New Toshiba HDD hdparm problems

[Rosika (Rosika Schreck)]

Hi again,

Taking up @kovacslt ´s recommendation to look here I gave it a shot:

while true; sleep 3; /usr/sbin/smartctl -g apm /dev/sda > /dev/null; end

(fish syntax)

While it seemed to have worked well with the user there it didn´t with me.
The Load_Cycle_Count kept growing nevertheless.

But the prevent_head_parking.sh script, especially when modified as suggested by László (see post #15), seems to work flawlessly.
Even after running for more than one hour: no increase of the value in question.

Thanks a lot @kovacslt and all.

Many greetings from Rosika

[kovacslt (László Kovács)]

@Rosika

What do you think?

In the first place I would not worry about load/unload cycles. If it wants to grow, then do it. Especially if this is a rarely connected drive.
I don’t think keeping it awake this way would do any harm. After all drives are there to be used, and if it would be a system drive, the at least the syslog, but probably all othe log entries would keep it awake

Frequent spin down / spin up however wears out mechanics sooner than it would die only spinning continously.

When you see load/unload increasing, do you hear spin-down / spin-ups too?

[kovacslt (László Kovács)]

@Rosika

smartctl -g apm /dev/sda

…of course for you it is /dev/sdb right?

@Rosika

But the prevent_head_parking.sh script, especially when modified as suggested by László (see post #15), seems to work flawlessly.

Oh, yeaaaahhh!!!
Rock’n’roll! &

[Rosika (Rosika Schreck)]

@kovacslt :

@kovacslt

…of course for you it is /dev/sdb right?

Yes, you´re very attentive, László.

Yesterday it was sdb, today it´s sda. It varies from boot to boot.

@kovacslt

In the first place I would not worry about load/unload cycles. If it wants to grow, then do it. Especially if this is a rarely connected drive.

Right.
As it´s my backup drive it is connected once a month or so.
Yet I dread to think of it as being my daily drive, using it for my OS.

It´s a good thing then that I made this experience right now.

This leads me to the conclusion: For running any Linux OS I´d certainly prefer a WD HDD, which seems to respect hdparm settings…
… or use an SSD to that end.

@kovacslt

I don’t think keeping it awake this way would do any harm

Thanks for your evaluation, László.

@kovacslt

After all drives are there to be used, and if it would be a system drive, the at least the syslog, but probably all othe log entries would keep it awake

Yes, that´s true, of course.

@kovacslt

When you see load/unload increasing, do you hear spin-down / spin-ups too?

No, can´t say that I do.

It´s only when I set the spindown time to 10 secs (see m,y post #5) when I actually heard it spinning down.
But the value still kept growing.

Thanks a lot for your help and many greetings from Rosika

[easyt50 (Howard)]

@Rosika

Writing a small amount of data to the disk at regular intervals should cause significantly less wear on the drive compared to the frequent head parking and unparking.

Reading more about disk management, I read keeping the disk active will cause the drive temperature to increase. Increase heat can also cause ware on a HDD. Maybe less wear then the constant parking / nu-parking. I also read most HDD are rated at 600,000 Load_Cycle_Count.

[Rosika (Rosika Schreck)]

Hi Howard,

thanks for reading up on the subject.

Yes, I think the information you provided is correct indeed.

There´s also a quite extensive article on this subject here.
As it´s part of ubuntuusers wiki the article is in German but can be read in English via e.g. TranslateLocally for Firefox" add-on.

Some key statements are:

• As became known shortly after the release of Ubuntu 7.10, Ubuntu and other Linux distributions unusually often allow the writing heads of certain models of notebook hard drives to a parking position, which increases the wear of the hard disk and thus reduces the service life. Depending on the hard drive model, Windows users may also be affected.

• Current notebooks park the writing heads of their hard drives if there is no read/write access and thus save power for battery operation. However, the parking processes are wear-resistant and therefore limited in number. The maximum number varies from manufacturer to manufacturer and usually lies 300,000 to 600,000 parking operations.
  this corresponds to what @easyt50 found out.

