Page 3 of 10

Posted: Sun Jan 11, 2009 2:32 pm
by signal64
From a general use with XP perspective, I've been running two WD500AACS-00ZUB0 with XP for about a year now.

One has the OS while the other is for data (i.e. not raid 0).
It's a general useage PC that is mostly powered off when not in use.

Looking at the smart data in everest shows:

Drive 1 - WinXP
Power-On Time Count: 3899
Power Cycle Count: 228
Load/Unload Cycle Count: 109959

Drive 2 - General Data Use
Power-On Time Count: 3813
Power Cycle Count: 186
Load/Unload Cycle Count: 62955

Both drives were purchased at the same time, but drive 1 was on it's own for awhile before the second was added.

Posted: Fri Jan 16, 2009 7:28 pm
by akromatic
whoa , i've not realised that WD green power had this issue untill i came across this thread.

well after reading this, i got confused and would like to know if my 750GB drive is having issues as well but i have no idea how

i tried to download Active@Hard Disk Monitor but it doesnt support my drive as its connected via USB(as im using a 5yr old laptop and have no desktops in the house)

so what can i use to check my drive(preferably free) and "fix" it as i do have quite alot of important/critical stuff in it?

Posted: Fri Jan 16, 2009 10:57 pm
by bgiddins
I think there's still the question of whether this is really even an issue - it's entirely possible that it's simply the wrong value being reported by firmware.

WD Green drives aren't exactly failing in droves like the Seagate Barracuda 7200.11 drives are at the moment.

Posted: Mon Jan 19, 2009 11:17 am
by zds
Nick Geraedts wrote:As Mike mentioned before, there's been no evidence of the high load/unload cycle counts being the true values (are the raw values or the actual number, as watchtower pointed out). Furthermore, nobody has confirmed if the data provided by SMART does/should correspond to the specs in the datasheets. Is SMART counting Intellipark cycles, when the datasheet lists full load/unload cycles?
The SMART data reported by WD drive itself confirms this: the drive tells it's been rated for 600k cycles and be replaced at that point. If WD thinks this information is incorrect, they should rise up and say so. Right now they are not denying or confirming anything, even when pressed, which is the worst they can do.

What WD should do is to either come up and say "ok, we goofed up the SMART data, your drives are fine" or then "your drives have firmware that's killing them, here's how to fix it". That's how responsible manufacturers work.

This way or that way, I would not buy GP drives myself until this issue has been cleared by WD; any drive can fail, but it does not make sense to by a drive that's *known* to have issues.

Posted: Mon Jan 19, 2009 12:42 pm
by bgiddins
zds wrote:This way or that way, I would not buy GP drives myself until this issue has been cleared by WD; any drive can fail, but it does not make sense to by a drive that's *known* to have issues.
You're right, I'm going out to replace my WDs with Seagate 7200.11s right now :?

Posted: Mon Jan 19, 2009 1:27 pm
by Nick Geraedts
bgiddins wrote:
zds wrote:This way or that way, I would not buy GP drives myself until this issue has been cleared by WD; any drive can fail, but it does not make sense to by a drive that's *known* to have issues.
You're right, I'm going out to replace my WDs with Seagate 7200.11s right now :?
Huh? Aren't you going around the wrong way here?

The 7200.11's have serious firmware issues right now, and there doesn't seem to be very good response from Seagate regarding the issue. The SMART data reported from the WD drives could be alarming, but there haven't been any significant reported cases of them failing in mass numbers like the Seagates are doing.

For what its worth, here's my SMART data just a few weeks since my last reply:

