7

i just realized that some of my HDD's have a huge Load_Cycle_Count when reading out their S.M.A.R.T data. Some are short before failing and i am asking myself why this is the case and if there is anything i can do to prevent them from dying.

 alex@ga-P55A-UD5:~$ sudo smartctl -a /dev/sdb smartctl 6.5 2016-01-24 r4214 [x86_64-linux-4.4.0-142-generic] (local build) Copyright (C) 2002-16, Bruce Allen, Christian Franke, www.smartmontools.org === START OF INFORMATION SECTION === Model Family: Western Digital Caviar Green (AF) Device Model: WDC WD10EARS-00Y5B1 [...] Vendor Specific SMART Attributes with Thresholds: ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE 4 Start_Stop_Count 0x0032 090 090 000 Old_age Always - 10281 9 Power_On_Hours 0x0032 062 062 000 Old_age Always - 28456 193 Load_Cycle_Count 0x0032 001 001 000 Old_age Always - 611460 
 alex@ga-P55A-UD5:~$ sudo smartctl -a /dev/sdc smartctl 6.5 2016-01-24 r4214 [x86_64-linux-4.4.0-142-generic] (local build) Copyright (C) 2002-16, Bruce Allen, Christian Franke, www.smartmontools.org === START OF INFORMATION SECTION === Model Family: Western Digital Caviar Green Device Model: WDC WD6400AADS-00M2B0 [...] Vendor Specific SMART Attributes with Thresholds: ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE 4 Start_Stop_Count 0x0032 093 093 000 Old_age Always - 7615 9 Power_On_Hours 0x0032 057 057 000 Old_age Always - 31743 193 Load_Cycle_Count 0x0032 053 053 000 Old_age Always - 442121 
 alex@silent-ssd:~$ sudo smartctl -a /dev/sdd smartctl 6.5 2016-01-24 r4214 [x86_64-linux-4.4.0-142-generic] (local build) Copyright (C) 2002-16, Bruce Allen, Christian Franke, www.smartmontools.org === START OF INFORMATION SECTION === Model Family: Western Digital Green Device Model: WDC WD20EARX-00PASB0 [...] Vendor Specific SMART Attributes with Thresholds: ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE 1 Raw_Read_Error_Rate 0x002f 200 200 051 Pre-fail Always - 0 4 Start_Stop_Count 0x0032 098 098 000 Old_age Always - 2477 9 Power_On_Hours 0x0032 085 085 000 Old_age Always - 11176 193 Load_Cycle_Count 0x0032 181 181 000 Old_age Always - 59646
Improve this question

5 Answers 5

Reset to default
10

My findings so far:

The Cause

  • Regarding to Western Digital and various websites 1, 2, 3, 4, 5, 6 the high number in S.M.A.R.T Attribute 193 Load_Cycle_Count is related to a technique introduced by WesternDigital named Intellipark.
  • Intellipark is implemented in some of their hard drive models, especially in their green series.
  • It is designed to reduce power consumption if the drive is not beeing used.
  • In some usecases, especially when combined with a Linux operating system, this intellipark-feature tends to shorten hdd's live.

Solutions

  • Western Digital explains it's not their features fault, it's the bad configured operating system and they give some advices on how to properly configure linux.
  • Western Digital also released a DOS tool to modify the intellipark-feature on some devices.
  • For the Linux platform Christophe Bothamy released idle3-tools to modify that intellipark-feature - big thank you from my site.
  • as mentioned in the comments below, hdparm -J does either modify the wd idle3 timer but as another commentary pointed out: This implementation is not as thorough as the official WDIDLE3.EXE

What I've done

Now I downloaded idle3ctl and turned off intellipark completely. Hopefully this will help to prevent the disks from failing quick. But anyway at least one disk is almost dead, regarding to S.M.A.R.T.

To disable the intellipark-feature i followed the idle3-tools instructions.

First I read out the idle3 timer value of this intellipark feature: sudo ./idle3ctl -g /dev/sdx

Than i disabled the timer sudo ./idle3ctl -d /dev/sdx

A power off/on cycle is necessary to take effect sudo hdparm -Y /dev/sdx

After that i rechecked the idle3 time and did the same after a reboot:

 alex@silent-ssd:~/idle3tools/idle3-tools-0.9.1$ sudo ./idle3ctl -g /dev/sdd Idle3 timer is disabled

Edit 2023

I just wanted to give an update about the state of the drives after 4 years of usage. All drives are still alive so far but the two disk with the Load Cycle Counts beyond 500.000 are not in use anymore, or at least only used from time to time for unimportant stuff. The third drive is still in active use. It hold the btc blockchain and is almost running 24/7 since four years now. The number of Load Cycle Counts only increased by a few thousands. For spin down I am using hd-idle with a timer of 3 hours that often times does not take effect because of the almost permanently updating blockchain.

