WD10EADS WD Caviar Green - 1TB >> excessive Load Cycle Count

[NAS-t]

I have two WD10EADS 1TB drives in my Duo. The first was installed a months ago as Drive 2, mirroring the 500 GB Seagate that came with the Duo. The WD drive seemed to work just fine, so I recently bought a second and installed it as Drive 1.

In just a week of use, the newer drive is exhibiting a large and growing number of Load Cycles. After a little research, I found that the WD10EADS drive comes in a number of flavors. The first configuration had 4 platters (each approx 250 GB). A second variant had 3 platters. The latest vintage, which is the kind I just bought, has a pair of 500 GB platters. I am seeing vastly different behavior from the two WD10EADS units, as reported by the Duo's SMART reports.

Drive 2: WDC WD10EADS-00L5B1. This is a 3-platter drive, installed in my Duo last month.
- power-on hours = 1012
- Load Cycle Count = 670
- Power Cycle Count = 78

Drive 1: WDC WD10EADS-00M2B0. this is the newer 2-platter drive, installed last week.
- power-on hours = 74
- Load Cycle Count = 3477
- Power Cycle Count = 10

The LCC value on Drive 1 increases by 2 or 3 every minute or so.
I have my Duo configured to spin-down the drives after 10 minutes. Even so, the LCC continues to climb, even when the drives should both be spun down.

I've done some web research and found extensive coverage of this issue at QNAP, Synology and Quiet PC Review forums. The root cause appears to be that the WD power-saving 'green' design parks the heads after 8 seconds of inactivity, while the Linux OS tickles the drive every 20 seconds or so, thereby unparking the heads. The Green series of drives are rated by WD for 300K load cycles during the drive's lifetime. Thus, there is some concern that the constant park/unpark operations will wear out the drive prematurely.

One customer received back from Western Digital a troubling statement:

...The unit is not recommended for NAS or RAID use, due to its end-user design. But, under normal use, rather than strenuous or highly demanding tasks, there should be not much issue with the unit.
Johnny T.
Western Digital Service and Support

I can't explain why my two WD10EADS drives are behaving differently. I can think of three possibilities:

- the two different models have different firmware, and they park/unpark the heads using different algorithms. (note that both my drives have the same 01.00A01 firmware).
- the two models are both parking/unparking the same, but are reporting differently through SMART. In other words, my Drive 2 also has large LCC count, but I just don't know it. It's been suggested by some experts that WD deliberately masked the actual operational statistics reported by some of their other drives in order to quell consumer concerns.
- my Drive 1 is operating as the Linux boot drive, and is being frequently accessed by the kernel as the kernel goes about its routine tasks. But Drive 2 is merely a RAIDed data disk that does not get repetitive accesses.

Detailed discussion and debate on this topic can be found here:
http://forum.qnap.com/viewtopic.php?f=23&t=14273
http://www.silentpcreview.com/forums/viewtopic.php?t=51401&postdays=0&postorder=asc&start=0
http://forum.synology.com/enu/viewtopic.php?f=124&t=11682

Based on my observed SMART reports, I feel some caution is warranted re the use of WD10EADS drives in RAIDed ReadyNAS systems. Be aware that all such drives are not created equal: you need to know the full part number to know if you have a 4-platter, 3-platter, or 2-platter unit.

[Han Solo]

I can't explain why my two WD10EADS drives are behaving differently. I can think of three possibilities:
[list]- the two different models have different firmware, and they park/unpark the heads using different algorithms. (note that both my drives have the same 01.00A01 firmware).

This is the one I most agree with. WD created an algorithm to handle this load/unload issue. If the drives see excessive load/unloads it is supposed to set the timer to higher value something like 5 minutes (can't remember the exact value at this time). I think that sometimes for whatever reason the algorithm is not working exactly as expected. I talked to WD about the utility that a user can use to manually change this setting and they provided us with the same utility but the user does not have to pick a time but rather the software just sets it to 5 minutes. We will be posting a link to this software after we try it out here.

