4 disc Raid5: S.M.A.R.T attributes show errors - S.M.A.R.T. Test PASSED!

  • Hi,
    i'm running a 4 disc mdadm RAID5


    sda (3TB)
    sdb (3TB)
    sdc (3TB)
    sdd (3TB)


    One drive (sdb) shows a red dot in OMV SMART settings (Popup message: Drive has few bad blocks).
    Here a log of Smart-Attributes:


    Not so good: :/
    Raw_Read_Error_Rate: 8791 (other discs show 0)
    Reallocated_Sector_Ct: 3 (other discs show 0)
    Current_Pending_Sector: 142 (other discs show 0)


    Should i be concerned? The Raid-Status is clean, all is working fine, except the strange Smart-Status log.


    sdb has few bad sectors which are pending. I tried to get the bad block numbers to reallocate these, but Smart health-test PASSED! ?(

    Code
    root@Server:~# smartctl -H /dev/sdb
    === START OF READ SMART DATA SECTION ===
    SMART overall-health self-assessment test result: PASSED

    Did selftests and they completed without errors and do not show bad blocks in LBA_of_first_error! Why? There are pending sectors!


    Code
    root@Server:~# smartctl -l selftest /dev/sdb
    SMART Self-test log structure revision number 1
    Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error  
    # 1  Short offline       Completed without error       00%      7195         -                   
    # 2  Extended offline    Completed without error       00%      7180         -                   
    # 3  Short offline       Completed without error       00%      7172         -

    tune2fs, fdisk need the exact block numbers to reallocate/overwrite pending blocks, but there are no block faulty blocks under LBA_of_first_error
    How can i fix the pending sectors without loosing raid5 or data-loss?


    Please help
    pappl

    • Offizieller Beitrag

    I have very little experience with RAID and SMART errors.


    But I would do this:


    1. Order a new replacement HDD at once!
    2. Check that your backups are safe.
    3. Consider not using the NAS until you have replaced the HDD, creating extra backups may push the disk over the edge, perhaps along with more disks, resulting in total loss of all data. But you have backups, don't you? So that is not a big problem.

  • 5 Reallocated_Sector_Ct must be 0, you have still 140 sector to avoid data loss, so you have time to order a new disk or return in guaranty your defect disk

  • Thanks for your help!


    5 Reallocated_Sector_Ct must be 0, you have still 140 sector to avoid data loss, so you have time to order a new disk or return in guaranty your defect disk

    Good news!


    Which drive vendor? In case it's Seagate you might want to educate yourself about 'numbers without meaning' (at least when interpreted as decimal numbers which is wrong:( http://www.users.on.net/~fzabk…Seagate_SER_RRER_HEC.html

    WD Red 3TB, thanks for the information.


    I have very little experience with RAID and SMART errors.


    But I would do this:


    1. Order a new replacement HDD at once!
    2. Check that your backups are safe.
    3. Consider not using the NAS until you have replaced the HDD, creating extra backups may push the disk over the edge, perhaps along with more disks, resulting in total loss of all data. But you have backups, don't you? So that is not a big problem.

    Backups are made for all shares i don't want to lose.


    How do i replace one drive of the 4 disc raid5 system?


    - Buy a new 3TB drive
    - Power off the server (OMV 4.1.17)
    - Replace the faulty hdd
    - Power on and log into OMV web administration
    - Raid5 will resync automaticly for hours? Or do i need to keep additional settings in mind?



    pappl

    • Offizieller Beitrag

    Power off the server (OMV 4.1.17)
    - Replace the faulty hdd
    - Power on and log into OMV web administration
    - Raid5 will resync automaticly for hours? Or do i need to keep additional settings in mind?

    No, all this will need to be completed from the cli.


    If you search for replacing a hard drive in an mdadm raid 5 you'll get some answers. This is just one the only thing I'm not sure about is if the array has to be stopped first.

  • After reading about Raid5 array data loss, i don't want to wait until the suspicious drive gets more errors.
    If a second drive fails all data is lost, i should have done a more secure Raid6 array. I have a backup of all important data, but not for all data due to high costs.


    I'm a little afraid of swapping the drive and rebuild process, because some users get problems after swapping a still functioning drive in a clean Raid5 array. (Rebuild not possible, array afterwards not visible, array mount not possible, data lost,...) =O
    So the drive which has to be replaced must be removed from the array via cli first.
    Every tutorial seems to have another command options (--force,...).
    Interesting there is no noob-proof sticky tutorial thread about swapping a still working drive/faulty drive and rebuilding Raid5 after all these years.




    pappl

    • Offizieller Beitrag

    Interesting there is no noob-proof sticky tutorial thread about swapping a still working drive/faulty drive and rebuilding Raid5 after all these years.

    Nothing is noob proof :) here is a thread you may find interesting, for the average home user the use of a raid set up makes no sense, @Adoby has an excellent thread on here describing his own set up, that is a much better option than using raid. Raid is easy to set up can be a PITA recovering.

  • If a second drive fails all data is lost


    It's worse than that with most RAID-5 implementations. A single URE (unrecoverable read error) on one of the remaining disks occurring when rebuilding a RAID-5 can stop the whole rebuild and your whole array is lost.


    Traditional RAID is not about data safety but only about data availability and nobody at home needs this. Unfortunately almost everyone at home playing RAID forgets about backup.

Jetzt mitmachen!

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