Western Digital WD20EARS and read head parking every 8 seconds!?

[graysky]

Western Digital GreenPower drives - particular the WDXXEARS series have this annoying "feature" that parks the read heads of the drive once every 8 seconds! You can see this reflected in the Load_Cycle_Count value from smartmontools for example.

WD support provides tool called RE2GP Idle Mode Update Utility short for "wdidle" to fix this problem in other drives models -- but according to the disclaimer on the page, it is NOT for use with the EARS series of HDDs!

Additionally, you can find reports of folks complaining about this "feature" all over the web. In ad-hoc fix is to use hdparm to force a temporary reset from once every 8 seconds to once every 62 minutes like this:

hdparm -S 242 /dev/sdX

What do people in here think/know about this issue. Has anyone owning one of these HDDs successfully used the WD tool despite the warning on the website and if so what results have you met with? Any other fixes or thoughts are welcomed.

 

[Donovan]

My understanding is that you can disable the head parking even on the newer WD drives...what you cannot do is enable TLER which can be important for RAID setups. I haven't personally tested this on an EARS (I switched to buying Samsung drives when WD got restrictive with TLER) but I have disabled the head parking on an EADS and many posters to the TivoCommunity.com forums have done it with EARS drives.

There's a lot of good discussion of this issue on the Tivo hacking sites because the head parking interferes with soft reboots on Tivo boxes and must be disabled. In particular, read this FAQ entry as well as the last several pages of the very long thread it appears in:

http://www.tivocommunity.com/tivo-vb/showthread.php?t=370784#drey


Start with v1.05 of wdidle3 (that should be the version on WD's site) and use "wdidle3 /d" to disable followed by just "wdidle3" to verify success. If that doesn't work, there are at least two other versions floating around on the TivoCommunity.com forums (v1.03 and v1.00), as well as discussions about other command-line parameters to use if /d doesn't work (you can set the timeout to its maximum value instead of 8 seconds).

 

[anonxlg_]

what's so bad about head parking every 8 seconds? noise and wearing?
I have a wd20ears coming

 

[ClockerXP]

what's so bad about head parking every 8 seconds? noise and wearing?
I have a wd20ears coming

Can somebody answer this question? I would like to know too.

Thanks,
CxP

 

[Mark R]

Can somebody answer this question? I would like to know too.

Thanks,
CxP

It's basically a noise and wear, as has been stated.

Loading/unloading is quite noisy - it makes a click-clack sound.

It also causes some wear on the heads. Most drives are rated for about 300,000 park cycles. However, a lot of people with the EARS drives are finding that they get through about 50,000-70,000 park cycles per month. Of course, the drives don't die when they get to 300,001 - but people are concerned as the drives are set to get to 10x their design limit over a 5 year period. However, as yet, there doesn't seem to be any real problem here.

 

[RebateMonger]

Can anybody explain or point me to a link on how folks measure how many head parks they've had on a WD disk?

Thanks!

 

[Mark R]

Any SMART tool can read the number of cycles. It's called -
load/unload count. (SMART ID number 193)

The raw value is the number of cycles.

 

[sub.mesa]

Some people claim to have 50% of the specified Load Cycle Count within a month, due to it happening every 8 seconds.

I'm not sure if this issue is related to Linux or also applies to other OS. I think it should be possible to disable it, though.

 

[Campy]

Well i have a WD20EADS drive and i'm up to 46241 load cycles now. I've had the drive for a year. This number is hardly indicative of a problem is it? I've never noticed any sound from it either, my raptors on the other hand i hear quite well, at least when they're working.
I have a WD20EARS drive on the way. Is there anything about the new drive that's significantly different that I should be aware of?

 

[Campy]

My understanding is that you can disable the head parking even on the newer WD drives...what you cannot do is enable TLER which can be important for RAID setups. I haven't personally tested this on an EARS (I switched to buying Samsung drives when WD got restrictive with TLER) but I have disabled the head parking on an EADS and many posters to the TivoCommunity.com forums have done it with EARS drives.

There's a lot of good discussion of this issue on the Tivo hacking sites because the head parking interferes with soft reboots on Tivo boxes and must be disabled. In particular, read this FAQ entry as well as the last several pages of the very long thread it appears in:

http://www.tivocommunity.com/tivo-vb/showthread.php?t=370784#drey


Start with v1.05 of wdidle3 (that should be the version on WD's site) and use "wdidle3 /d" to disable followed by just "wdidle3" to verify success. If that doesn't work, there are at least two other versions floating around on the TivoCommunity.com forums (v1.03 and v1.00), as well as discussions about other command-line parameters to use if /d doesn't work (you can set the timeout to its maximum value instead of 8 seconds).

I currently have three WD drives in my PC. should I unplug the two drives that don't need this fix (a couple of old raptor drives) for safety or does the wdidle tool autodetect/let you choose which drive to run it on?

 

[FishAk]

I currently have three WD drives in my PC. should I unplug the two drives that don't need this fix (a couple of old raptor drives) for safety or does the wdidle tool autodetect/let you choose which drive to run it on?

Unplug them. It's quick, simple, and guarantees no problem.