Code: Select all

Drive 0
  9 Power_On_Hours          0x0032   090   090   000    Old_age   Always       -       7572
193 Load_Cycle_Count        0x0032   090   090   000    Old_age   Always       -       332227
Drive 1
  9 Power_On_Hours          0x0032   090   090   000    Old_age   Always       -       7570
193 Load_Cycle_Count        0x0032   097   097   000    Old_age   Always       -       311617
Drive 2
  9 Power_On_Hours          0x0032   093   093   000    Old_age   Always       -       5250
193 Load_Cycle_Count        0x0032   074   074   000    Old_age   Always       -       378011
Drive 3
  9 Power_On_Hours          0x0032   093   093   000    Old_age   Always       -       5251
193 Load_Cycle_Count        0x0032   074   074   000    Old_age   Always       -       378736
Drive 4
  9 Power_On_Hours          0x0032   093   093   000    Old_age   Always       -       5251
193 Load_Cycle_Count        0x0032   075   075   000    Old_age   Always       -       377991
Drive 5
  9 Power_On_Hours          0x0032   093   093   000    Old_age   Always       -       5251
193 Load_Cycle_Count        0x0032   072   072   000    Old_age   Always       -       386726
Drive 6
  9 Power_On_Hours          0x0032   093   093   000    Old_age   Always       -       5253
193 Load_Cycle_Count        0x0032   001   001   000    Old_age   Always       -       671329
Drive 7
  9 Power_On_Hours          0x0032   093   093   000    Old_age   Always       -       5251
193 Load_Cycle_Count        0x0032   073   073   000    Old_age   Always       -       383658
There's only one drive that's got a cycle/hour ratio that's significantly higher than the others, and that's the one that was constantly dropped from the array. All of the others have actually had their cycle/hour ratio drop since last time.

Posted: Mon Jan 19, 2009 1:34 pm
by whiic
zds: "The SMART data reported by WD drive itself confirms this: the drive tells it's been rated for 600k cycles and be replaced at that point."

Nope. Threshold is 0. By SMART definition, value can be 1-255 so it can NEVER reach 0 even if you had 6 billion cycles. It will be stuck at 1 forever when reaching 600000 cycles.

Threshold 0 means that the attribute is for informative purposes only and isn't used to determine whether HDD is dead or not.

"ok, we goofed up the SMART data, your drives are fine"

They did NOT goof it. SMART values can be pretty freely used as manufacturers see fit. "You are doing it wrong". This "problem" has nothing to do with SMART: it has to do with unload cycle rating and the rate unloads are performed. Whether SMART increments or not does not change the graveness of the (potential) issue.

And so far we are only talking about potential issue for long-term reliability.

"This way or that way, I would not buy GP drives myself until this issue has been cleared by WD; any drive can fail, but it does not make sense to by a drive that's *known* to have issues."

Potential = possible, not confirmed, not known.

bgiddins: "You're right, I'm going out to replace my WDs with Seagate 7200.11s right now"

LOL! Talking about KNOWN issues... 7200.11 actually has one confirmed, OFFICIALLY confirmed issue: they disappear from BIOS, typically after a few months of use. Most affected drives have firmware SD15 and affected capacities range from 500 to 1.5 terabytes.

And before you think I'm a liar, Seagate itself actually admitted the flaw. Firmware update available for drives that have not yet died. Free data recovery services offered for those who already had their "superior" Seagate die. WTF? Free? Yes, then it must be a friggin' serious issue because services like that aren't really offered that often.

http://www.msfn.org/board/index.php?showtopic=128092 (43 page discussion thread... and growing)
http://techreport.com/discussions.x/16232 <-- one of dozens of web based news sites reporting it
http://techreport.com/discussions.x/16246 <-- Seagate admits the issue
http://seagate.custkb.com/seagate/crm/s ... cId=207931 <-- Seagate's support page on the issue

If you think that potential but yet unconfirmed longterm issue with rapidly increasing load/unload count on WDs is worse than Seagate's confirmed and admitted series of deaths at just few months of age... well, be free to think so. But I won't agree with you. Nor will I recommend anyone to agree with you. But in the end, it's your own choice. I recommend to avoid 7200.11 like plague.

EDIT: Here's one more article:
http://www.theregister.co.uk/2009/01/16 ... re_plague/

Posted: Tue Jan 20, 2009 12:55 am
by bgiddins
bgiddins wrote:You're right, I'm going out to replace my WDs with Seagate 7200.11s right now :?
...does no-one on this board get sarcasm? :roll:

I'm very happy with my WDs (mine don't appear to be reporting high cycles), but I don't believe it's an issue anyway until these drives start dying.

Data that's not backed up isn't data you want anyway.

Posted: Tue Jan 20, 2009 2:31 am
by whiic
lol. I didn't read your previous post. Would I have read it, sarcasm would have become very obvious.