Here is the smart data of that drive:

 === START OF INFORMATION SECTION === Model Family: Western Digital Green Device Model: WDC WD20EARX-00PASB0 [...] SMART Attributes Data Structure revision number: 16 Vendor Specific SMART Attributes with Thresholds: ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE 1 Raw_Read_Error_Rate 0x002f 200 200 051 Pre-fail Always - 0 4 Start_Stop_Count 0x0032 098 098 000 Old_age Always - 2713 9 Power_On_Hours 0x0032 046 046 000 Old_age Always - 40109 193 Load_Cycle_Count 0x0032 180 180 000 Old_age Always - 61955
As a side note to the topic about the lifespan of a HDD:
Thanks to this posting ( https://superuser.com/questions/197862/how-harmful-is-a-hard-disk-spin-cycle ) I found a nice paper / study ( https://www.usenix.org/legacy/event/fast07/tech/full_papers/pinheiro/pinheiro_html/index.html ) about hard drives failures and their conclusion on how to interpret S.M.A.R.T values.
Improve this answer
3
  • 3
    See also hdparm’s -J option and the discussion in the manpage. Commented Mar 4, 2019 at 10:53
  • Nice answer. I can confirm everything here. I can suggest a further solution "Don't buy WD drives." This is what I decided and don't have any problems with head parking and clicking anymore Commented Jan 17, 2021 at 21:32
  • 1
    Even if not clearly mentioned in hdparm's manpage, trying calling it with -J results in "Use of -J is EXTREMELY DANGEROUS. This implementation is not as thorough as the official WDIDLE3.EXE. Use at your own risk! Please also supply the --please-destroy-my-drive flag if you really want this. Program aborted." So, better off using idle3-tools. Commented Jan 26, 2021 at 15:56
4

I have a 1TB WD Blue Mobile WD10SPZX with LCC rated endurance of 600k. It's was increasing fast too (about 7.7 cycles per hour):

coutinho@discovery:~$ sudo smartctl -a /dev/sdb smartctl 6.6 2017-11-05 r4594 [x86_64-linux-4.19.0-13-amd64] (local build) Copyright (C) 2002-17, Bruce Allen, Christian Franke, www.smartmontools.org === START OF INFORMATION SECTION === Device Model: WDC WD10SPZX-75Z10T1 [...] Vendor Specific SMART Attributes with Thresholds: ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE 4 Start_Stop_Count 0x0032 100 100 000 Old_age Always - 119 9 Power_On_Hours 0x0032 096 096 000 Old_age Always - 3040 193 Load_Cycle_Count 0x0032 193 193 000 Old_age Always - 23541

I tried both WD's WDIDLE3 DOS tool and idle3-tools, none worked with my disk.

Finally, I managed to reduce its LCC increase less than 1 per hour by setting APM value to 254 with hdparm adding the following lines to /etc/hdparm.conf:

/dev/sdb { apm = 254 spindown_time = 0 }
Improve this answer
6
  • Almost one month later, I observed my hd had a LCC of 23798 and 3436 power on hours. It increased 257 cycles in 396 hours, which gives 0.65 cycles/hour. Commented Jan 20, 2021 at 5:31
  • Oh my god... ANOTHER thing that I have to be worried about while using Linux... it's getting very tiresome at this point... the main reason for me to use Linux is the fact of it allowing me to have Live Sessions from memory, not ever touching my disk (only for storage that I explicitly choose) Commented Feb 14, 2022 at 1:27
  • So please, my friend, tell me: do you have any update on this matter after about a year? Any new findings? Is there hope? ... (from what I get reading your last comment there is that this solution detailed on your answer did not yield the expected results, correct?) Commented Feb 14, 2022 at 1:28
  • Just for reference, these are my readings on a laptop with WD Green after 1 year and 3 months of usage: _____ 9 Power_On_Hours=10627_xxx_4 Start_Stop_Count=2772_xxx_193 Load_Cycle_Count=9903______ What is the "Start Stop Count" ?? Commented Feb 14, 2022 at 1:31
  • After almost 2 years since the post, my readings are: Power_On_Hours = 12650 , Load_Cycle_Count = 28062 . It increased 4521 cycles in 9610 hours ( 0.47 clycles/hour). So setting APM to 254 reduced LCC incerease to a manageable level, it will take a century to LCC reach its limit. Commented Dec 6, 2022 at 2:49
2

WD Green disks are designed to park the heads relatively quickly after the disk becomes idle. As a result you'll get a high load count.

An old article, but still pretty accurate, https://www.pugetsystems.com/labs/articles/Western-Digital-Green-vs-Red-Hard-Drives-602/ writes

A load/unload cycle is when the disk spins up its platter to get ready for operation. Typically this happens when you turn on a system, resume from standby, or when the drive has been idle long enough for the OS to power down the drive. [..] the 300,000 cycles the Green drives are rated for is really quite a lot already. Even if you turn on/off your system or let it idle long enough to power down the drive 20 times a day, 365 days a year, a Green drive should still last for over 40 years.

