of RAM | HD 4670 (passive)
7 TB of storage: 1x 1TB 1st gen GP, 1x 1TB 2nd gen GP, 1x 2TB 3rd gen GP, 1x 7200rpm F1, 2x 5400rpm F2 EcoGreen

#9 brickman

brickman

    Member

  • Member
  • 5 posts

Posted 26 December 2008 - 11:15 PM

The cause is SpeedFan utility that monitors temperatures of mobo, CPU and HDDs (via SMART). Other temperatures are polled every 10 seconds or more frequently but even SMART polling is done once a minute. WD unloads faster than that, even at most conservative setting in wdidle tool and result is always 1440 cycles.

I have "solved" the problem by stopping using SpeedFan


whiic, you're a genius. SpeedFan is indeed the problem. I disabled it and it solved it. I would have never guessed it. Thank you.

#10 K15

K15

    Member

  • Member
  • 389 posts

Posted 27 December 2008 - 09:15 PM

But at this rate, it will reach 2.6 million cycles within 5 year service life. However, it may survive over 2 million cycles... and 5 years is a long time anyway even without unload cycles. I have the same problem with my GreenPowers and would rather not have 1440 unloads a day. The cause is SpeedFan utility that monitors temperatures of mobo, CPU and HDDs (via SMART). Other temperatures are polled every 10 seconds or more frequently but even SMART polling is done once a minute. WD unloads faster than that, even at most conservative setting in wdidle tool and result is always 1440 cycles.


Oh I know. I wasn't trying to justify the drive's behavior, just trying to ease the original poster's concerns.
As for speedfan, are you saying that every time it polls the drive temperature, the drive unloads the heads? That would certainly explain things.



Seagate U-series. ewww. Those are pretty bulletproof suckers. Especially U5 was pretty horrible performer even when compared to other drives of it's era. I have one. It warms up decently for 2-platter 5400rpm but it doesn't get hot. Not bad sectors, no reported errors. After cold-starts, during seeks, it sometimes makes the most odd and load screeching noises. It might also corrupt data sometimes...



They really are awful aren't they? The drive is a U6 and has AAM enabled. For a satellite receiver, seek time is irrelevant, so it doesn't much matter. I must admit that the drive is very quiet for a ball bearing drive and the seeks are nearly inaudible. Just for fun, when I had it out to rip shows off it, I ran HDtach. Access time was something like 23ms. :P
I also have a U5. Back when 40GB was big, it was used purely for data (ie MP3) storage. In that computer the 8GB Maxtor (boot/app drive) was FAR faster.
I would die if I had to use a U-series drive as a boot/app drive. Something I don't get is how Seagate always kept the "official" seek time spec on those drives as 8.9ms. Who were they kidding?
Posted Image

#11 whiic

whiic

    hauptsturmführer

  • Member
  • 921 posts

Posted 28 December 2008 - 06:14 AM

K15: "As for speedfan, are you saying that every time it polls the drive temperature, the drive unloads the heads? That would certainly explain things."

Temperature of a HDD is read by polling (all) SMART values, then taking temperature attribute and interpreting raw data field. And polling SMART values does wake up a HDD from unload mode or spin it up if it's spun down. If I use any SMART monitoring too, the result is the same. It's just that SpeedFan runs on the background and it polls all connected drives automatically with 1 minute intervals that creates the problem.

Instead of trying to hide certain HDDs from SpeedFan (without preventing other monitorin tools from accessing them), it would probable be easier to crack the SpeedFan and make it not poll HDDs. How? I dunno.

K15: "They really are awful aren't they? The drive is a U6 and has AAM enabled. For a satellite receiver, seek time is irrelevant, so it doesn't much matter. I must admit that the drive is very quiet for a ball bearing drive and the seeks are nearly inaudible. Just for fun, when I had it out to rip shows off it, I ran HDtach. Access time was something like 23ms."

Even if you disable AAM and allow it to make twice the amount of seek noise, it'll still be 20+ ms and the amount of seek noise is just unbelievable for such slow seeks. Performance is Bigfoot-like but despite smaller actuator, noise is worse. (Also, considering lower rotational speed, 3600rpm, to obtain same access time, Bigfoot has to actually seek faster.)

I think not many modern HDDs are that slow seekers even with AAM enabled. Maxtor DM9 and DM16 were 20+ ms in silent seek mode but they at least became pretty completely inaudible and their performance mode was on par with competitors.

K15: "I would die if I had to use a U-series drive as a boot/app drive."

Somehow, I'm alive. I actually used it as it came with a pre-built computer. Win98SE installed so it didn't take a whole year to boot up. Later I once installed even WinXP on it and that is when I noticed that the performance of this drive was even worse than I had expected. I had performed HDTaching on it previously but it was even worse than the awfully bad synthetic benchmark results would have let me believe.

K15: "Something I don't get is how Seagate always kept the "official" seek time spec on those drives as 8.9ms. Who were they kidding?"

OEMs? No, they cannot be kidded... they just didn't care. If customer didn't return the product as defective, it was good enough. Unfortunately U-series was cheap and reliable, and thus got itself into many pre-built machines. Same probably applied to Bigfoots... even though they weren't notoriously reliable. But they at least were much cheaper. U-series was cheaper by only a small margin.
Antec 1200 | HX520W | Commando | Q6600 G0 @ 3.15GHz | Noctua NH-U12F | 8GB of RAM | HD 4670 (passive)
7 TB of storage: 1x 1TB 1st gen GP, 1x 1TB 2nd gen GP, 1x 2TB 3rd gen GP, 1x 7200rpm F1, 2x 5400rpm F2 EcoGreen



2 user(s) are reading this topic

0 members, 2 guests, 0 anonymous users