Posted: Tue Jan 20, 2009 6:06 am
by zds
bgiddins wrote:
zds wrote:This way or that way, I would not buy GP drives myself until this issue has been cleared by WD; any drive can fail, but it does not make sense to by a drive that's *known* to have issues.
You're right, I'm going out to replace my WDs with Seagate 7200.11s right now :?
Fortunately these two are not the only two drives in existence *grin*. With this data I'd just pick regular WD or Samsung F1.
whiic wrote: "This way or that way, I would not buy GP drives myself until this issue has been cleared by WD; any drive can fail, but it does not make sense to by a drive that's *known* to have issues."

Potential = possible, not confirmed, not known.
Earlier in this thread watchtower posted WD response telling the head unloads reported in SMART data are real unloads, and they indeed go beyond what the drive was designed to go within normal lifespan. I'd call that an issue and an confirmation.

I might not cause drives to die instantly or in masses, but as WD admits, it increases risks compared to drive that operates within it's rated limits. So no reason to panic, but a reason to pick some other drive instead.

Posted: Tue Jan 20, 2009 7:44 am
by whiic
zds: "Fortunately these two are not the only two drives in existence *grin*. With this data I'd just pick regular WD or Samsung F1."

Why not 7K1000.B? Like those two, it's also 7200rpm with 3 platters. Why would WD's 7200rpm be any better? And didn't/doesn't F1 have some incompatibility issues with some controllers?

There is no perfect drive but there is a valid reason for prefering 5400rpm drive over 7200rpm. If you don't know what it is, take a look at top left corner of your browser. It reads there on the title.

"Earlier in this thread watchtower posted WD response telling the head unloads reported in SMART data are real unloads, and they indeed go beyond what the drive was designed to go within normal lifespan."

Exceeding the minimum sustained unload count.

"I'd call that an issue and an confirmation."

Unless drives actually start dying after exceeding the specified count, I won't call it an issue.

"I might not cause drives to die instantly or in masses, but as WD admits, it increases risks compared to drive that operates within it's rated limits."

It might increase risk of failure. But if you choose Samsung, it's CSS based, and that choice of technology itself is risky due to stiction and lower non-operational shock tolerance. And WD's 7200rpm (if it doesn't have the same "issue" as GP) spends more time flying it's heads over the platters which spin at speeds of over 100km/h, each second increasing the likelyhood of a headcrash that could possibly not occur if heads were unloaded as they are most of the time with GP.

GreenPower has no known issue. It has one known potential issue.

Posted: Thu Jan 22, 2009 1:12 pm
by Devonavar
WD has officially responded to this issue, acknowledging that the 8 second unload timer can conflict with certain software, causing these unduly high spin-up counts. They recommend taking care of it on the software end (i.e. using and OS and applications that don't hit the hard drive every 10 seconds). However, they also offer a firmware patch to disable head unloading, as well as a utility that allows for a lower power, longer spin-up cycle. The unload timer can also be adjusted between 8 seconds and 5 minutes.

Also listed are the exact model numbers of the affected drives.

More here in WD's support database.

Details of the Low Power Spin-up feature here.
WDC wrote:Affected Models:
WD1000FYPS-01ZKB0, WD7500AYPS-01ZKB0, WD7501AYPS-01ZKB0 drives Load/Unload counter for S.M.A.R.T ATTRIB 193 continues to increase?

Symptom:
WD drives are designed to reduce power consumption, in part by positioning the heads in 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, operating systems, 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 Western Digital’s RE2GP, and artificially increases the number of load-unload cycles.

Solution:
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 to 30 seconds or so, thereby gaining substantial power savings and eliminating superfluous activity.

Most customers, when made aware of the unnecessary activity caused by their systems, have modified their utility, operating systems, or applications to take advantage of Western Digital's advanced power-saving mode. Other customers have requested a utility (See link below) 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. This update is described in WD's Process Change Notice PCN 2579-701324-A02 (see attached PDF file).
Please click on this link, RE2GP Idle Mode Update Utility, to download the utility.

NOTE: 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, WDSpinUp Utility, also available at the RE2GP Download page.

Posted: Thu Jan 22, 2009 1:16 pm
by zds
Devonavar wrote:WD has officially responded to this issue, acknowledging that the 8 second unload timer can conflict with certain software, causing these unduly high spin-up counts. They recommend taking care of it on the software end (i.e. using and OS and applications that don't hit the hard drive every 10 seconds). However, they also offer a firmware patch to disable head unloading, as well as a utility that allows for a lower power, longer spin-up cycle. The unload timer can also be adjusted between 8 seconds and 5 minutes.
Thank you very much for digging this up for us!