• However, it is possible to correct the APM level settings using the operating system. It is not easily possible to find values that are suitable for each hard disk. Nevertheless, Ubuntu now sets reasonable standard values, which in many cases already avoid increased wear

• example:
  sudo smartctl -A /dev/sdX | grep -E "(Load_Cycle_Count|ID)"
  This command produces at properly specified drive (letter X, first hard drive is sda) and activated SMART function an output of the following form:

ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
193 Load_Cycle_Count        0x0032   95    95    000    Old_age   Always       -       32578

• The best overview of the number of “consumed” parking processes is available here VALUE. The value below is usually equal to 100 for a new hard drive and then approaches the critical value 0 during the lifetime (THRESH). In this example, the hard drive has already carried out 5% of its maximum intended parking operations.

• n the column RAW_VALUEis the actual number of parking operations. In network operation, the value issued here may only increase very slowly. For example, 100 load cycles per day certainly do not restrict the lifetime of a hard drive. However, a significantly faster change can be determined, which can also be determined with falling VALUEvalue is associated, appropriate measures should be taken.

Many greetings from Rosika

[Rosika (Rosika Schreck)]

Hi all,

just a side question regarding this Toshiba HDD.

Here is a pdf file showing its manual in different languages.

As far as “Dismount and disconnect the drive” is concerned it says:

User’s Manual - CANVIO PARTNER EN_5

Dismount and disconnect the drive:
It is strongly recommended not to unplug the drive by just unplugging the cable connection as it may be that data is
written to the drive at that time. Such data will be lost and the drive may be damaged physically!

I understand the data loss part but what about “may be damaged physically!”

Does anyone have further knowledge about it?

Many thanks and many greetings from Rosika

[easyt50 (Howard)]

@Rosika

I understand the data loss part but what about “may be damaged physically!”

This warning pertains mostly to when the disk is reading or writing to the physical disk platters. Of course, we usually un-mount the HDD before unplugging it from the USB connector.

" A sudden loss or spike of power can short out your hard drive’s control board. It could cause your hard drive’s read/write heads to crash (into the platter), its motor to seize, or its platters to become damaged."

I believe most new HDD automatically retract the heads with power loss.

Kind Regards,
Howard

[Tech_JA (Jorge Augusto)]

Hi Rosika,
I’m sorry, but I’m totally lost on this topic, so I’m asking:

• Do you want to avoid parking heads or do you want to do parking heads whenever you’re not using the disk?

Jorge

[Rosika (Rosika Schreck)]

Hi all,

thanks for your replies.

@easyt50 :

Thanks, Howard, for enlightening me.

@easyt50

" A sudden loss or spike of power can short out your hard drive’s control board. It could cause your hard drive’s read/write heads to crash (into the platter), its motor to seize, or its platters to become damaged."

Oh dear, that doesn´t sound good.

@easyt50

I believe most new HDD automatically retract the heads with power loss.

Yes, I think I read that somewhere too.

Suppose there´s a sudden power outage…

But why would the Toshiba people have felt the need to include the disclaimer then?
Perhaps just to be on the safe side, legally, I guess

Thanks a lot, Howard.

@Tech_JA :

Don´t be sorry, Jorge. Feel free to ask anything you like.

Actually I want to avoid the parking mechanism from kicking in, so that the Load_Cycle_Count stops increasing.

The write script which @kovacslt perfected in post #15 seems to be the best solution to the problem.

Cheers from Rosika

[Rosika (Rosika Schreck)]

@ all:

Hi again,

just want to add some background information to my latest question:

During the tests regarding the Load_Cycle_Count I had to unplug and plug in the HDD more than just once.
I got tired of doing it manually, so I looked up the Toshiba HDD with lssub -t and found out the correct port of it: It´s 2-1.2.

Instead of unplugging it I could then use the command:

echo '2-1.2' | sudo tee /sys/bus/usb/drivers/usb/unbind

and instead of plugging it in I used:

echo '2-1.2' | sudo tee /sys/bus/usb/drivers/usb/bind

Works perfectly and has the same effect.

But before unplugging the HDD this way it´s certainly best to unmount it first.

Well, I cannot for the life of me remember whether I actually unmounted it before using the unbind command. I very much fear that I forgot to do that.

