Spin down external hard drive [now obsolete]

[junior466]

Greetings!

I have a 1TB external drive plugged into my pi and tried installing hdparm to get it to spin down when not in use but I believe "write cache is not possible for the drive". I was following a guide here https://www.htpcguides.com/spin-down-an ... pberry-pi/

Any suggestions? I wasn't even aware that I had to worry about spinning down the drive but I came across that article recently and started to wonder.

[epoch1970]

Hdparm is slowly falling into obsolescence.
Of the modern USB drives I've used with a Pi, none were supported by hdparm.
But they would spin down by themselves when not in use. The best way IMHO to tell a modern drive it is no longer in use is to unmount it. It will spin down shortly after.
You can automate that with an automounter or scripts.

HTH

[junior466]

Subject: Spin down external hard drive

epoch1970 wrote:
Hdparm is slowly falling into obsolescence.
Of the modern USB drives I've used with a Pi, none were supported by hdparm.
But they would spin down by themselves when not in use. The best way IMHO to tell a modern drive it is no longer in use is to unmount it. It will spin down shortly after.
You can automate that with an automounter or scripts.

HTH


The problem is I have an IP camera recording to it over FTP whenever it detects motion so I can't unnmount it since I need it on standby.

Will I really shorten its life if I don't spin it down?

[InsulationTape]

IME keeping hard drives on gives them life, it is the power cycling that finishes them off.

I back up my "research" with the anecdote lots of people will have heard about hard drives, "it was OK last time I used it".

A spun-down hard drive might save battery life, if applicable (ie laptops, and not most Pis), but to keep a drive going for years I think keeping them powered-up and spinning is best. If an HDD is being run flat-out most of the time and is insulated then power control might extend its life, but just dealing with the heat is probably a better aim.

I had a USB HDD for a while and I couldn't get it to stay powered on. In the end I set up a cron'd script to just tickle the drive frequently enough that it wouldn't spin down. As it happened the drive did die anyway, some cheap Samsung 1.5T IIRC from a computer fair.

It might be worth looking at smartd and have the drive test itself periodically, and send a notification if the drive looks to be failing. I say this with an HDD here like this:

Code: Select all

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     0x000f   100   100   046    Pre-fail  Always       -       184533
  2 Throughput_Performance  0x0005   100   100   030    Pre-fail  Offline      -       10420380
  3 Spin_Up_Time            0x0003   100   100   025    Pre-fail  Always       -       1
  4 Start_Stop_Count        0x0032   098   098   000    Old_age   Always       -       6181
  5 Reallocated_Sector_Ct   0x0033   100   100   024    Pre-fail  Always       -       0 (2000 0)
  7 Seek_Error_Rate         0x000f   100   100   047    Pre-fail  Always       -       3381
  8 Seek_Time_Performance   0x0005   100   100   019    Pre-fail  Offline      -       0
  9 Power_On_Seconds        0x0032   001   001   000    Old_age   Always       -       89116h+20m+03s
 10 Spin_Retry_Count        0x0013   100   100   020    Pre-fail  Always       -       0
 12 Power_Cycle_Count       0x0032   100   100   000    Old_age   Always       -       4015
192 Power-Off_Retract_Count 0x0032   100   100   000    Old_age   Always       -       237
193 Load_Cycle_Count        0x0032   057   057   000    Old_age   Always       -       869830
194 Temperature_Celsius     0x0022   100   100   000    Old_age   Always       -       44 (Min/Max 6/56)
195 Hardware_ECC_Recovered  0x001a   100   100   000    Old_age   Always       -       455
196 Reallocated_Event_Count 0x0032   100   100   000    Old_age   Always       -       0 (0 6906)
197 Current_Pending_Sector  0x0012   100   100   000    Old_age   Always       -       0
198 Offline_Uncorrectable   0x0010   100   100   000    Old_age   Offline      -       0
199 UDMA_CRC_Error_Count    0x003e   200   200   000    Old_age   Always       -       0
200 Multi_Zone_Error_Rate   0x000f   100   100   060    Pre-fail  Always       -       9807
203 Run_Out_Cancel          0x0002   100   100   000    Old_age   Always       -       429595295995

I manually keep an eye on this 40gig drive as smartd would likely moan far too much. The HDD was originally in a Tosh laptop I bought in ~2006. It is now in a first gen Mac Mini, which is a firewall for a shoddy IoT camera, and some other general network tasks (rsyslog for the router, etc.). But the total power on time for the HDD is over 10 years! The test log has all "Completed without error".

I know some story on the internet does not prove much, but frankly that blog post is just one rung above a forum post I am shamelessly advocating sleep-depriving that drive like a tasteless Guantanamo joke.

These are the settings for the Mac's HDD:

Code: Select all

/dev/disk/by-id/ata-FUJITSU_MHV2040AH_NT43T6329P59 {
spindown_time = 0
acoustic_management = 254
apm = 255
}

[junior466]

That makes a lot of sense. I think I wil just leave it be and make sure I have a backup but you are right, keeping it spinning definetely makes more sense.

What would be the best way to test this drive for errors? Is it possible to get it done in the pi itself? I use OSX as my daily computer so I may also be able to plug it into that?

[startrek.steve]

I always thought you couldnt run SMART on a USB Attached drive? I could never get SMART to see it?

[rpdom]

I always thought you couldnt run SMART on a USB Attached drive? I could never get SMART to see it?

Sometimes you need to add extra options to support the controller. I've used SMART and hdparm on USB attached drives to control and check them. I used to have a USB HDD that was dedicated for backups and was spun up automatically once a week for that purpose.

[InsulationTape]

I always thought you couldnt run SMART on a USB Attached drive? I could never get SMART to see it?

Ah yes, I've seen that. IIRC some USB-SATA bridges simply don't support what's necessary. Maybe even some USB-PATA ones, too. I'm sure I've seen this problem right back when PATA was more common.

One of the bajillion projects revolving around my Pi is some CCTV hackery. I've got a webcam and video grade HDD in a USB enclosure (originally from a Sky+ box or similar, I think) plugged into my Pi. I'm eye-balling zoneminder....

I can confirm that smartctl can query the drive connected to my Pi, using one of these:

Code: Select all

Bus 001 Device 005: ID 1bcf:0c31 Sunplus Innovation Technology Inc. SPIF30x Serial-ATA bridge

The enclosure is a Maxam branded aluminium box, USB2. So, the archetype of generic.

hdparm isn't installed on my Pi so can't say right now if parameters will set with that
These two options didn't work (though I did not try every hdparm feature):
-Z Disable Seagate auto-powersaving mode
-i Display drive identification
I do have a Seagate drive beyond the USB bridge, so -Z is not outside the realm of possibility. It looks like with my USB-SATA bridge I can do things like set drive power otherwise, the performance test. I'm not sure I believe the -C option, "Check drive power mode status".

Code: Select all

root@sentry:~$ touch /mnt/vid/scratch/delme ; sync
root@sentry:~$ rm /mnt/vid/scratch/delme ; sync
root@sentry:~$ hdparm -C /dev/sda

/dev/sda:
 drive state is:  standby

Looks like hdparm's functionality could be hit-and-miss with my bridge.... which reflects what epoch1970 said.

[InsulationTape]

What would be the best way to test this drive for errors? Is it possible to get it done in the pi itself? I use OSX as my daily computer so I may also be able to plug it into that?

I'm pretty sure this is the right package name for the command line program.

Code: Select all

apt-get install smartmontools

This will show drive info, and how long tests should take.

Code: Select all

smartctl --all /dev/sda | less

Start one of the tests. The long test on this drive is 104 mins

Code: Select all

smartctl --test=long /dev/sda

Running this again will show the progress

Code: Select all

smartctl --all /dev/sda | less

Code: Select all

#1  Extended offline    Self-test routine in progress 90%       430         -

Once the test is complete the line will change to be something like

Code: Select all

# 2  Extended offline    Completed without error       00%       243         -

If your drive is currently "blank" or otherwise empty then you could also read and write to every sector (cue the program dd), and check dmesg afterwards for any drive-like errors. No timeouts, bus resets, etc. or it could be sign of problems.

[junior466]

Excellent! Thank you so much for your help!

[junior466]

So it appears that my drive doesn't support testing. Here's the output of sudo smartctl -a /dev/sda2

Code: Select all

=== START OF INFORMATION SECTION ===
Vendor:               SAMSUNG
Product:              HD103SI
Revision:             
User Capacity:        1,000,204,886,016 bytes [1.00 TB]
Logical block size:   512 bytes
Device type:          disk
Local Time is:        Tue Jul 18 20:22:01 2017 EDT
SMART support is:     Available - device has SMART capability.
SMART support is:     Enabled
Temperature Warning:  Disabled or Not Supported

=== START OF READ SMART DATA SECTION ===
SMART Health Status: OK

Error Counter logging not supported

Device does not support Self Test logging

and then:

sudo smartctl --test=long /dev/sda2

Code: Select all

smartctl 6.4 2014-10-07 r4002 [armv7l-linux-4.9.35-v7+] (local build)
Copyright (C) 2002-14, Bruce Allen, Christian Franke, www.smartmontools.org

Long (extended) offline self test failed [unsupported scsi opcode][/size]

So I guess I'm out of luck?

[InsulationTape]

I think that'll be your bridge not passing things through properly, rather than the drive itself. smartctl says SMART is supported by the drive, but then fails to get the SMART data. And then you can't tell it to run a test. IMHO the "unsupported scsi opcode" is due to what's controlling the drive rather than the drive itself.

Try another USB-SATA adapter if you have one. Or if not you could hook the drive up to a normal PC for a while to test it. Smartmontools is available for Windows, https://www.smartmontools.org/wiki/Down ... owspackage , but you could use a Linux live CD/DVD/USB and run smartctl in that. Debian's LXDE live CD is likely very similar to using a Pi, the default username and password for Debian live CDs is user and live. https://www.debian.org/CD/live/

Having a Google there is evidence that the drive's SMART does work, this bug report contains smartctl reading SMART data from a HD103SI https://sourceforge.net/p/smartmontools ... /26399651/ . This screenshot from a German forum looks exactly like SMART data https://www.computerbase.de/forum/attac ... 1266023150 .

But those Samsung green drives had a terrible reputation, or at least one manufacturer's green product line. They are so keen on powering down they heat/cool far too much and just kill themselves. IIRC trying to reign in their keenness to spin-down is either impossible or very tricky. By all means use the drive, but have good backups!

[DougieLawson]

Here's the output of sudo smartctl -a /dev/sda2

Try running sudo smartctl -a /dev/sda as that's more likely to work.