HDD shutdown behaviour (Linux Lite vs MX)

Hardware
1

Hi all, :waving_hand:

I have a possibly very curious question regarding the actuator of my hard disk drive (HDD) in my dual boot setup (Linux Lite and MX Linux).

Is it possible at all that Linux Lite and MX treat my external HDD differently in any way when it comes to shutting down (power off) the system ?

I´m almost inclined to think so. :thinking:

Here´s some background information about my setup:

  • dual boot Linux Lite (older system) and MX Linux (freshly installed).

  • MBR setup.

  • Both are installed on an external HDD (Western Digital Technologies, Inc. Elements 25A2), connected to my PC Lenovo-H520e via USB 2.0, and I boot my OS´es from there. Both systems are using ext3.

  • desktop environment in both systems is Xfce and both use systemd.

And here´s some further information about my hardware setup:

  • my external HDD has a power supply of its own. I switch it on manually (by a power strip with on/off switch).

  • As soon as power is established the HDD starts spinning.

  • After that I power on the PC (Lenovo H520e) and then start my session.

  • Before shutting down the PC I set the HDD to:
    sudo hdparm -S 120 /dev/sdb

  • So although the PC is already in shutdown state the HDD keeps spinning for ten minutes. That´s how it´s meant to be.
    After that it stops spinning and I can power off the HDD´s power supply.

Well, my research led me to the following:

  • An actuator in a hard disk drive (HDD) is a mechanical component that moves the read/write heads across the spinning disks to access data.

  • A very quiet clicking sound from an external HDD when shutting down is typically normal, as it often indicates the hard drive’s head is parking.
    -This sound is likely the read-head parking—moving to a safe “landing zone”—as it shuts down

  • Single, Soft Sound: It happens exactly once, just as the computer shuts off.

  • Parked Mechanism: Modern drives unload the read-head from the platter onto a ramp, which can produce a single clicking or tapping noise.

So, after having gathered this information my question is:

Is it possible that the quiet clicking sound from the external HDD which I hear is slightly louder when shutting down MX than it is when shutting down Linux Lite :red_question_mark:

In fact it´s a rather gentle noise, not really hard clicking sound, more of a slight, gently whirring sound. Very short, just a fracture of a second.

I don´t mean the hard clicking sound it produces should the power suddenly go off unexpectedly.

The performance of the HDD is fine. Nothing to complain about.

Just to be clear: I can hear the sound in question on Linux Lite as well (as part of the shutting down process), but it definitively is louder in MX than in Linux Lite. Just a bit louder, but I can hear the difference.

So once again my question:

Is it possible at all that Linux Lite and MX treat my external HDD differently in any way when it comes to shutting down (power off) the system?

Thanks a lot in advance. :heart:

Many greetings from Rosika :slightly_smiling_face:

2

Hi Rosika,

I am at a lost to even try to answer this question. From what I have read on the web the parking the heads of a HDD should no longer be a concern. Even after a power lost, the HDD will park the heads before a complete shutdown of the disk.

But your procedure does not hurt a thing and may be a good path of caution.

I submitted your question to Co-pilot and the short answer was “yes” the two different Linux distros can shut down the external HDD differently. This “can produce a slightly different head‑parking sound.”

A more detail explanation from Co-pilot was;

"MX Linux and Linux Lite ship with different:

udev rules

udisks2 settings

power‑management defaults

USB autosuspend behavior

These can affect:

when the drive receives a STOP UNIT command

whether the drive is told to flush cache before stopping

whether the USB bridge is powered down earlier or later

A slightly different timing can change the sound of the head parking."

It went on to said maybe the two different kernels could be the cause or even MX may apply it’s own power management after the hdparm command was issued.

I can message you a link of all the output I got if you like.

Hope this helps.
Howard

3

Hi Rosika,
I think Howard has answered your question .
It may be interesting next time you use Clonezilla, to listen when it shuts down. Clonezilla live is Debian. Clonezilla usually does everything comprehensively.

Regards
Neville

4

Hi @easyt50 and @nevj :

@easyt50 :

Thank you so much for digging up that valuable information, Howard. :heart:

That is the most important aspect, of course.

I think that this may very well be case when comparing the two distros.
That must be it.
Plus the fact that the kernels differ:

  • in MX: 6.12.74-2
  • Linux Lite: 5.15.0.173.161

Well, all the other points you listed may add to the differences in the whirring sound.
All of that adds up to a great explanation.
I can rest more assured now, as everything seems to be fine.

Thanks a lot, Howard.

@nevj :

Good idea, Neville.
I will have to remember that as I usually invoke a reboot rather than a shotdown when clonezilla has finished.
And a reboot doesn´t produce the head parking sequence, so I don´t hear that whirring sound at all. Neither more nor less loudly.

Thanks so much to both of you. :heart:
Many greetings from Rosika :slightly_smiling_face:

5

Additional information:

Unrelated to the topic, but not totally:

Setting the value of my HDD to:

sudo hdparm -B 254

is part of my daily routine. It prevents unnecessary head-parking actions of my HDD. I normally issue the command first thing after login.

