High LCC count on WDEARS drives

Well, looks like I get to eat crow .  I've been ignoring the advice to adjust WDEARS drives with wdiddle because I hadn't personally seen any issues with the several WDEARS drives I've been using for months.  Well, now it looks like I finally am starting to see some problems:

(click for full size)

I just noticed the issue, so I haven't had a chance to see if the LCCs are continually increasing, or if they are stable at those numbers.  I'll compare them to this screenshot over the next few days and weeks.

Obviously I want to avoid all of these WDEARS drives failing at once.  Should I start preemptively replacing them?  I do have a single 2 TB Hitachi I could throw in the array right away.  The others would have to wait a few days or weeks.

Also, I know about the HPA on the cache drive, and I don't care to fix it.  It isn't harming anything.  The parity drive's errors have also been stable for a long time, so I'm not too worried about it either.

Well, looks like I get to eat crow .  I've been ignoring the advice to adjust WDEARS drives with wdiddle because I hadn't personally seen any issues with the several WDEARS drives I've been using for months.  Well, now it looks like I finally am starting to see some problems:

 

(click for full size)

 

I just noticed the issue, so I haven't had a chance to see if the LCCs are continually increasing, or if they are stable at those numbers.  I'll compare them to this screenshot over the next few days and weeks.

 

Obviously I want to avoid all of these WDEARS drives failing at once.  Should I start preemptively replacing them?  I do have a single 2 TB Hitachi I could throw in the array right away.  The others would have to wait a few days or weeks.

 

Also, I know about the HPA on the cache drive, and I don't care to fix it.  It isn't harming anything.  The parity drive's errors have also been stable for a long time, so I'm not too worried about it either.

I set the LCCs in myMain to go off at a pretty low level so that people would see them before the numbers got crazy high.  There is nothing you can ever do to lower them - but you can slow down the increase.

BTW, if you set the "hpa_ok" attribute to 1 on the drive settings page, you can turn off that HPA warning.  Just click on the drive's ID (last few digits of the SN).  The drive settings dialog will open.  Under the bottom setting (Other Notes) enter hpa_ok, and to the right enter 1.  Click save.  Go back and refresh and you'll see that the warning is gone.  Just don't forget if you want to assign it to the array.

Raj, You're counts are high when i compare them to my drives.

Same firmware

Your:(sdf) 2878 Power on Hours with 25912 LCC counts

Mine: (sdb) 3666 Power on Hours with 8138 LCC counts

Mine: (sde) 2504 Power on hours with 4557 LCC counts

Other firmware:

Your: (sdg) 7162 Power on hours with 170700 LCC counts ( WOW!!! )

(Don't know to what limit these drives can go without trouble)

I did adjust all of my WD20EARS drives with WD-iddle after a few weeks working with UnRaid because of the high counts with the standard setting on everyday base.

Now it's oke i guess, only a few per day.

My counts for (sdb) on 10/07 were 3380 Power on Hours with 8089 LCC counts

That means for 286 hours only 49 counts  

Now the total average is going down comparing to the Power on Hours

I did not change anything for that WD20EARX. That model did not got those high counts in the first few weeks after installing.

Now 825 Power on Hours with only 442 LCC counts.

I replaced a previous drive with 150K LCC count with my local dealer without a problem.

If that (your sdg) drive was mine i replaced it, and for the others , adjust them with Wd-iddle for now.

(click for larger image)

I replaced SDG with a spare Hitachi drive.  Drive rebuild and subsequent parity verify completed without errors.

I will RMA SDG.  I will run wdiddle on my remaining WD EARS drives and hopefully prolong their life a bit.  Once they reach more than 100,000 LCC I will RMA them as well.  Hopefully I'm able to stagger the RMAs such that I'm never down more than one disk at a time.

It is definitely a pain to have to go through all of these steps, but at least MyMain gave me plenty of advance warning.  Thanks for that, Brian!  Thanks also to Rembro for your advice.

Wow, i'm glad I ran across this today.  I had not heard

I have one WD20EARS and it has 35900 LLC's for 7759 pwr on hours.  

Is this enough to get WD to RMA the drive?  You just tell them it has high LLC's?

I did find a note in another forum (reading on this issue) that said the specified limit is 300,000 or in another place it said this was the minimum at which the chance of a problem increases.  At what point would they RMA the drive?

Also, where do you find the correct wdiddle to use with this drive?  The only one I could find on WD's website is

wdidle3_1_05.zip

This utility is designed to upgrade the firmware of the following hard drives: WD1000FYPS-01ZKB0, WD7500AYPS-01ZKB0, WD7501AYPS-01ZKB0.

EDIT:

I also noted this count is coming up on my Samsung green drives, though notably much smaller numbers.  The WD has 4.6 cycles per hour of use and the Samsungs are starting out with about .5 cycles per hour (they are really new).

Is this typical for any green drive?  Should I do anything with the Samsung drives?

Yikes  I never really paid much attention to that number, but now I'm starting to worry...

According to wikipedia, the limit for 2.5" laptop drives is 300-600k, so that's a pretty useless figure. Google couldn't really give me any more definite numbers.

