WD10EACS: dramatic high load cycle counts

[tw_switzerland]

Hello!

I have a Synology NAS Disk Station DS-209+ with 2 built in and 2 external
SATA harddisks of type Western Digital WD10EACS Green Power.

I chose those disks based on the compatibility list on the web page of Synology.

I realized that the disks never hibernate. Investigations showed, that there
seems to be a problem concerning power management and hibernation.

Even worse: the disks seem to be stressed abnormally by loading/unloading the
heads 112 times per hour. The new disks reached 31'000 load cylces in only
281 hours of operation. This would mean, the specified 300'000 load cycles of
the disks would be reached within a few months. This does not sound very good
to me and could end in a disaster...

If you google the web, you find many discussions about the problem. But I think there must
be a real fix - no ugly workarounds.

My questions:
- Does anynody know the problem of the load cycles?
- Can the problem be solved by setting the jumper 3-4 (enable Power Management 2) on the disks?
- Does anybody know why the disks never hibernate?
- Is there a real solution?
- Is there a firmware for the disks or the Synology fixing the problem?

NAS from Synology:
DS-209+
Firmware: DSM 2.0-0728

Disks from Western Digital:
3 x WD10EACS-00C7B0 Firmware 01.01B01
1 x WD10EACS-00D6B1 Firmware 01.01A01

S.M.A.R.T. Info (excerpt):
Start_Stop_Count 8
Load_Cycle_Count 31'259
Power_On_Hours 281


I will stop the operation of the NAS until the problem is resolved.

Thank to anybody giving some hints.

Thomas

[Volpone2]

Hello,
a) You should search this forum about about WD Green / Load Cycle Count / WD10EACS / drive compatibity ... there is a lot of contributions about this issue !
b) Contact synology and / or forum administrator about the problem, Synology position on this issue is not acceptable nor professional ...
c) it seems new versions of WD Green Power firmware fix the problem (but only on the new drives).

Regards

[tw_switzerland]

Thank you for your answer Volpone2.

I did read all information I could find. But I'm simply not satisifed by the speculation and ugly workarounds I found.

There must be a clear statement on the problem coming from Synology and from Western Digital. And they should address the problem by new firmwares for both the NAS and the disk.

I contacted the support of Synology and Western Digital. Still no answer on this load cycle problem.
[Edit 2008-11-27: The Synology support has answered and is investigating the problem in detail. Franklin from Synology support started an experiment. If you are affected by the load cycle problem, you should post your numbers in his Experiment thread]

I am optimistic - the solution will come soon.

Thomas

[Volpone2]

Thank you for your answer Volpone2.

I did read all information I could find. But I'm simply not satisifed by the speculation and ugly workarounds I found.

I agree !

There must be a clear statement on the problem coming from Synology and from Western Digital. And they should address the problem by new firmwares for both the NAS and the disk.
I contacted the support of Synology and Western Digital. Still no answer on this load cycle problem.

I obviously agree too, but unfortunatly:
- Synology still claim "WD GP are fully compatible" and they didn't communicate on the issue (for commercial reasons i think) !
- Western Digital fix is wdidle3.exe utility ... and the new disks firmware

I am optimistic - the solution will come soon.

I was optimistic and constructive too, but i've lost too much time with this issue.
I've now replaced my WD GP by Hitachi drives !
Regards

[tw_switzerland]

Volpone2: My local supplier (where I bought the NAS and the disks) told me that I should set the jumper 3-4 on the disks. That would enable "Power Management 2" features and solve the high load cylce count. Do you have any information on that?

Thomas

[tw_switzerland]

The answer from Western Digital Support has just arrived:

[...]
The firmware in our devices are only for win and mac. Obviously technical specifications about our products are related only to this 2 OS.
Western Digital technical support only provides jumper configuration and physical installation support for hard drives used in systems running the Linux/Unix operating systems. For setup questions beyond physical installation of your Western Digital hard drive, you would have to contact the vendor of your Linux/Unix operating system.
[..]

According to this information one can no longer recommend disks from Western Digital for use in other OS than Windows and Mac OS. I regret having bought 4 disks from WD for the NAS when these disks are not supported.

I am getting frustrated now as I really hoped WD would have at least a firmware fixing the problem.

[fabitux]

Hi everybody !
Sorry for my english (I'm french and not good at english speaking)
I post in "experiment" topic too.

I have tried all solutions :
1- WDIDLE3.EXE solution with -S 255 option
No effect on LCC count : between 60 to 110 LCC/hour

2- Noatime solution
http://www.synology.com/wiki/index.php/Spindown_issues
I add :

Code: Select all

mount -o remount,noatime /
mount -o remount,noatime /volume1
mount -o remount,noatime /volume2

on /etc/rc.local
No effect : 70 to 100 LCC/h

3- Centisecs solution
http://www.synology.com/enu/forum/viewtopic.php?f=120&t=10504&start=75)
add this line in /etc/rc.local

Code: Select all

echo 2200 > /proc/sys/vm/dirty_expire_centisecs

No effect : sometimes 40 LCC/h, sometimes more...

4- Listening music
When I'm reading some music on first drive and some video on the second drive, it seems to reduce the LCC/h.
But it isn't a solution because of, sometimes, it happens that I close my pc lol

5- Actual solution : downloading with bittorrent
I have a permanently download in bitorrent : zero LCC/h on each disk !!
Bitorrent files are saved on an only disk, but LCC is the same on each disk : zero !!
At this time, my /etc/rc.local is this one :

Code: Select all

#!/bin/sh
#exit
#####solution noatime
mount -o remount,noatime /
mount -o remount,noatime /volume1
mount -o remount,noatime /volume2
#optionally disable the swap file
swapoff /dev/md1

#####solution cinhil
#/sbin/sysctl -w vm.dirty_expire_centisecs=2000

#####hack centisecs
/opt/bin/syncer &
#sets flush time to 22 secs just under WS 25.5 unload max
echo 2200 > /proc/sys/vm/dirty_expire_centisecs

# Optware setup
mkdir -p /opt
[ -e /volume1/opt ] && mount -o bind /volume1/opt /opt
[ -x /opt/etc/rc.optware ] && /opt/etc/rc.optware
[ -e /bin/killall ] || ln -s /bin/busybox /bin/killall

I think it is not a good thing that there are no (zero !!) LCC.... but ????? Perhaps it is more bad for the disk ???
To your mind, downloading permanently without any LCC is it good ?

What about hibernation ?
I don't know...
One time, my syno hibernate on the night. I'm sure because I have "ear" (listened to), and the prompt whith ssh root@192.XXXX takes 5 or 10 seconds before done.
I'm really poor of idea about hibernation :
- Is my syno hibernnates sometimes ?
- Should my syno hibernate ? (Is it a good thing if he hibernate ? If he doesn't ?)

For conclusion : I was enjoy to buy Synology and Green/Western Digital because of environnement preoccupation, but I'm very deception now...

Of course, any help / comment would be appreciated
à+ / fabitux

[tw_switzerland]

Hello fabitux!

Thank you for your information. Looks like a nice summary of the workarounds I saw.

Head unload/load occurs when disk is switching its state between idle and active. So when you have permanent access to a disk, then LCC should not be incremented at all (as no idle state is detected by the disk).
When Linux systems are in idle state, there must something interfere with the Green Power management of the disk so as the disk seems to be stressed by switching permanently between idle and active.
Instead of a "green behavior" the disk consumes even more power than normally - as much as the disk would run seek operations.

So it is a normal and good thing that LCC is not increasing under load.

Thomas

[dizzzz]

I checked the SMART status of the two drives in my DS209+ (DSM 2.0-0728) and found the same behaviour; High LLC numbers [and increasing] while the box is installed for just a few days now, ouch. Well I guess it could be fixed with a firmware update, in the mean time couldn't we use a hdparms-like application to set the drive settings?

[tw_switzerland]

Hibernation:

I read in the german forum that sometimes you have to set hibernation time to a new value in order for the function become active. So i switched hibernation time from 20 minutes to 10 minutes. And see: the log now showed that the disks went into hibernation and waked up again.

Unfortunately this is what is standing in the log only. The disk itself seems to be rotating as before! So the hibernated WD10EACS seem to be spinning...?

[tw_switzerland]

@dizzzz: Please add your data to the Experiment Thread by Franklin.

I'm not shure if this problem can be fixed so easily if we even not knwo exactly why, when and in what configurations it happens...

[Volpone2]

Hello fabitux!

....
So it is a normal and good thing that LCC is not increasing under load.

Thomas

Perfect analysis "tw_switzerland" ...

Thank you (and Fabitux) to summarize the facts with detailed informations. I hope some reactions (information, workaround, fix) from Synology !
High load LCC issue is, IMHO, not a specific Synology issue (see Google about Linux / WD GP) but Synology fault is not to consider the problem and lack of communication.

This is a very good thing Frankin decide to "experiment" ... LCC issue is "opened" since march 2008, i hope time are changing ...

Regards

[tw_switzerland]

Hello

I have two different versions of the WD10EACS: three 00C7B0 and one newer 00D6B1.
This morning I have tested both versions mounted internally and in an external eSata case.

What I can say now is, that how the disks are attached to the SAN does not change anything:
-> The idle WD10EACS-00C7B0 increments the LCC +2 every minute.
-> The idle WD10EACS-00D6B1 never increments the LCC. In fact its LCC is identical to the start stop counter.

The following questions are in the room now:
Is the newer WD10EACS-00D6B1 version improved and the LCC problem is really absent?
Or does the new firmware simply show another value (equals to start stop counter) as load cycle counts?
Or is the intellipark feature of the newer disk disabled by default?

Thomas

[nboss]

I've 2 internal WD10EACS in a DS209+ fw728 and no problem with LCC. See values below:

Disk 1:
Laufwerkmodell:
WD10EACS-00D6B0
Seriennummer:
WD-WCAU43852386
Firmware-Version:
01.01A01
Gesamtgröße:
931.51 GB
Power On Hours: 75
LCC: 11

Disk 2:
Laufwerkmodell:
WD10EACS-00D6B0
Seriennummer:
WD-WCAU43370750
Firmware-Version:
01.01A01
Gesamtgröße:
931.51 GB
Power On Hours: 75
LCC: 13

Hope i will help for further investigation.

Regards,
Nicolas

[nop]

http://www.onderka.com/2008/12/03/synol ... b-auf-1tb/

Still syncing, but after 2:45+ hours LCC is at 7 respectively 8 for my 2 WD10EACS with firmware 00D6B1. Wish me luck, I will report later.