[NAS-t]

Mr. Solo,
Thank you for replying. I am delighted to hear that the Jedi Masters are aware of and working on this little problem.

Some additional information:
I provided incorrect specs regarding the firmware in my two WD drives ( I really must get new glasses). My older (3-platter) drive has firmware 01.01A01. The newer (2-platter drive) has firmware 01.00A01. It is the latter that is experiencing the increasing LCC.

Yesterday I exchanged my drive for another identical model. I was hoping that my local store might still have some of the older stock (the 3-platter drive with the 01.01A01 firmware). Unfortunately, they only had one in stock, and it is from the same manufacturing lot as the one I was returning. The replacement drive is now installed as Drive 1 in my Duo, and the LCC is growing by 2 or 3 counts per minute. I can hear the heads park/unparking. So I conclude that the LCC behavior I was seeing was not due to a defect in the particular drive that I had, but is inherent in the 01.00A01 firmware.

I have opened a ticket with WD asking them about the difference between 01.00A01 and 01.01A01, and how they propose to address the issue. I have not yet received a reply. If I hear anything from them, I'll update this thread.

As you know, WD issued a DOS utility some time ago (wdidle3.exe) that a user could invoke to modify the drive's timer. The utility was intended for a different line of drives, and has since been withdrawn by WD. They claim it will void the warranty if used on the EADS drives.

Cheers,
Dave

[NAS-t]

It's been a week since I swapped my WD10EADS-00M2B0 for a replacement WD10EADS-00M2B0. It's been running in my Duo as Drive 1. The drive has 124 power-on hours. The Load Cycle Count continues to grow, with just over 3000 park/unpark cycles.

My other drive (Drive 2) is a WD10EADS-00L5B1. It has just over 1200 power-on hours, and a mere 780 Load cycles.

Let's do the math.
The older -00L5B1 with firmware 01.01A01 has .65 load cycles per hour of operation.
The newer -00M2B0 with firmware 01.00A01 has 24.2 load cycles per hour.

It's clear the two different variants behave differently.

A trouble ticket has been open at WD for over a week. They have yet to reply.

[pfile]

so i assume that if one has their box configured to spin the disks down, that the load/unload count will probably not increment, at least while the drives are spun down? unless this is actually still a bad idea on WD drives; i can't tell if those caveats about WD and spindown in the official compatibility list is old information by now.

[NAS-t]

so i assume that if one has their box configured to spin the disks down, that the load/unload count will probably not increment, at least while the drives are spun down?

@pfile,
I do have my Duo configured to spin-down, after 20 minutes. You are correct in that the heads are not parking/unparking while the drives are spun down: the LCC count doesn't change and I don't hear the heads chattering away. The LCC would be much higher if I did not have the drive spinning down.

The caveat you are referring to in the Official Drive Compatibility list pre-dates version 4.1.5. The Netgear footnote indicates that the WD GP drives should be ok to spin down on 4.1.5 and later releases.

[lindebrand]

Hello everyone.

I bought a Duo 2 days ago with one 500GB (Seagate Pipeline) drive in it. So far I'm impressed, even though I hardly know anything about this stuff .

I was planning to use 2x1TB drives in my Duo, and I ordered 2 of these drives (Two WD10EADS-00M2BO were delivered). I've so far just inserted one of the drives in the before empty slot.
And I'm getting a bit worried about these drives not going to last. My 1TB-drive has a power hour of 7 hours, and just broke 100 LCC. (without Spin-down enabled).

Is this going to be a huge problem? I want the drives to last as long as possible of course, and as I've understood WD can't support more then 300.000 LCC and the warranty is canceled after that? I've read that the typical lifetime a disk can have 300.000-600.000 LCCs before breakdown, but I've also read about people with more then 1.5 million of them. Should I worry?

I'm going to call my supplier and see if I can get the one drive I've used replaced. If I can, should I?
Or should I just stick with both of them even though rapidly increasing LCCs? As I've not used one of them I can atleast send one back, and try and get another on the compatibility list.

Is there any advantages of having two identical discs instead of two different?

I hope you can answer my questions. Thanks in advance.
/lindebrand

[NAS-t]

@lindebrand

We're in exactly the same boat. (actually, a similar boat, as I have only one of the newer -00M2BO drives)

Based on my research to date, it's not clear that excessive LCC will lead necessarily to early drive failure. There's one body of opinion that high LCC is not a fundamental problem on its own, and that it's nothing to worry about. Others feel that if the drive does fail early (i.e. before the 3 year warranty expires), WD might refuse to RMA it, claiming that the drive was abused by the constant parking and unparking.

Personally, I have a concern about premature failure. I'm also bothered by the noise. My Duo sits right on my desk, and until my drives spin down, I can hear the heads on the -00M2B0 constantly parking and unparking. It's as if they're calling out to me "we're wearing out, we're wearing out..".

I opened a ticket with WD on this 10 days ago. They have yet to respond. I think it's likely that they messed up with the firmware in the -00M2B0. The previous version (with 3 platters) does not have this problem. But there are reports that the first version of the WD10EADS (with 4 platters) also experienced excessive LCC. Makes me think that WD tweaked the firmware for the -00L5B1 drive (rev 01.01A01) to address the issue, but forgot to propagate this fix into the firmware for the -00M2B0 (rev 01.00A01).

[lindebrand]

Thanks for the answers NAS-t.

I'm guessing that I'm not going to be able to return the drive I put in the NAS as it's used. So, if I can't return it I've got a few options.

1. Use the drive in the NAS, and hope for the best.
I had a Seagate Barracuda 7200.11 1TB die on me a while back that I'm gonna get replaced, hopefully the drive that I'm getting instead of it is okay and compatible. I can then use that one in the NAS together with the WD.

2. Try and get the WD-drive sold. And then buy one that is compatible (or another Seagate-drive). And use it with the Seagate-replacement.

3. Try and get both the WD-drive and the replacement Seagate sold. (I hear they still have problems with their firmware? That was the reason I chose these WD-drives instead. My Barracuda died after just 5 months, so I'm worried if their problems still remain it will happen again. Everyone gives the Seagate drives 1 of 5 stars and writes CAUTION in big letters in the comments of the webshop doesn't really help either hehe )

Not sure what I'm gonna do yet.

Asking the same question as in my last post: Is there any advantages of having two identical discs instead of two different?

/lindebrand

[NAS-t]

Asking the same question as in my last post: Is there any advantages of having two identical discs instead of two different?

No real advantage. Indeed, the ReadyNAS is designed expecting that users will not have identical disks. This makes it very easy to grow the size of the array.

There is one 'soft' benefit of having two identical: you can watch the SMART parameters and perhaps tell if one of the drives is beginning to fail by comparing their SMART stats. All drive vendors do not implement/measure the various parameters the same way, so it can be hard to compare some of the numbers side by side on dissimilar drives.

There is a disadvantage. If you have two identical drives and they both have the same design defect, there is a chance that you could lose both of them at the same time, and then 'poof' your RAID redundancy is gone. I'm thinking of the Seagate drives with the known brick-my-drive firmware flaw earlier this year (I had one of them). But statistically, the chance of a fatal firmware defect killing both drives at the same time is vanishingly small. You're more likely to lose all your data from a lightning strike zapping your entire system, or from theft.

[lindebrand]

Got some news. I talked with my webshop today and I was able to return both discs and get the money back. Even though one is used

So, I'm going to return both of them. Get my money back, then wait for the replacement of the broken Seagate to arrive.
Hopefully it's compatible and everything (sigh, never thought it would be such a hassle with drives). And then I can order a identical Seagate or another drive.

I've got no idea if a execcive LCC leads to drive failure, and I'm not bothered by the noise as you are NAS-t. I've got my Duo beside my computer on the floor and it's more noisy then the Duo make by far.

But, I've got the chance to replace my drives so I will, just in case.


I just remembered that I've got a WD Caviar Green a while back and put in my stationary. I don't know what type it is, 2-platter or not. I'll check when I get home and check with some S.M.A.R.T-program how many LCCs that drive has.
But as I've understood this is a "problem" not only caused by the NAS right? Just that you'd like more reliable drives in a NAS?