According to http://forums.anandtech.com/showthread.php?t=2085685 it should be possible to set this head parking timer using hdparm. Does anyone have any experience with that?

... It is definitely a pain to have to go through all of these steps, but at least MyMain gave me plenty of advance warning.  Thanks for that, Brian!  Thanks also to Rembro for your advice.

You are welcome. I remember when I added that warning thinking it may be warning people too early about an LCC issue that may or may not be a problem. But I figured it was better to know early, before the value was so high you'd feel compelled to RMA a drive.

I was a huge WD fan in the 1T EACS / EADS days. But since the EARS have been Hitachi exclusively. Hope Hiachi acquisition helps WD get their shit together. No reason these load cycle counts need to be this high.

Is this enough to get WD to RMA the drive?  You just tell them it has high LLC's?

No idea, but I'll let you know as soon as I find out!

According to this article, the drives have been certified for up to 1 million load/unload cycles. They also list an hdparm option to avoid problems:

Linux users add following (hdparm -B 255 /dev/sdX where X is your hard drive device). ATA users can disable APM usually controlled via BIOS and/or OS.

I'll probably try this on my cache drive to begin with, see if it has any effect.

According to this article, the drives have been certified for up to 1 million load/unload cycles. They also list an hdparm option to avoid problems:

 

Linux users add following (hdparm -B 255 /dev/sdX where X is your hard drive device). ATA users can disable APM usually controlled via BIOS and/or OS.

 

I'll probably try this on my cache drive to begin with, see if it has any effect.

Thank you!  I'll try it on my SDG drive above before RMAing it.

I tried running hdparm and I'm getting the following error:

root@Tower:~# hdparm -B 255 /dev/sde

/dev/sde:

setting Advanced Power Management level to disabled

HDIO_DRIVE_CMD failed: Input/output error

HDIO_DRIVE_CMD(identify) failed: Input/output error

Any ideas?

Also, will the drive still spin down after doing this?

Hi Raja,

Not that I try to be a smarta** or anything, but I think you should avoid those WDC Green for the moment. I've had a few bad experiences (5 times to be exact) with WDC Green drives. They were great at first, and just went slower suddenly, continued by deterioration from then on. They were not used as main hard drive, just data storage and were not accessed extensively.

The last drive in my system is an EARS that I used the wdidle3 tool when I got the drive to disable the timer. I'm now glad I did that. I think I need to look into doing that with my other WD drives.

root@Media02:~# smartctl -a /dev/sda | grep  Load_Cycle_Count
  9 Power_On_Hours          0x0032   086   086   000    Old_age   Always       -       10515
193 Load_Cycle_Count        0x0032   193   193   000    Old_age   Always       -       23984
root@Media02:~# smartctl -a /dev/sdb | grep  Load_Cycle_Count
  9 Power_On_Hours          0x0032   083   083   000    Old_age   Always       -       12923
193 Load_Cycle_Count        0x0032   190   190   000    Old_age   Always       -       31369
root@Media02:~# smartctl -a /dev/sdc | grep  Load_Cycle_Count
  9 Power_On_Hours          0x0032   096   096   000    Old_age   Always       -       3361
193 Load_Cycle_Count        0x0032   199   199   000    Old_age   Always       -       4662
root@Media02:~# smartctl -a /dev/sdd | grep  Load_Cycle_Count
  9 Power_On_Hours          0x0032   085   085   000    Old_age   Always       -       11276
193 Load_Cycle_Count        0x0032   181   181   000    Old_age   Always       -       59054
root@Media02:~# smartctl -a /dev/sde | grep  Load_Cycle_Count
  9 Power_On_Hours          0x0032   096   096   000    Old_age   Always       -       3591
193 Load_Cycle_Count        0x0032   200   200   000    Old_age   Always       -       582

I checked this at home as well last night, and saw my EADS and EARS drives are in the 200K region!

What the hell!! 

I too ignored the warnings from earlier, but now will be applying the fix asap.

For anyone looking for a quick synopsis of the issue and how to fix, and don't feel like searching through the forums again:

Link

Thanks for the reminder Raj!

This site http://tibi78.dyndns.org/wordpress/2011/07/wd-load-cycle-count-issue/ might be helpful? Have not tried it myself but am concerned since have data on discs already.

I checked this at home as well last night, and saw my EADS and EARS drives are in the 200K region!

 

What the hell!!  

 

I too ignored the warnings from earlier, but now will be applying the fix asap.

 

For anyone looking for a quick synopsis of the issue and how to fix, and don't feel like searching through the forums again:

 

Link

 

Thanks for the reminder Raj!

Wow, that link was really useful.

Yea, I used the instructions from that link to disable the idle timer on all of my WD20EARS drives.

Only difference is that I used unetbootin to create a FreeDOS bootable USB - http://unetbootin.sourceforge.net/

Oh, and I did download and use the 1.05 version of wdidle3.exe from here: http://support.wdc.com/product/download.asp?groupid=609&sid=113〈=en

Just to note, the 1.05 version of wdidle3.exe worked fine on both my EARS and EADS drives to disable intellipark.