Dear me.

That disclaimer about potential physical damage got me worried then.
The only thing I do know for certain is that I didn´t read anything from the HDD, nor did I write to it while doing
echo '2-1.2' | sudo tee /sys/bus/usb/drivers/usb/unbind .

I very much hope that nothing serious had happened to my new disk.

@easyt50

It could cause your hard drive’s read/write heads to crash (into the platter), its motor to seize, or its platters to become damaged."

Well, it doen´t make any funny noises and using it for a while it seems to work perfectly.

The only other thing I could do for verification was taking a look at smatctl:

Any thoughts

Thanks a lot and many greetings from Rosika

[kovacslt (László Kovács)]

@Rosika

Instead of unplugging it I could then use the command:

echo '2-1.2' | sudo tee /sys/bus/usb/drivers/usb/unbind

and instead of plugging it in I used:

echo '2-1.2' | sudo tee /sys/bus/usb/drivers/usb/bind

.
That’s great!!!

@Rosika

I very much hope that nothing serious had happened to my new disk.

I really doubt that it can really physically damage the drive. When unplug/detach without unmounting, the buffers are not guaranteed to be flushed, so data loss is or file system corruption is possible.
Run fsck on it, if you are in doubt.

[Rosika (Rosika Schreck)]

Thank you very much, László, for your reply .

@kovacslt

That’s great!!!

Thanks for the praise.

I actually modified the command

echo '1-1.2.4.1.1' | sudo tee /sys/bus/usb/drivers/usb/unbind; and sleep 4; and echo '1-1.2.4.1.1' | sudo tee /sys/bus/usb/drivers/usb/bind

for dealing with the HDD.

I need to use the command above once in a while (pretty seldom, actually) when for some reason my keyboard gets a tad bit laggy.
After applying the command it works flawlessly again.

@kovacslt

I really doubt that it can really physically damage the drive.

That´s quite some relief to me. Thanks a lot for your evaluation, László.

@kovacslt

Run fsck on it, if you are in doubt.

I tucked the HDD away yesterday but I might do in the near future. Thanks for the tip.

Many greetings from Rosika

[easyt50 (Howard)]

@Rosika

Perhaps just to be on the safe side, legally, I guess

Hi Rosika,

That is exactly what I believe. And don’t worry about your HDD. Over the past 25 year, I must have not have dis-mounted a HDD correctly well over a 100 times due to I forgot to un-mount it,
power failure, PC lock up and PC shut down.
To my knowledge, the HDD was never damaged.

[Rosika (Rosika Schreck)]

Thanks, Howard, for sharing your experience.

That´s good to know.
Nothing beats exchanging personal knowledge.

Many greetings from Rosika

[Tech_JA (Jorge Augusto)]

Hi Rosika,

@Rosika

Actually I want to avoid the parking mechanism from kicking in, so that the Load_Cycle_Count stops increasing.

I’m sorry again, but I don’t understand why you want to avoid the parking system.
Many years ago, and older users may remember, before turning off the PC, we had to park it manually, there was a DOS command to park it.

Nowadays, fortunately, HDDs already do auto park and, even when they’re running, if the disk isn’t being accessed, it’s better to park the heads than to leave the heads ad aeternum on top of the HDD platters (cylinders).

If this is what you’re discussing, because I’m having some difficulty translating the topic, imagine the example of a turntable: we have the turntable spinning all the time and we have the needle constantly on top of the record, if after a while we want to take a break from playing the record, we lift the arm of the turntable (we do a kind of park). We want to hear the music again, we lower the turntable arm again and the whole needle is on the record.
That’s how I think about having the heads constantly on the platters (cylinders) of the HDD.

This is a very weak example, but the longer the needle is on the disk, the greater the chance of scratching it if, for example, the HDD is hit.
If the HDD has the option to park the heads after some idle time, I wouldn’t change these settings.

I’m really sorry if I’ve misinterpreted the topic.

Jorge

[Rosika (Rosika Schreck)]

Hi Jorge,

theoertically you´re right, of course.