Also

Head parking delay 8 seconds

In your case, your two disks have been powered on for 31743 hours and 11176 hours, respectively (approximately 3 years 6 months and 1 year 3 months, at 24x7). For that period of time the number of unload/load cycles really isn't unreasonable.

If you are running 24x7, mind, you should be using WD Reds instead of Greens.

Improve this answer
3
  • yes it might be true that the number should be higher after such a long period of time but compared to all the smart values this Load_Cycle_Count value is the worst i have and might be the cause why i have to replace my disks sooner than later. As you can recheck here the rest of the values are quite fine: pastebin.com/P47S9CfZ Commented Mar 4, 2019 at 11:13
  • Given their age those two disks look pretty good to me. No data errors reported. Commented Mar 4, 2019 at 11:47
  • Line 102 of that pastebin: Value 001 Worst 001 Thresh 000 - from my knowledge this is means it is almost dead. The other two hard drives are in a better shape, indeed. Commented Mar 4, 2019 at 12:20
1

Intellipark on newer WD drives cannot be switched off even with original WDIDLE3 utility.

As a workaround you can add to hourly cron this command

smartctl -t offline /dev/sdX

it prevents parks. Performance degradation is unnoticeable for me.

Improve this answer
9
  • I tried googling that but it didn't explain what it does in the context mentioned here. What is that command used for? What does it do? It runs a full set of disk tests? (I would be afraid that puts stress on the disk even more?) Commented Feb 14, 2022 at 1:41
  • Do you have updates on this? Do this measure of hourly cron job prevents the Load_Cycle_Count from increasing? Commented Feb 14, 2022 at 1:42
  • This drive died after couple of months (no kidding). But I'm not sure that it is real consequence. Maybe it triggered some bug in firmware, like variable/counter overflow. I've blacklisted WD for myself now... Commented Feb 14, 2022 at 8:24
  • My goodness... One of the main reasons for me to use Linux Live Sessions from USB primarily is exactly to spare random writes from the disk at all times, and now I find that Linux introduces new ways of damaging the disk aside from that. Very disappointing. I don't have spare income to keep buying stuff like that, are all the Kernel devs rich? (they are surely well paid, that's a fact) Commented Feb 16, 2022 at 15:36
  • But anyway, I have WD Green... This week, I used hdparm -B command to change my power saving scheme from 254 (no powersave, never sleep) to 128 (maximum powersave while not suspending disk) Two days later, I go check my SMART statistics again and I find in these 2 days the disk had additional 600 times of "Load Unload Cycles" (it jumped from 9.900 after 1 year, to about 10.500 in 2 days) I think it means the disk was parking heads more often, I think that's what Load Unload Cycle means... Reverted back to 254 now and will monitor it... 9.900 value means that disk parked head 25x / day? (scary) Commented Feb 16, 2022 at 15:41
0

Adding to the discussion: I supervise a Synology NAS which has 5 drives, and is populated with WD Red 3TB drives

  • Drives 1,2,3 and 5 have Power_On_Hours = 49760, +/- 2.
  • Drive 4 has Power_On_Hours = 26959, as it replaced a failed drive at some point.

The replacement was one of the batch of 8 drives purchased originally, 5 installed, and 3 saved as replacements. So all drives are from the same time point, though bought from two different vendors to avoid getting all from one batch.

  • Drives 1,3 and 5 have Load_Cycle_Count = 345 +/-2
  • Drive 2 Load_Cycle_Count = 656
  • Drive 4 (the one with only 27k hours) Load_Cycle_Count = 93144 !!

For what it's worth, I'm looking at the SMART data due to error warning emails about drive 2,

  • Drive 2: Raw_Read_Error_Rate = 4520.
  • Drives 1,3,5 Raw_Read_Error_Rate = 0
  • Drive 4 (again the 27khr unit) Raw_Read_Error_Rate = 10232 !!

So for some reason, the one drive that was installed later than the rest has an enormous Load_Cycle_Count and Raw_Read_Error_Rate.

Not sure what to make of all that.

Improve this answer
2
  • 2
    What exact hard drive models do you use? Have you tried googeling them, in case people encountered similar problems? Is this a raid drive? Are your HDDs CMR or SMR? Manufactures tend to change things silently over time and sometimes these changes do things become worse. I am sorry that I am of no help to your specific problem. Commented Jun 19, 2023 at 8:12
  • @AlexOnLinux Thanks for your comment -- I was not so much asking a question as simply documenting this oddly different behavior between seemingly same drives all in the same Synology box. I will at some point look at the detailed model Ids etc and see if there is some rhyme or reason to the difference. Commented Jun 26, 2023 at 22:23

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.