Posted: Fri Jan 23, 2009 8:44 am
by Nick Geraedts
Thanks for the link Devon. I was wondering if anyone knew if the firmware was compatible for the non-RE drives as well. I know that most of the features in the various firmwares are compatible (you can enable TLER on non-RE drives for example) across a particular series, but I'm not sure if that applies to enterprise drives vs consumer drives.

Posted: Fri Jan 23, 2009 4:36 pm
by whiic
What are models WD1000FYPS-01ZKB0, WD7500AYPS-01ZKB0 and WD7501AYPS-01ZKB0? Are these supposed to be Greenpower drives? It does refer to GP in the description... typical WD mess.

If the tool supports WD10EACS, I'll probably set the value to slightly over 1 minute, so that it doesn't cause high unload counts if I leave SpeedFan running and polling SMART periodically. Optimum would be if I could prevent SpeedFan from polling HDD temperatures and I could leave the unload at 8 seconds.

EDIT: Is it a tool or is it a firmware? Or do you have to first update firmware, then run utility to customize? Or do you run the customization utility which creates a corresponding firmware image in-situ and writes the firmware to disk?

Ok... these are RE edition GPs only. I wonder why only RE. And why is 1000GB RE2GP WD1000FYPS instead of WD10EYPS? It has NOTHING to do with RE feature as WD7500AACS vs WD7500AYPS use same A. Why do WD create the oddest imaginable naming convention that is difficult to read, contains absolutely no relevant data, and still doesn't even use the naming convention consistently? Just so unbelievably WD.

Posted: Fri Jan 23, 2009 11:04 pm
by cipher7
WDC wrote:Affected Models:
WD1000FYPS-01ZKB0, WD7500AYPS-01ZKB0, WD7501AYPS-01ZKB0 drives Load/Unload counter for S.M.A.R.T ATTRIB 193 continues to increase?
So does this mean that the 1TB WD10EACS-00ZJB0 and WD10EACS-00D6B0 drives aren't affected?

Posted: Sat Jan 24, 2009 1:18 am
by akromatic
hmm im still confused wether there is a problem or not and which models are effected

well most importantly if my drive is effected (WD7500AACS) as I cant check its duty cycle because its in a USB enclosure and I've got no access to a desktop

Posted: Sat Jan 24, 2009 3:02 am
by whiic
"So does this mean that the 1TB WD10EACS-00ZJB0 and WD10EACS-00D6B0 drives aren't affected?"

No. It just means WD doesn't offer a solution for these drives. (Solution = firmware update or utility to configure unload delay.) To solve it, you need to tweak your OS and other software to fit your HDD's behavious instead of the other way around.

Both 00ZJB0 and 00D6B0 are definitely affected by "it". I'm still refusing to call it an issue, so it's just "it" instead.

Though, noisewise it may be an actual issue. The clicks during load and unload process are slightly louder than non-AAM seeks and are audible in decently quiet system... in a completely silent system, some silencing fanatics would probably consider them obtrusive. But for reliability, I wouldn't consider "it" a confirmed issue.

For people who are sensitive to seek noise, maybe giving Samsung EcoGreen 5400rpm is a good option? They are (AFAIK) CSS based like their 7200rpm counterparts, and thus they have no power-saving modes, thus no clicking sounds related to them. Though some say the seeks themselves are quite loud on EcoGreens... probably not load as per se, but with idle noise being less than 7200rpm F1, the 5400rpm variant might have more noticeable seeks due to lower noisefloor of the system.

Too bad Hitachi doesn't make 5400rpm drives as they use APM for unload configuration and the aggressiveness of power saving has been customizable since... the first model that actually had the unload technology? But Hitachi has declared that they won't try to "cut corners" in power and noise by reducing rpm to achieve results easy. This means that unless 5400rpm is re-introduced as an industry standard, Hitachi won't follow WD and Samsung. Seagate would probably be more likely to be the next to introduce 5400rpm but I wouldn't bet on it.

"hmm im still confused wether there is a problem or not"

It's wise to be confused because the reality is that we don't know if there is a problem or not. The nature of the potential issue (or "it") is long-term reliability. GPs have been around for a year and it's service life is 5 years. So, we'll probably have the answer to whether there is a problem or not in the next 4 years. :)

"and which models are effected"

My guess is: all GreenPowers, RAID and non-RAID, 1st generation (250GB/platter) and 2nd (333GB/pl), flagship capacities and downscales.