But please take a look at my post #26.
Main points:

@Rosika

Ubuntu and other Linux distributions unusually often allow the writing heads of certain models of notebook hard drives to a parking position, which increases the wear of the hard disk and thus reduces the service life.

(bold by me)

@Rosika

[…]the parking processes are […] limited in number. The maximum number varies from manufacturer to manufacturer and usually lies 300,000 to 600,000 parking operations.

I guess what it boils down to is:

If there are no write or read operations, why should the Load_Cycle_Count unnecessarily keep growing?

Thanks for your opinion, Jorge.

Many greetings from Rosika

[Tech_JA (Jorge Augusto)]

My friend Rosika,

@Rosika

Ubuntu and other Linux distributions unusually often allow the writing heads of certain models of notebook hard drives to a parking position

If the disk is idle, i.e. not reading or writing, this is fine, right?

@Rosika

which increases the wear of the hard disk and thus reduces the service life.

The HDD is made up of many components, both mechanical and electronic. This is one of the points that can cause the disk to wear out and, in turn, reduce its lifespan. Of course, there’s something wrong if you do it at an exaggerated pace.

---

@Rosika

If there are no write or read operations, why should the Load_Cycle_Count unnecessarily keep growing?

---

Changing the subject, in fstab, do you have this disk with the noatime option?
If not, try it, it may help to reduce disk activity and, in turn, the parking of the heads.

I can’t answer that, but no one better than the brand’s support team can - it could be a configuration of the brand itself, or a firmware bug.
For example, have you seen if there’s new firmware for that disk (I couldn’t find the model Toshiba Canvio Partner 1 TB)?

I found this info:
Problem with hard drive clicking

I hope you get answers to your questions, my friend.

Jorge

[Rosika (Rosika Schreck)]

Dear Jorge,

thank you so much for your extensive reply.

@Tech_JA

If the disk is idle, i.e. not reading or writing, this is fine, right?

Theoretically yes.
The continuous rising of the Load_Cycle_Count seems to suggest that for some reason the idle state doesn´t prevail for too long.

@Tech_JA

in fstab, do you have this disk with the noatime option?

No, actually I haven´t included the Toshiba HDD in fstab:

cat /etc/fstab
# <file system> <mount point> <type> <options> <dump> <pass>

UUID=411331c4-730c-4a53-963b-e58f5733e688	/	ext3	errors=remount-ro	0	1
UUID=FC63-685C	/boot/efi	vfat	umask=0077	0	1
UUID=7f7b5dd9-60ef-4dd1-98dd-f585c921f2cf	/home	ext3	defaults	0	2

As you can see I didn´t bother to put it there. It´s ext4-formatted.
I use it only for backup purposes (clonezilla and grsync), so I attach it roughly once a month for just a limited amount of time.

@Tech_JA

I can’t answer that, but no one better than the brand’s support team can

Oh, I think there´s some room for doubt here.
My experience is: as soon as you say you´re using a Linux distro the so-called support refuses to listen to you.

The Toshiba pdf only refers to WIN and MAC machines. Linux isn´t even mentioned.

@Tech_JA

have you seen if there’s new firmware for that disk

No idea, Jorge. I wouldn´t even know where to look for it.
But I wouldn´t want to “mess around” with a brand new disk this way. So I guess it´s no option for me.

Thanks a lot for providing this interesting link, Jorge.

I read it through and it seems the author tested quite a lot of different makes and models.

These seem to be be the key points:

• The click sound is usually caused by the drive parking its head

• The clicks seem to happen when the drive is idle and the power has been on for a significant period of time.

• Western Digital drives […] are also known to exhibit pronounced clicks.
  I can verify this with my main HDD, which is a WD one. Here, however, hdparm commands are respected, and the problem could be solved.

• The clicking sound appears to occur only when the drive is idle

I already tried all of these hdparm settings: 255, 254, 192. None of these had an impact on the Load_Cycle_Count.

@kovacslt ´s script still seems to be best solution. Running it during the time I need the Toshiba HDD to be connencted for doing my grsync backups really does handle the job quite efficiently. The Load_Cycle_Count didn´t grow even by the value of 1.

Thanks a lot for your kind help, dear Jorge.

Many greetings from Rosika