Critical Design Flaw Found in WD Caviar Green HDDs

[Regeneration]

It has come to our attention that Western Digital's Caviar Green HDDs suffer from a critical design flaw caused by an aggressive power-saving feature. Western Digital has developed a new technology called Intellipark (aka Idle 3 mode) and it is designed to reduce power consumption, in part by positioning the HDD's heads in a park position and turning off unnecessary electronics after 8 seconds of inactivity.

According to an in-house investigation and user reports', some software and operating systems are incompatible with the Intellipark feature causing endless head parking movement as the HDD continuously goes in/out of idle mode. This abnormal behavior creates stress on the HDD and that could lead to the following issues:

• Loud clunking/clicking/buzzing noises every few seconds.
• Artificially increases the number of load-unload cycles in S.M.A.R.T.
• Possibly shortens life-time of the HDD.
• Possibly reduces performance of the HDD.

Read more: http://www.ngohq.com/news/19805-critical-design-flaw-found-in-wd-caviar-green-hdds.html

 

[erixx]

i have 2 of those and i am very happy. no problems of parking under win7. but thanks

 

[Widjaja]

Western Digital have offered a way to disable the intellipark feature a while back.

 

[95Viper]

It has come to our attention that Western Digital's Caviar Green HDDs suffer from a critical design flaw

It took a while to come to your attention...
The referenced post over at SPCR, here:Is there a problem with head parks on WD Green HDDs? was posted back on Sat Dec 13, 2008 8:59 am.

I have one and never had the problems suggested. And, I have used it in a Linux machine, also, before.

Maybe, just got lucky or something; or, could be the way the owner/operator has set up their OS or apps.

But, thanks for the heads-up.

 

[inferKNOX]

What a time to leave us, Samsung!
j/k

 

[yogurt_21]

have 5 of them running at home. 2 in a nas in raid 1 and 3 in raid 0 in my main rig. No issues whatsoever. then again the specific model isn't mentioned. Mine are all EARS.

 

[W1zzard]

on a new 1.5 TB drive i have 737 hours on and 13864 load cycles.

737 * 3600 / 13864 = 1 park event every 191 seconds

at 300k estimated parks each 191 seconds = 15947 hours of lifetime = 664 days

clearly not "critical" .. wd's rma is very easy and fast anyway

 

[hat]

I'm pretty sure mine is never idle anyway, as I have a pagefile on it (and my system drive as well)...

Still, we should be aware of this.

 

[Sihastru]

Four of my many 2TB green drives stats:

1. WD20EARS
Power On Hours Count: 1274
Load Cycle Count: 9690

2. WD20EARS
Power On Hours Count: 1646
Load Cycle Count: 11836

3. WD20EARS
Power On Hours Count: 388
Load Cycle Count: 5148

4. WD20EARS
Power On Hours Count: 388
Load Cycle Count: 5095

I am not worried at all. There is no actual proof that links general drive failures to the head parking feature. It's just paranoia.

 

[newtekie1]

I swear this problem has been around for ages, it caused RAID controllers to freak out when the WD Green drives were first introduced, though a firmware fix helped the issue, I've been complaining about it for a while now. It is the reason I don't buy WD Green drives. If I want a low power drive I'll buy a Seagate LP.

Edit: To give you an idea of just how long this problem has been known, here is a post from Jan 2010: http://forums.whirlpool.net.au/archive/1367904

Edit2: Here is one from April 2009: http://www.networkedmediatank.com/printthread.php?tid=20686

 

[Sihastru]

I think some RAID controllers used to freak out because of the absence or presence of TLER and not because of the head parking feature.

 

[newtekie1]

I think some RAID controllers used to freak out because of the absence or presence of TLER and not because of the head parking feature.

The absences of TLER was a problem with the Black drives in RAID, but the head parking issue was been a problem with Green drives since long before the TLER issues.

 

[Regeneration]

Yes, this issue was discovered in 2008, but nobody bothered to post an article about it – so a lot of people are clueless. And by the way, this issue still appears even on the latest models.

I consider it critical as not all PC users monitor SMART. Those Green series are quite popular... you know, cheap hardware sells. Look at the SPCR's thread, some of the load cycles data is really troubling.

I've experienced this issue with a brand new WD10EARS drive from Feb/2011 on Windows 7 (fresh installation with single 3rd party software – FileZilla).

Again - take in mind - not all PC users are experienced like us

 

[yogurt_21]

Yes, this issue was discovered in 2008, but nobody bothered to post an article about it – so a lot of people are clueless. And by the way, this issue still appears even on the latest models.

I consider it critical as not all PC users monitor SMART. Those Green series are quite popular... you know, cheap hardware sells. Look at the SPCR's thread, some of the load cycles data is really troubling.

I've experienced this issue with a brand new WD10EARS drive from Feb/2011 on Windows 7 (fresh installation with single 3rd party software – FileZilla).

Again - take in mind - not all PC users are experienced like us

oh? how are you "experienced" in a way that I'm not? Can you beat 10 years of server management?

again i've seen no issues with any of my wd greens. This is a paranoia non issue.

I swear this problem has been around for ages, it caused RAID controllers to freak out when the WD Green drives were first introduced, though a firmware fix helped the issue, I've been complaining about it for a while now. It is the reason I don't buy WD Green drives. If I want a low power drive I'll buy a Seagate LP.

Edit: To give you an idea of just how long this problem has been known, here is a post from Jan 2010: http://forums.whirlpool.net.au/archive/1367904

Edit2: Here is one from April 2009: http://www.networkedmediatank.com/printthread.php?tid=20686

did you bother reading any of the comments on either article? both articles we're basically pointed out as incorrect in the comments below them. In all it seems most have been running for years without issue. Drives fail sure, but you can't just take a single wdgreen failure and enter a conspiracy theory. Wdgreens run fine, my raid 1 config has been up and running for 1 year without issue on my nas, no slowing no random park errors in smart, no issues whatsoever. We have several WDgreens running in servers at work that have been up for over 2 years consecutively, no issues on those drives either. At worst the feature can slow the drive down. Considering it's a damned green drive that's exactly what it supposed to do. lol If you want performance you should be going for the black editions anyways.

 

[newtekie1]

oh? how are you "experienced" in a way that I'm not? Can you beat 10 years of server management?

again i've seen no issues with any of my wd greens. This is a paranoia non issue.

did you bother reading any of the comments on either article? both articles we're basically pointed out as incorrect in the comments below them. In all it seems most have been running for years without issue. Drives fail sure, but you can't just take a single wdgreen failure and enter a conspiracy theory. Wdgreens run fine, my raid 1 config has been up and running for 1 year without issue on my nas, no slowing no random park errors in smart, no issues whatsoever. We have several WDgreens running in servers at work that have been up for over 2 years consecutively, no issues on those drives either. At worst the feature can slow the drive down. Considering it's a damned green drive that's exactly what it supposed to do. lol If you want performance you should be going for the black editions anyways

Read his post again, he said not all PC users are experienced "like us". You would be part of the "us" that is experienced.

This problem will show it self in different usage senarios. Just because you haven't seen the issue doesn't mean it doesn't exists. I'm sure I could find someone out there that still has a functioning Antec Smartpower PSU, that doesn't mean they weren't poorly designed pieces of garbage with insanely high failure rates.

The fact is these drives are designed for storage drives. When used as a storage drive, that isn't constantly accessed, there isn't a problem. However, when used as an OS drive, or a drive that is accessed a lot, it becomes a problem. I know I've seen a oddly large number of WD Green drives come into my shop failed that were used as OS drives, and I would venture to bet this is why. People buy them, or heck even OEMs buy them, thinking they are cheap and huge, so it is a good deal. They pop them in their machines to replace the older smaller OS drive, and think nothing of it.

And I've personally seen them freak out RAID controllers. I bought several of the first generation green drives for a RAID5 array on a highpoint controller. About once a week, the controller would mark one of the disks as failed. Reboot the machine and the RAID controller would see the drive again and rebuild the array and all was well again. Used the utility provided by WD to change the park time to 30 seconds and never had a disk falsely marked as bad again.

Is it a huge problem? No, not as big as many make it out to be. Will it shorten the life of the drive? You bet, but probably not to the point that is really matters for most users. Is it something people should be made aware of? Yes. Is it something WD should fix? Hell yes.

The part that probably gets most "angry" is the fact that it is just an easy fix, and WD has just not done it. Simply changing the firmware to park the heads after 30 seconds instead of 8 would reduce the problem to near non-existance in any usage senario.

 

[silkstone]

I have a new WD EARS

Power On Hours Count: 286
Load Cycle Count: 40
Whatever that means

Only Error is a UDMA CRC Error - but that is prolly due to the cable being squashed at the back of my case. i just did some cable management and the cable was twisted an akward way, when i fit it all back together, causing Bios not to see the drive. I don;t think you can ever clear that error but I should really get a new cable, but Meh.

Oh, maybe this would be useful to you: http://www.macobserver.com/tmo/article/french_retailer_data_offers_ssd_failure_rates/
A failure rate reported by a french retailer, not sure of the sample size tho.

 

[W1zzard]

Power On Hours Count: 286
Load Cycle Count: 40
Whatever that means

that means that in 286 hours your drive used up 40 of the estimated 300,000 cycles it can do.

some basic math will give you an estimate how many hours you can expect to get out of the drive.

 

[erixx]

senerios

fak yeah! haha scenarios lol Forcing the boundaries of our imagination&understanding, we, the non native English speaking majority!

 

[silkstone]

I know I've seen a oddly large number of WD Green drives come into my shop failed that were used as OS drives,

This is kind of common sense really tho. The more a drive is used, the more likely it is to fail. Hence an OS drive is more likely to fail than a backup drive, right? I'm sure of all the failed drives that companies recieve for RMA, the majority are OS drives, or drives that are used a lot. Furthermore, i'm ure the majority of systems would have only the 1 drive, hence that drive being an Os drive. Ergo, an os drive is the most likely kind of drive to be returned for replacement.

I'm sorry, but to back up claims we really need un-bias facts, not rumors from the inter-webs.

that means that in 286 hours your drive used up 40 of the estimated 300,000 cycles it can do.

some basic math will give you an estimate how many hours you can expect to get out of the drive.

Thanks. That makes sense, i should have known that, but i didn't realize a drives lifetime was based on the Load Cycle Count. What exactly is load cycle count in laymans terms? i'm assuming it means the amount of times the disk has been spun up to full operating speed and the heads moved in to position?

 

[yogurt_21]

Read his post again, he said not all PC users are experienced "like us". You would be part of the "us" that is experienced.

This problem will show it self in different usage senerios. Just because you haven't seen the issue doesn't mean it doesn't exists. I'm sure I could find someone out there that still has a functioning Antec Smartpower PSU, that doesn't mean they weren't poorly designed pieces of garbage with insanely high failure rates.

The fact is these drives are designed for storage drives. When used as a storage drive, that isn't constantly accessed, there isn't a problem. However, when used as an OS drive, or a drive that is accessed a lot, it becomes a problem. I know I've seen a oddly large number of WD Green drives come into my shop failed that were used as OS drives, and I would venture to bet this is why. People buy them, or heck even OEMs buy them, thinking they are cheap and huge, so it is a good deal. They pop them in their machines to replace the older smaller OS drive, and think nothing of it.

And I've personally seen them freak out RAID controllers. I bought several of the first generation green drives for a RAID5 array on a highpoint controller. About once a week, the controller would mark one of the disks as failed. Reboot the machine and the RAID controller would see the drive again and rebuild the array and all was well again. Used the utility provided by WD to change the park time to 30 seconds and never had a disk falsely marked as bad again.
Is it a huge problem? No, not as big as many make it out to be. Will it shorten the life of the drive? You bet, but probably not to the point that is really matters for most users. Is it something people should be made aware of? Yes. Is it something WD should fix? Hell yes.

The part that probably gets most "angry" is the fact that it is just an easy fix, and WD has just not done it. Simply changing the firmware to park the heads after 30 seconds instead of 8 would reduce the problem to near non-existance in any usage senerio.

that's a raid controller issue, not a drive issue. If the drive works and the raid controller is posting errors you need to take a look at the raid controller.

besides anyone who think's it's great to use large drives as os drives has their own problems. All storage servers I've ever built have a seperate os drive or a seperate os raid. Not to mention power saving on a boot drive is going to result in painfully long load times.