I basically just created a bootable DOS usb stick, put wdidle3.exe on it, removed my unraid usb stick from my server, booted up with the dos usb stick and ran "wdidle3 /d" and it did the job on all my WD drives.

I've already done this on my two WD20EARS drives and I don't hear any constant "clicking", but I can't seem to figure out how to find out how high my LCC count is.

How would I go about finding out what the current number is?

I've already done this on my two WD20EARS drives and I don't hear any constant "clicking", but I can't seem to figure out how to find out how high my LCC count is.

 

How would I go about finding out what the current number is?

 

 

Get a SMART report for the drive -- unMENU is the easiest route to that.

I've already done this on my two WD20EARS drives and I don't hear any constant "clicking", but I can't seem to figure out how to find out how high my LCC count is.

 

How would I go about finding out what the current number is?

 

 

 

Get a SMART report for the drive -- unMENU is the easiest route to that.

Ah, yes, that was quite easy.

Thanks.

Just to note, the 1.05 version of wdidle3.exe worked fine on both my EARS and EADS drives to disable intellipark.

 

I basically just created a bootable DOS usb stick, put wdidle3.exe on it, removed my unraid usb stick from my server, booted up with the dos usb stick and ran "wdidle3 /d" and it did the job on all my WD drives.

Hi, Did you do 1 disc at a time or did it do all the discs at once?  Also once you loaded DOS where you still able to network into UnRaid box or did you have to reattach a head to the box? Sorry for the nooby questions. Have 6 EARS 2TB already loaded with data in unraid box and just trying to be careful. Thanks.

Once, you load dos from the USB, you cannot access unraid. You do need a monitor and keyboard to do this. Maybe you can get a way without a monitor, but that's something that can be discussed later if it's absolutely necessary.

Anyways, to do one drive at a time, I unplug all but one drive and used wdidle3 and kept doing it one after the other.

OK, I'm finally getting around to doing this.  I tried unetbootin first, but for some reason it didn't work for me (I would get to an A:\ prompt instead of a C:\ prompt, and it wouldn't let me switch to C:\, so I couldn't run WDIDLE).  I instead used the HP Tools/Win98 method which took longer but worked just fine.  Using a test server with only the single WDEARS drive connected, I disabled the timer on one of my WDEARS (my original SDG, with over 170,000 LCCs).  I'm now preclearing that drive (using the -n option to skip the preread) and I will add it back into the array as a new data drive.  I'm only preclearing it to bypass unRAID's clearing downtime - preclearing the drive after running WDIDLE is NOT a required part of this process.  As far as I know, you can run WDIDLE on a parity, data, or cache drive without putting that drive's data at risk.  I decided not to RMA the drive since it is rated up to 1 million LCCs, and 200,000 is still pretty far from that.  I think for now I'm going to leave my other drives as they are, unaltered.  I'll keep an eye on them over the next few weeks and months.  If the other drives break 200,000 LCCs within a few weeks or months then I'll run WDIDLE on them as well.

I'm curious how much power is wasted by disabling the WDIDLE timer.  I would like to test it, but I think I would probably have to run a longitudinal study over the course of several months to really find any significant data, and I simply don't have the resources to do that at the moment.

By the way, here's the latest status of the LCCs on the rest of my drives:

(click for full size)

You can see that I replaced the former SDG with a Hitachi drive.  I started a Calc spreadsheet to help me monitor the drives.  Here's what I've got so far:

The projected failure rates don't look too scary at the moment.  Even if SDE were to fail after 6 years of use, I would be happy with that lifespan.  In either 25 or 62 years I'm sure the TB scale will be laughably insignificant, somewhat like KB are today.  Granted, 16 days isn't much of a sample period, so I'll keep monitoring these drives every so often and adding to the spreadsheet.  I'll report back if I find anything alarming.  However, I believe that as long as these drives stay within acceptable limits, I would prefer the power savings to the potentially extended lifespan.

An interesting read about intellipark:

http://blog.fosketts.net/2011/04/27/western-digital-intellipark-feature-design-flaw/

This is no problem for operating systems like Microsoft Windows, which have been tuned to leave disk drives in the idle state for as long as possible. But Linux and RAID devices are not nearly as friendly. Assuming that hard disk drives are always spinning, many of these systems write data much more frequently, often every 10 to 20 seconds.

 

When Western Digital Caviar Green drives are used in systems that write data very frequently, IntelliPark can become a serious liability. Continually parking and on parking the heads causes wear and tear, potentially leading to drive failure. The difference between a park operation every 10 seconds and one every 5 minutes is dramatic, both in terms of drive longevity and power savings.

 

Enthusiasts have been quick to suggest that this feature is a critical design flaw, causing otherwise good hard disk drives to fail. Predictably, Western Digital sees things differently. They contend that the Caviar Green series was designed to be used in operating systems like Microsoft Windows, and suggest using other drive models in Linux and RAID systems. They also offer an idle mode update utility which allows end-users to tune this parameter or turn it off entirely.