/lindebrand

[NAS-t]

But as I've understood this is a "problem" not only caused by the NAS right? Just that you'd like more reliable drives in a NAS?

/lindebrand

This is a contentious issue - there has been a lot of discussion and (heated) debate on the other forums that I cited in the first post on this thread. I don't think it's fair to say that this problem is "caused by the NAS". It's more of an interaction between the type of disk I/O that the NAS Linux-based OS generates, and the WD green design.

The root cause is that the WD "green" design aggressively parks the heads in order to reduce power consumption. They call this feature "IntelliPark". I understand that there is an 8-second timer in the WD firmware -- if the drive has not been asked to do anything for 8 seconds, the heads get parked. It seems that the RAIDiator OS interacts with the drive frequently (some experts claim every 20 seconds or so) which briefly unparks the heads. Then after 8 seconds of no activity, Intellipark parks the heads. Rinse and repeat, with a few park/unpark cycles happening every minute. Ironically, if the NAS was doing any real work (ie copying files), the heads would never get a chance to park, and thus the Load Count would not keep growing.

Frankly, I do not know if this behavior makes the Caviar Green drives any less reliable. We won't really know until a couple years from now whether there are large numbers of drives failing with huge LCC counts.

[93turbo]

I can't explain why my two WD10EADS drives are behaving differently. I can think of three possibilities:
[list]- the two different models have different firmware, and they park/unpark the heads using different algorithms. (note that both my drives have the same 01.00A01 firmware).

This is the one I most agree with. WD created an algorithm to handle this load/unload issue. If the drives see excessive load/unloads it is supposed to set the timer to higher value something like 5 minutes (can't remember the exact value at this time). I think that sometimes for whatever reason the algorithm is not working exactly as expected. I talked to WD about the utility that a user can use to manually change this setting and they provided us with the same utility but the user does not have to pick a time but rather the software just sets it to 5 minutes. We will be posting a link to this software after we try it out here.

Have you found any solution in this case yet?

[NAS-t]

No solution found. Indeed, WD refuses to acknowledge there re even is a problem. They replied to the trouble ticket I opened and told me:
- "We have not recognized the "load/unload" issue as a problem for those drives".
- "We have no firmware upgrades for this drive".

Meanwhile the Load Cycle Count on the affected WD10EADS drives continues to climb.

Disk1 WD10EADS-00M2B0
- firmware = 10.00A01
- LCC = 23600
- Start/stop = 480
- Power-on Hours = 571
- - - - - - - - - - - - - - -> 41 Load Cycles per hour

Disk2 WD10EADS-00L5B1
- firmware = 10.01A01
- LCC = 1187
- Start/stop = 1187
- Power-on Hours = 1656
- - - - - - - - - - - - - --> 0.7 Load Cycles per hour

My newest WD10EADS drive has a third the power-on hours of my first such drive, yet has twenty times the number of park/unpark cycles.

There is another anomaly between the two drives: the 'good' drive reports exactly the same number of Load Cycles and Start/Stops. That's not the case with the other one. It's apparent that the firmware in the two drives is behaving quite differently.

Note that I have disk spindown configured in FrontView (20 minute timer). Once the ReadyNAS decides to spin down the drives, they both go into sleep mode, and the constant park/unpark cycle ceases (as evidenced by the absence of noise).

Other than the constant head chatter, the system is working fine. Neither drive reports any read errors or reallocated sectors.

-Dave

[93turbo]

Here is my info

I have tree WD10EADS with following status:

Disk 1
WD10EADS-00M2B0
Power On Hours 44
Power Cycle Count 6
Load Cycle Count 1781

Disk 2
WD10EADS-00M2B0
Power On Hours 19
Power Cycle Count 3
Load Cycle Count 1153

Disk 3
WD10EADS-65L5B1
Power On Hours 1356
Power Cycle Count 18
Load Cycle Count 28

I don't buy Green power again. I go for the Black edition when this disks have give up.