 

[newtekie1]

This is kind of common sense really tho. The more a drive is used, the more likely it is to fail. Hence an OS drive is more likely to fail than a backup drive, right? I'm sure of all the failed drives that companies recieve for RMA, the majority are OS drives, or drives that are used a lot. Furthermore, i'm ure the majority of systems would have only the 1 drive, hence that drive being an Os drive. Ergo, an os drive is the most likely kind of drive to be returned for replacement.

Yes, but the point was that I don't see them coming in when used as storage drives. The usage matters. I see more WD Green drives failed as OS drives than any other type of drive when used as an OS drive.

I'm sorry, but to back up claims we really need un-bias facts, not rumors from the inter-webs.

Ok, you want some facts, try reading some. WD released a tool to help the issue, how much more of a fact do you need?

that's a raid controller issue, not a drive issue. If the drive works and the raid controller is posting errors you need to take a look at the raid controller.

Not when every other drive used with the controllers worked fine. The delay to unpark the heads when the drive went idle caused the controller to think the drive was dead and mark it as bad. Same issue with the TLER, but for a different reason.

besides anyone who think's it's great to use large drives as os drives has their own problems. All storage servers I've ever built have a seperate os drive or a seperate os raid. Not to mention power saving on a boot drive is going to result in painfully long load times.

Large drives as OS drives are a must. Why you ask? Because density increases performance, so a large drive, with denser platters, is faster than a small drive with less dense platters. So even if you are only going to use 250MB, it is better to have a 1TB drive than a 250GB drive, because the 1TB will be faster(assuming all the rest of the specs are the same).

As for your claim that power saving will result in "painfully long load times", it really makes me question if you have ever really used any of these drives, because the really aren't all that noticeably slower than something like caviar blue.

 

[cheesy999]

Large drives as OS drives are a must. Why you ask? Because density increases performance, so a large drive, with denser platters, is faster than a small drive with less dense platters. So even if you are only going to use 250MB, it is better to have a 1TB drive than a 250GB drive, because the 1TB will be faster(assuming all the rest of the specs are the same).

As for your claim that power saving will result in "painfully long load times", it really makes me question if you have ever really used any of these drives, because the really aren't all that noticeably slower than something like caviar blue.

the drive doesn't have too be larger though, the 500gb f3 is actually slightly faster then the 1tb f3, just needs high gb/platter

 

[newtekie1]

the drive doesn't have too be larger though, the 500gb f3 is actually slightly faster then the 1tb f3, just needs high gb/platter

You're right, but that is the exeption, not the rule. Generally, all things being equal, the larger drive will be faster.

 

[cheesy999]

You're right, but that is the exeption, not the rule. Generally, all things being equal, the larger drive will be faster.

i'd reply with something about the 5000 rpm 2-3tb drives you get but i feel like that would be trolling - especially as this is what a lot of wd greens are and were off-topic slightly anyway

none of this changes the fact that instead of just disabling this feature they should off fixed it, through a firmware update or through a hardware recall

and power saving dosn't kick in on a boot drive normally as most of the time on most people pc's leaving it in idle causes the av to kick in or some other background program carries on running, i've never had my drive fall to sleep as an os drive but then again its a hitachi deathstar so it is probably compleatly different in design eg:no intelli park

 

[95Viper]

It is not FUD or un-founded rumor or any other BS.
And, I am not pointing fingers, but some need to check facts for themselves before they make posts that dispute the facts.:shadedshu

Just because it has not happened does not mean it could not. Read the WD FAQs!

Did any of you nay sayers take the time to look at the link to the WD Knowlrdge Base FAQ(The S.M.A.R.T Attribute 193 Load/Unload counter keeps increasing on a SATA 2 hard drive)
that links to here, too, The S.M.A.R.T Attribute 193 Load/Unload counter continue to increase for the WD RE2-GP SATA II hard drives?

I don't believe they would put out these FAQs for sh*ts and grins... must be truth in there somewhere.