 

[Campy]

Ok i found something really interesting:

http://community.wdc.com/t5/Desktop...-after-short-operation-time/td-p/15731/page/2

Sorry, but our agent didn't know that this policy was just changed. Current WDIDLE3 works with the RE and GP drives listed below.

RE Drives - WD1000FYPS-01ZKB0, WD7500AYPS-01ZKB0, WD7501AYPS-01ZKB0

GP Drives - WD20EADS, WD20EARS, WD15EADS, WD15EARS, WD10EADS, WD10EARS, WD8000AARS, WD7500AADS, WD7500AARS, WD6400AADS, WD6400AARS, WD5000AADS, WD5000AARS

WDIDLE3
http://support.wdc.com/product/download.asp?groupid=609&sid=113&lang=en

So the new GP drives are officially supported. Good news!

 

[sneakrnet]

I've got a WD20EARS drive that I've been using for a little under a month, and it already has >18,000 load/unload cycles (using Windows 7 Pro 64-bit). I just used the utility (wdidle3) to change the "idle time before head park" from 8 seconds to 300 seconds.

Some info that people were asking about:

The utility scans for WD drives that it supports, and sets the idle time to the specified time for each supported drive that it finds; you cannot selectively change the idle time, except by disconnecting any drive that you don't want changed before you run the utility.

In my case, it changed the idle time on the following models:
WD20EARS, WD10EARS, WD4000YS (which, before changing it, had the idle time disabled; in my case, it doesn't matter, but I didn't expect that).

So the utility apparently works on many models besides those advertised, so be careful. For example, the WD4000YS is a model intended specifically for use in RAID arrays, and probably shouldn't have its idle time changed.

 

[taltamir]

My understanding is that you can disable the head parking even on the newer WD drives...what you cannot do is enable TLER which can be important for RAID setups.

TLER actually means "disable deep recovery".
Deep recover is a feature where the drive will take 2 minutes to attempt to recover a bad sector. TLER simply disables it and immediately gives up on it. TLER means that any write error will cause file corruption rather then being recovered.

The reason TLER is needed is because your standard shitty RAID controller will notice that the drive is taking 2 minutes... and conclude the drive must be broken, and drop it from the RAID array.

The solution is ofcourse to use a good RAID controller / software.
ZFS for example isn't bothered one bit by either situation.

 

[TJCS]

now i know why my WD20EARS drive seems to power down state so frequently. If one is not concerned with saving power, why not just disable the feature altogether? I see a lot of people setting it to 300 secs and wonder which to choose. Only WD green drives has intellipark, other WD drives seem to run fine without it. Any thoughts?

 

[hddtech]

reason for green series to power down frequently is energy saving. This series supposedly saves power while idle. I got one of these drive for my RAID 0, MDL: WD20EARS-00MVWB0 <spam removed?>I don't see any differences in performance comparing to my 2x1TB WD1001FALS RAID system.

 

[quakeguy]

Hi,

So I've been running 6xWD20EARS in RAID5 (mdraid) for two and a half years.
I was sadly unaware of the load cycle issue with these drives until today. I've just corrected this using idle3-tools (http://idle3-tools.sourceforge.net/).

All 6 drives are operating without issue. Here's some data from the SMART output.
Note the whopping 600,000 load cycles. I wonder when these will begin failing. I'll post when they do.

All drives are model WD20EARS-00S8B1 (4-platter) except /dev/sdd, which is WD20EARS-00MVWB0 (3-platter).

# smartctl -a /dev/sda |egrep "(Load_Cycle|Power_On)"
9 Power_On_Hours 0x0032 069 069 000 Old_age Always - 23213
193 Load_Cycle_Count 0x0032 001 001 000 Old_age Always - 600457
# smartctl -a /dev/sdb |egrep "(Load_Cycle|Power_On)"
9 Power_On_Hours 0x0032 069 069 000 Old_age Always - 23212
193 Load_Cycle_Count 0x0032 001 001 000 Old_age Always - 597664
# smartctl -a /dev/sdc |egrep "(Load_Cycle|Power_On)"
9 Power_On_Hours 0x0032 069 069 000 Old_age Always - 23212
193 Load_Cycle_Count 0x0032 001 001 000 Old_age Always - 601381
# smartctl -a /dev/sdd |egrep "(Load_Cycle|Power_On)"
9 Power_On_Hours 0x0032 067 067 000 Old_age Always - 24598
193 Load_Cycle_Count 0x0032 001 001 000 Old_age Always - 605695
# smartctl -a /dev/sde |egrep "(Load_Cycle|Power_On)"
9 Power_On_Hours 0x0032 069 069 000 Old_age Always - 23211
193 Load_Cycle_Count 0x0032 002 002 000 Old_age Always - 596397
# smartctl -a /dev/sdf |egrep "(Load_Cycle|Power_On)"
9 Power_On_Hours 0x0032 069 069 000 Old_age Always - 23211
193 Load_Cycle_Count 0x0032 002 002 000 Old_age Always - 596772

 

[0roo0roo]

reason for green series to power down frequently is energy saving. This series supposedly saves power while idle. I got one of these drive for my RAID 0, MDL: WD20EARS-00MVWB0 I don't see any differences in performance comparing to my 2x1TB WD1001FALS RAID system.

Yea I idled my drives long ago.

Its greenwash if it murders the drive.