How often and which self test should I schedule?

  • Hi everyone,
    I have 4 HDDs in my NAS.
    Everyone is already monitored, but I saw that there is also a schedule tab with 4 different type of self test.
    I wanted to ask a few things:
    1) What's the difference between normal monitoring and self test?
    2) Wich one should I do? Short, long, conveyance or offline?
    3) How often should I do that? does it shorten the HDD life span?


    Thanks in advance for the help!

    Intel G4400 - Asrock H170M Pro4S - 8GB ram - Be Quiet Pure Power 11 400 CM - Nanoxia Deep Silence 4 - 6TB Seagate Ironwolf - RAIDZ1 3x10TB WD - OMV 5 - Proxmox Kernel

  • Good question! Currently I have mine at once a month. I too would like to know this answer.



    My homebrew Nas
    OMV 4
    ASRock760GM-HDV
    AMD Phenom II X6 1065T
    Viper 3 16GB DDR3 1600MHz
    PNY 120GB SSD
    Two: Hitachi 4TB
    In a HP Pavilion case :D

  • I was thinking about once a week a short test, e once a month for the long one.
    But I'm not sure if it can hurt the disk doing it so often. I never found a complete answer for that, and I think that a "suggest schedule" should be in the wiki since it's a rather complex argument.

    Intel G4400 - Asrock H170M Pro4S - 8GB ram - Be Quiet Pure Power 11 400 CM - Nanoxia Deep Silence 4 - 6TB Seagate Ironwolf - RAIDZ1 3x10TB WD - OMV 5 - Proxmox Kernel

  • no one know the difference, or can link me some guide? I can't find anything usefull :(

    Intel G4400 - Asrock H170M Pro4S - 8GB ram - Be Quiet Pure Power 11 400 CM - Nanoxia Deep Silence 4 - 6TB Seagate Ironwolf - RAIDZ1 3x10TB WD - OMV 5 - Proxmox Kernel

    • Offizieller Beitrag

    This is more of a matter of preference.


    I do a short test on all drives once a week, afterhours. I believe that's enough to increment the counts on the "maybe it's time to worry" SMART attributes:


    SMART 5 – Reallocated_Sector_Count.
    SMART 187 – Reported_Uncorrectable_Errors.
    SMART 188 – Command_Timeout.
    SMART 197 – Current_Pending_Sector_Count.
    SMART 198 – Offline_Uncorrectable.


    I don't do long tests as a matter of routine - only after something odd appears. (Follow up.)

  • I think if we keep looking the web will find it :)

    I don't even know what software OMV is using to check the SMART data :(



    thanks!

    Intel G4400 - Asrock H170M Pro4S - 8GB ram - Be Quiet Pure Power 11 400 CM - Nanoxia Deep Silence 4 - 6TB Seagate Ironwolf - RAIDZ1 3x10TB WD - OMV 5 - Proxmox Kernel

    • Offizieller Beitrag

    I don't even know what software OMV is using to check the SMART data

    The command line program is smartctl .
    A sample command is: smartctl -x /dev/sda


    If you're asking how to check SMART in the GUI (note that you'd have to enable devices in the SMART settings tab).



    The extended information tab contains extensive details on each hard drive.

  • My homebrew Nas
    OMV 4
    ASRock760GM-HDV
    AMD Phenom II X6 1065T
    Viper 3 16GB DDR3 1600MHz
    PNY 120GB SSD
    Two: Hitachi 4TB
    In a HP Pavilion case :D

    • Offizieller Beitrag

    @neurotone
    I wouldn't worry just yet. While 1 reallocated sector is not good, it's not really bad either. (A few extra sectors are designed into a drive for this purpose - media errors.)
    I would keep an eye on it. If reallocated sectors start to increment upward, it's just a matter of time and it might go fast.


    Of note is, if the drive is between 4 to 7 years old (or the equivalent in hours) that's the rough expected life. They don't last forever.

    • Offizieller Beitrag

    I only run smart tests manually, when I intend to actually look at them. I never run them automatically. It might be once per month or so. Or even less often.


    The smart tests are good for early warning before the drive fails totally. I used to run tests daily but stopped. I don't really care about early warning. If the drive fails it fails. I run OMV on several single drive SBC NAS. I have good backups and redundant servers.


    It might be a different story if you run a big raid array with many drives with no redundant server. Then it might be critical to get early warning and swap drives before they fail for real.

  • I do a short test on all drives once a week, afterhours. I believe that's enough to increment the counts on the "maybe it's time to worry" SMART attributes

    Wrong. If you want to increase «the counts on the "maybe it's time to worry" SMART attributes» you need to run a long test.


    The difference is

    • short test: tests electronics (vendor specific), servos/motors (vendor specific), does some limited surface tests, finishes within 2 minutes
    • long test: the 2 min limitation is gone and whole surface will be tested, as such if there's a problem with a sector anywhere only the long test will show it


    A short test might result in the drive later switching the SMART overall-health self-assessment test from PASSED to FAIL (since the drive controller realizes position time of the heads increasing and such a servo about to die soon) but this will not increase values of those attributes you seem to be interested in.


    We do monthly long tests to get drives reporting uncorrectable errors and potentially mapping away bad sectors followed by ZFS resilvers the other day (scheduled on weekends since those are all business installations). If there's data corruption at the drive level it will be compensated by ZFS at the filesystem level directly afterwards.

    • Offizieller Beitrag

    Wrong. If you want to increase «the counts on the "maybe it's time to worry" SMART attributes» you need to run a long test.

    Wrong. SMART attributes can increment with no user scheduled tests at all.

  • thanks a lot for your answers!
    I wanted to ask a last thing: is it possible from the gui to set the long test on the first,second, third an fourth saturday/sunday of the month?
    Since it take 10 hours, I can only o it during saturday or sunday where I have my NAS up for that much (I put it on sleep at nigh/when I'm not at home).

    Intel G4400 - Asrock H170M Pro4S - 8GB ram - Be Quiet Pure Power 11 400 CM - Nanoxia Deep Silence 4 - 6TB Seagate Ironwolf - RAIDZ1 3x10TB WD - OMV 5 - Proxmox Kernel

  • with that it will do it every saturday right? I would like it to set it only 1 saturday at month

    Intel G4400 - Asrock H170M Pro4S - 8GB ram - Be Quiet Pure Power 11 400 CM - Nanoxia Deep Silence 4 - 6TB Seagate Ironwolf - RAIDZ1 3x10TB WD - OMV 5 - Proxmox Kernel

  • SMART attributes can increment with no user scheduled tests at all

    Exactly. But the context was your wrong believe that a short test could help increasing these SMART values which is not the case:

    I do a short test on all drives once a week, afterhours. I believe that's enough to increment the counts on the "maybe it's time to worry" SMART attributes

    Again: for this only a long test is sufficient since it's about to tell the drive controller to scan/verify the whole drive surface.

Jetzt mitmachen!

Sie haben noch kein Benutzerkonto auf unserer Seite? Registrieren Sie sich kostenlos und nehmen Sie an unserer Community teil!