WD Black is probably unaffected as it's performance oriented and recovering from unload would cause a 0.5 to 1.0 second delay in requested IO operation.

WD Blue? I don't know, but probably unaffected. This is just a guess as well.

Posted: Sat Jan 24, 2009 6:29 am
by zds
akromatic wrote:hmm im still confused wether there is a problem or not and which models are effected

well most importantly if my drive is effected (WD7500AACS) as I cant check its duty cycle because its in a USB enclosure and I've got no access to a desktop
With current information the only way to be sure is to take the drive out of the enclosure, plug directly to a computer and check SMART data. Or then just hope for the best.

Posted: Sat Jan 24, 2009 7:02 am
by whiic
SMART is not the problem. It definitely is not the problem.

SMART monitoring keeps track of load/unload count with 00ZJB0 variant but with 00D6B0 the attribute (which is still present) is not updated when unload occurs. Whether load/unload count is kept track by SMART does not change how much wear the drive suffers during loading and unloading.

The ONLY sure way of telling whether the drive has the "issue" is to listen to it or otherwise measure vibration (either acoustic frequency or sub-audible vibration). The easiest and cheapest instrument to make the measurement would be your ears. Especially as the drive is in a USB enclosure, listening to it should be easy as it's separated by distance from other system's noise sources.

So forget SMART already. The panicked masses already made WD cripple SMART to hide the "issue" from your eyes. Now you have to trust your ears instead.

I know zds likes to spread panic and probably thinks the fact that WD is hiding the "issue" means that there is a real issue. It's not necessarily so. If there's panic, there's lost revenue regardless whether it's true or not. Truth doesn't weigh a shit in modern society.

Posted: Sat Jan 24, 2009 7:10 am
by zds
whiic wrote:The ONLY sure way of telling whether the drive has the "issue" is to listen to it or otherwise measure vibration (either acoustic frequency or sub-audible vibration). The easiest and cheapest instrument to make the measurement would be your ears.
This is sure, but it helps *only* when the drive is already mechanically reaching end of it's lifespan. It does not help at all determining if your drive is hi or low risk unit.

Whether you call it "problem", "issue" or "gift from heaven" does not change the fact that there *is* a difference in wear and tear if your drive does 5 head unloads per month or 50k. While this data can be hidden from SMART tools, some drives do reveal it and thus keep you informed. And when you know your drive is hi-risk one, you can prevent the damage beforehand by using wdidle3.exe or similar.

Waiting for the drive to fail before touching an issue that can be semi-easily prevented is just plain stupid.

Posted: Sat Jan 24, 2009 2:55 pm
by Cirkus
I have three WD drives in my HTPC and I have run WDIDLE3 and disabled the idle 3 timer (APM?) completely on all of drives.

The main reason I did this is because there is an annoying clicking sound when the head is loaded/unloaded. Before I ran WDIDLE3 there was an annoying clicking sound quite often. I also noticed that the Load/Unload cycle count would increase at every click. Now my drives don't click anymore and the Load/Unload cycle count value doesn't increase all the time. Thank you for the useful tips and links in this thread.

The drive that was clicking a lot was actually the drive with my Windows Vista installation (C drive). It is a WD Scorpio Blue 2.5" drive (WD5000BEVT).

Right after I had run WDIDLE3 my computer wouldn't boot up. I got scared that I had actually destroyed three perfectly good drives but after removing a USB memory stick attached to the computer I was able to boot Windows. I don't know what happened since it usually works to boot with the USB memory stick connected but it works now and I am happy.

Cirkus

Posted: Sat Jan 24, 2009 3:20 pm
by whiic
"This is sure, but it helps *only* when the drive is already mechanically reaching end of it's lifespan."

WTF?! Are you saying that load and unload operations should be completely inaudible and the fact that I can actually hear clicking sound each time they either load or unload means that all my three GPs have been dying since the second I started using them? Isn't it miraculous that out of these three drives that are supposed to already be reaching the end of their life spans since first unload cycles (as unload clicks were always audible) they have hundreds of thousand of cycles and months of completely faultless operation? Come on. Making clicking sound during load/unload is perfectly normal.

"It does not help at all determining if your drive is hi or low risk unit."

On the contrary: the more often I hear it click, the more often I know it performs the load/unload cycle. If I take an average interval between clicks and determine the frequency, then multiply it by whatever duration I want (like one year) and then divide that with number 2 (two) (as one click is unload, second click is reload, equals only 1 full cycle), I get the approximate number of cycles the drive has gone through last year or the approximate number of cycles the drive will go through during the next year (assuming interval between cycles remain the same).

It's not rocket science and to determine whether your drive is unloading more often than necessary, a 50% margin of error is still acceptable. We only need to roughly determine how frequent the unloads are: once in 10 seconds, once a minute, once every 5 minutes, or something approximate like that.

The minimum sustained load cycle count also more than likely has a very high margin of error. For example the minimum unload counts of laptop drives have been exceeded by millions of cycles without a problem. It may be that either the minimum unload count specs are rated lower than they could be, or that only that failure likelyhood increases only slightly after exceeding the limit. We don't know which one, but the fact remains that laptop drives rated for hundreds of thousands can survive millions.

"Whether you call it "problem", "issue" or "gift from heaven" does not change the fact that there *is* a difference in wear and tear if your drive does 5 head unloads per month or 50k."

Same could be said that it's an undeniable fact that wear and tear is bigger for a drive that has it's heads flying for 720 hours/month instead of 72 hours/month. And GreenPower does exactly that: it reduces hours heads spend flying, thus reducing likelyhood of critical headcrash and degree of gradual wear of the platter surface. It powers of some electronics when unloaded, saving not only electricity but risk of those circuits failing in use.

For you to say that the 5 vs 50k unload matters more than 720h vs 72h of flight time you are either
A) a HDD engineer who has spent a year or two studying the issue or
B) incredibly dumb to think you know it when you are barely guessing.

"While this data can be hidden from SMART tools, some drives do reveal it and thus keep you informed. And when you know your drive is hi-risk one..."

Eh???

You acknowledge that some drives with the "issue" hide the "issue" and that SMART cannot be trusted, yet you recommend that instead of listening to the drive to determine how quickly it unload during idle, you should trust the SMART you know it untrustworthy...

I believe you are not the HDD engineer type...

"Waiting for the drive to fail before touching an issue that can be semi-easily prevented is just plain stupid."

Trusting any HDD to work flawlessly is plain stupid. Also, trusting SMART values (which you yourself acknowledge untrustworthy) to determine presence of this "issue" is plain stupid.

Of course the risk caused by a potential issue should be reduced by using the best available methods (which vary from scenario to another). Avoiding potential issues is wise and I intend to tweak my system to avoid unnecessary unloads. Having the heads loaded over extended periods of idle is not the optimal solution: optimal solution would be to cut the unnecessary disk IO requests at it's source: OS, SpeedFan, whatever.

However, the potential issue isn't grave enough to panic about it and so far we have nothing that indicates that potential issue is a factual issue. The minimum sustained load cycle spec is probably concervatively rated so reaching 2 million cycles in 5 year service life might only mean slight increase in failures related to wear of unload mechanism or the surface where heads are reloaded. That slight increase of failures might be negated by reduced failures caused by overheating or headcrashes. Honestly I don't know. To claim to know it without doing research would be utterly stupid at this point of time when there's no statistical evidence. Unless you have a laboratory where you have a statistically significant amount of GreenPowers performing unloads at 6+ cycles a minute you don't have data because the drives in typical use get much less.

Having a flamewar over potential issue really is not worth it. Keep making back-ups and it doesn't much matter. Quite likely you won't need that back-up.

I'm support all discussion in minimizing the number of cycles by whatever means: wdidle3.exe, new firmwares, new WD utilities, configuring SpeedFan properly to reduce SMART polling, configuring the OS to reduce I/O traffic. Not only is it a potential risk but it's source of noise.

This "GreenPowers have an issue, avoid them, buy Black/Velociraptor/7200.11/whatever instead" is seriously uncreative. There's obvious reasons why we would like to stick to GreenPower. The only worthy competitor is Samsung EcoGreen and that has poor availability. 500GB laptop drives cost like they were made of gold and performance would be inferior; with 320GB laptop drives you'd need three and performance would be inferior; with 7200rpmn laptop drives you'd need at least three or more, it'll be more noisy, and cost like they were made of gold; and with SSDs you'd need to be a millionaire and need to use all SATA ports on mobo and possibly a few PCI slots as well to mount enough of them to have a decent capacity, leaving no room for future expansion.

Let's stick to solutions, no product placement. We can find decent alternatives on other threads of Silent Storage section.

How to disable SpeedFan SMART polling on selected HDDs

Posted: Sat Jan 24, 2009 3:34 pm
by whiic
Optimal solution is to tweak the system so that unnecessary unloads do not occur but when system is idle, HDD would unload and stay that way for long periods of time. Disabling unload altogether makes absolutely certain no unloads occur except when spun down (STAND-BY or powering down). Setting a higher value reduces number of clicks but reduces time spent unloaded. Finding optimum depends on situation.

Truely best solution would be to solve problem at it's source (OS, programs, background processes) and keep the unload delay to minimum, maximizing both power saving and minimizing clicks. A very tricky thing to do. For OS disk, I'd say it's impossible, and prefer using wdidle3.exe to either disable or set it to 1+ minutes (which practically both mean the same for OS drive: to never unload).

To contribute to prevention of noise issue, here's how to
prevent SpeedFan from periodic polling of SMART values
I found that in SpeedFan, going to Configure -> Temperatures and un-ticking the GreenPowers, they are no longer SMART polled once a minute. While I've been running my system for a few months without SpeedFan (I load it on boot to set low fan rpms but I quit the program after boot-up). I can now leave SpeedFan on after system reboot as the HDDs are no longer polled and thus I need not sacrifice on fan speed control.

During a 20 minute period, I heard zero clicks from 00ZJB0 and when I verified it by reading SMART, load/unload count had only increased by one... and that was caused by me polling SMART values, not by SpeedFan, not by Windows.

Posted: Sun Jan 25, 2009 12:16 pm
by Devonavar
Some thoughts...

I've been running a 1TB GreenPower in my HTPC for over a year now. I've clocked ~60K cycles, at a rate of roughly one every 30 seconds (my power-on hours are about 2K). Maybe I'm not a typical power user, but, even though my cycles are higher than they could be, I'm still not going to hit 300K (rated) until the 5 year lifespan of the drive is up anyway. I won't hit the 600K that SMART says is a problem until 10 years. For my usage pattern, this problem simply isn't.

Some anecdotal evidence: The same system has a Samsung MP0402H notebook drive that has clocked 800K cycles in about three years. It's still going strong (according to SMART, this will become a problem at 1,000,000 cycles). Load / unload technology has been around for a while, and it hasn't caused a rash of HDD failures. To be fair, desktop heads are bigger and may wear faster (or they may not), but it's not unreasonable to assume that WD has experience building for high cycle counts. On an unrelated note, the same drive has a power-on hours value of 84 YEARS. SMART data is *definitely* not as reliable as it could be.

Posted: Mon Jan 26, 2009 4:47 am
by whiic
Power-on "hours" 84 years... do you mean that power on time is around 735000 units? Because I don't think it's standardized to mean "hours". How about minutes? That way it would be 12250 hours or 510 day or 1.4 years. With 3 years of use, it would mean average 5 hours powered on per day. Does that sound roughly correct for the way you have used your computer?

When it comes to many values, SMART is very accurate. It counts events relatively accurately. Only failure prediction isn't accurate, because future failure is hard to predict even knowing accurate event counts such as accurate number of bad sectors today. It's hard to predict how many there is tomorrow. SMART cannot do that. And then there's failure types that do not show any bad events before the HDD is suddenly gone: like toast PCB (common on WDs) or corrupted firmware (common on recent Seagates).

When event counts are times an event has been triggered and value increment, and thus accurate, certain attributes are notoriously inaccurate due to inaccurate sensors. This is not SMART's blame. Temperature readings will be all over the place if the sensor is bad or badly located.

Some attributes, while they may count events accurately, may be impossible to interpret without insider knowledge. Stuff like raw error rate and hardware ECC recovered definitely fall under this category.

Due to sensor inaccuracy and non-standardized nature of raw data field I would trust SMART values only when I have a point of comparison. I.e trust temperature reading only when I have verified it's accuracy with external sensor at atleast two different temperatures (preferably as far as possibly without overheating or freezing the drive) and the smoothness of temperature graph (SpeedFan) during warm-up.

Same applies to unload counts: if I hear the drive loading and unloading while load/unload cycle count remains either zero or equal to number of start/stop counts, I know the value isn't updated. Either the raw data field is not used (only value changes) or the attribute is just left there but not used at all. (Which raises the question why it's present if it's not used for anything.)

EDIT: obviously 1.4 power on years within 3 year period is not 5 hours a day but roughly twice that much.

Posted: Tue Jan 27, 2009 5:40 am
by maxcherry2
Hi There,

I have used Western Digital drives all my PC life starting with 36gb Raptors and currently have two 320GB in windows raid 0 and Mirrored on an Intel Matrix chipset the pc also has a 500GB as Storage and even an old 120GB sata for downloads.

More recently I have purchased an Icybox NAS and installed to GP 1TB drives in a span array. I brought these drives for there low noise/power as the NAS was always going to be the limiting factor on performance when transferring data.

I have had this set up for 4 months and store music, HD movies and photos on it I was up to about 700GB of data when one of the drives failed!!!

Now the box runs Linux and to be honest the NAS was much loader from day one then I thought it would be. The NAS has a 10 minute spin down so when not is use it is fine. However when downloading or streaming to my PS3 the things clanks and clicks almost constantly. I thought this was because the drives are not in a PC case so they can be heard easier but after reading this lot I think there are major issues with these drives in some instances. I mean 4 months and death?

Oh yeah all the data was on the drive that died.

Just my luck :cry:

Posted: Tue Jan 27, 2009 9:39 am
by Nick Geraedts
maxcherry2 wrote:Oh yeah all the data was on the drive that died.
I don't meant to sound harsh or blunt about this, but simply storing files on an external source is not a backup solution. A backup means that you have a duplicate copy of your file(s), preferably in different physical locations.

You could have started copying your data to a Seagate drive or a Samsung and had the exact same experience. Early deaths are not part of the discussion here - whether the SMART data actually means something in terms of drive longetivity is.

Posted: Tue Jan 27, 2009 12:15 pm
by Cryoburner
I'm curious why this thread is pinned with a title like "WARNING: WD Green Power drives may kill themselves". That just sounds like scaremongering to me. It's based on mostly groundless information as well, since as far as I've seen, no one's actually reported a GreenPower drive failing from having a higher than normal number of unload cycles.

Even if a drive's life was shortened by this effect, it wouldn't be because the drive 'killed itself', but rather because some software was waking it up on a regular basis. Granted, Intellipark could be redesigned to handle situations like that better. For instance, it could keep track of the length of time between disk accesses, and adjust the time before parking its heads accordingly. On a drive that was constantly being polled every 10 or 20 seconds, this would effectively disable Intellipark.

Needless to say, the title of this thread sounds like a tabloid headline, and its claims are comparatively speculative. If it's to remain pinned, it should at least be retitled, and have a link to WD's utilities included in the first post, to actually make it useful for anyone experiencing frequent head parks on their GreenPower drive.

Getting the runaround from WDC about firmware for WD10EACS

Posted: Tue Jan 27, 2009 1:15 pm
by spcruser
After following this thread for some time, and checking the info in the main article about WDC's existing firmware fix for (only) the RE2 drive (WD1000FYPS) I've been trying to see if they even have firmware in development to address this issue on the WD10EACS. However I have so far only received boilerplate replies (and worse ... "it's just a counter" being my 'favorite' so far) after several attempts at clarification and escalation :(

A major reason I bought these drives (I have 6 currently, two of which are RMA replacements, and I'm begin to understand why that might have happened) was their reputation for quiet, reliable data drive performance in these forums - at least in late 2007 when I started to put my systems together.

I've tried to reduce the issue with the suggested changes to SpeedFan but I need SpeedFan to properly monitor and control the two quiet PC systems I built with these drives, so not monitoring their temperatures (for now) is counterproductive for me.

Some of my WD10EACS drives have this #193 Load/Unload Count attribute up over a MILLION in not much more than 1 year of usage. This seems to be over three times the 'Reliabaility / Data Integrity' spec value, as in WDC document 2879-701229. I'm disappointed by WDC's approach to this problem, but I guess I shouldn't be surprised at 'mere' corporate ineptitude in a time when the general public is finally learning of the grotesque corporate corruption and incompetence that plagues our society.

Given that Seagate seems to be 'problematic' also, has anyone a recommendation for a good quality, decent value 1TB drive that is actually properly supported by it's manufacturer? I don't think I'll be rushing to buy any more WDC products.