On Linux Lite I use a script which takes care of that (hdparm_script.sh).
I had to re-write it for MX, as MX (Debian?) seems to get a different serial number of the disk.

It´s only today that I adapted the script to this effect, so until yesterday I used to issue the command manually in MX.

But due to yesterday´s excitement I must have forgotten to do so and as a consequence the Load_Cycle value climbed by a value of 132 during the afternoon. :neutral_face:

Now that my adapted script is in place everything seems to be back to normal again: The value usually climbs just by 3 per session. :wink:

Cheers from Rosika :slightly_smiling_face:

6

Do you mean a different UUID? No that is for partitions, not disks. You must mean a Volume Number?

7

@nevj :

Hi Neville, :waving_hand:

No, actually it´s the serial number my script uses.

This is part of my script for MX:

serial=WD-WXL1A56HU7XU
echo $(lsblk -o name,serial | grep $serial | awk '{print $1}')
[...]

I got the value from the command:

lsblk -o NAME,SERIAL
NAME   SERIAL
[...]
sdb    WD-WXL1A56HU7XU
├─sdb1 
├─sdb2 
├─sdb3 
├─sdb4 
├─sdb5 
├─sdb6 
├─sdb7 
└─sdb8 
[...]

The same part of the script for Linux Lite is:

serial=57584C314135364855375855
echo $(lsblk -o name,serial | grep $serial | awk '{print $1}')
[...]

Curious thing. I don´t have an explanation for that. :thinking:

Cheers from Rosika :slightly_smiling_face:

8

Hi Rosika,
That is handy to know.
I have 3 disks and have trouble identifying them in Clonezilla.
You use scripts more than I do.

Regards
Neville

9

Hi Rosika,

I found your post on head parking interesting, so I decided to perform a little bit of reading / research on the subject of parking heads on a HDD.

What I found out was ;

  • the park / unpark mechanical action is usually rated between 300,000 to 600,000 cycles.
  • if a HDD parks every 10 seconds it could hit the limit of cycles in 1–2 years
  • if a HDD spins continuously but parks the heads rarely it could last 10 years or more.

I thought this was very interesting, so thanks for your post on parking the heads of a HDD. I believe the above statistics assume the HDD is on all the time. But still, if you have a HDD connected for long periods, it looks like parking of the heads less often cuts down on the wear / tear of a HDD.

Kind Regards.
Howard

Source of the stats: the web & Co-pilot
Update
The info above was for HDD connected by USB.

10

Hi again, :waving_hand:

@nevj :

I´m glad you find that information interesting.

BTW:

I was doing some research on that and found out that this may be confusing indeed.

Hard disks actually have a tightly “burnt-in” serial number that should not change.
If Linux distributions display different values, it is usually because of how the kernel or tools read the hardware information.

One explanation might be:

  • Hardware serial number: The real number from the manufacturer (WWN or SN).

  • File system UUID: An ID generated by the operating system for the partition.

  • The difference: lsblk tries to read the hardware SN, but for some drivers, an identifier of the controller or a “Virtual ID” is displayed instead.

  • A newer kernel may read the serial number directly via SATA/NVMe protocol, while an older kernel displays an emulated ID via the USB bridge chip (if external).

(that´s from Google´s AI)

There may be other explanations but the ones above sound most likely.

@easyt50 :

thanks a lot, Howard.

Yes, the numbers you fond out sound plausible.
Unnecessary head parking should be avoided, of course. Not every HDD is prone to that. My WD HDD however has to be taken care of in that respect.

Years ago when I first started looking at my HDD´s smart data I realized that the Load_Cycle_count value was pretty high.

Then at one time in the past I read about the so-called “Notebook-Festplatten-Bug” on:

They said:

As became known shortly after the release of Ubuntu 7.10, Ubuntu and other Linux distributions let the writing heads of certain models of notebook hard drives change unusually often to a park position.

I hadn´t known of this issue before and was just using my Linux distro (Lubuntu) the normal way. That´s why the Load_Cycle_Count was pretty high.

After that I´ve been treating my HDD this way:

sudo hdparm -B 254 /dev/sda.

It´s the first thing I do after logging in.
As an effect the Load_Cycle_Count only climbs by a step of 3 or sometimes 4 for the duration of a whole session, no matter how long the session is.

Thanks a lot. :heart:

Oh, and as far as the original topic is concerned I also learned this:

The sound difference I noticed doesn’t equal more stress.

That little click (or rather whirring sound) which I hear is the actuator doing one very specific job: parking the read/write heads safely.

In other words: This is a designed mechanical action. The order of the actions carried out seem to be:

  • The drive moves the heads off the platter
  • places them on a ramp
  • locks them in place

The most important part is:

The force and movement of that action are controlled entirely by the drive’s internal firmware and not by Linux.
As a consequence the sound difference doesn´t matter at all. :wink:

Many greetings from Rosika :slightly_smiling_face: