Home / server / Questions / 1165865
Accepted
Scandinave
Asked: 2024-09-28 21:26:04 +0800 CST

Best way to replace a not yet faild drive

  • 772

I have a 4 drives Raid 5 array. One of the drive (sdb) show a lots of errors and my array show high latencies during access.

I currently have one new disk and another one is commanded.

As part of the replacement I also want to grow the raid level from raid 5 to raid 6. I identify to mean to achieve this :

By first growing to level 6

mdadm --manage /dev/md0 --add /dev/sdf1 
mdadm --grow /dev/md0 --raid-devices 5 --level 6 --backup-file /<path>/mdadm-backup
mdadm --manage /dev/md0 --replace /dev/sdb1 # Set faulty but keep it for replacement until a spare drive is available
# After my second drive arrived
mdadm --manage /dev/md0 --add /dev/sdg1

By first replacing the faulty drive then growing to raid level 6

mdadm --manage /dev/md0 --add /dev/sdf1 
mdadm --manage /dev/md0 --replace /dev/sdb1 --with /dev/sdf1
# after second drive is arrived
mdadm --manage /dev/md0 --add /dev/sdg1
mdadm --grow /dev/md0 --raid-devices 5 --level 6 --backup-file /path/mdadm-backup

Is there a difference between the two solutions ? Is one is safer than the other by imposing less operations on disks into the array ?

[Edit] adding smart result :

Sdb ( almost faulty )

ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  1 Raw_Read_Error_Rate     0x000f   082   072   006    Pre-fail  Always       -       22562487
  3 Spin_Up_Time            0x0003   093   093   000    Pre-fail  Always       -       0
  4 Start_Stop_Count        0x0032   100   100   020    Old_age   Always       -       124
  5 Reallocated_Sector_Ct   0x0033   077   077   010    Pre-fail  Always       -       29352
  7 Seek_Error_Rate         0x000f   087   060   030    Pre-fail  Always       -       4963671931
  9 Power_On_Hours          0x0032   026   026   000    Old_age   Always       -       65627
 10 Spin_Retry_Count        0x0013   100   100   097    Pre-fail  Always       -       0
 12 Power_Cycle_Count       0x0032   100   100   020    Old_age   Always       -       124
184 End-to-End_Error        0x0032   100   100   099    Old_age   Always       -       0
187 Reported_Uncorrect      0x0032   001   001   000    Old_age   Always       -       4793
188 Command_Timeout         0x0032   100   072   000    Old_age   Always       -       137445703785
189 High_Fly_Writes         0x003a   001   001   000    Old_age   Always       -       107
190 Airflow_Temperature_Cel 0x0022   073   056   045    Old_age   Always       -       27 (Min/Max 25/28)
191 G-Sense_Error_Rate      0x0032   100   100   000    Old_age   Always       -       0
192 Power-Off_Retract_Count 0x0032   100   100   000    Old_age   Always       -       82
193 Load_Cycle_Count        0x0032   100   100   000    Old_age   Always       -       296
194 Temperature_Celsius     0x0022   027   044   000    Old_age   Always       -       27 (0 14 0 0 0)
197 Current_Pending_Sector  0x0012   001   001   000    Old_age   Always       -       35384
198 Offline_Uncorrectable   0x0010   001   001   000    Old_age   Offline      -       35384
199 UDMA_CRC_Error_Count    0x003e   200   200   000    Old_age   Always       -       0

sbc

  1 Raw_Read_Error_Rate     0x000f   118   099   006    Pre-fail  Always       -       194860592
  3 Spin_Up_Time            0x0003   093   093   000    Pre-fail  Always       -       0
  4 Start_Stop_Count        0x0032   100   100   020    Old_age   Always       -       124
  5 Reallocated_Sector_Ct   0x0033   100   100   010    Pre-fail  Always       -       24
  7 Seek_Error_Rate         0x000f   087   060   030    Pre-fail  Always       -       654004767
  9 Power_On_Hours          0x0032   026   026   000    Old_age   Always       -       65631
 10 Spin_Retry_Count        0x0013   100   100   097    Pre-fail  Always       -       0
 12 Power_Cycle_Count       0x0032   100   100   020    Old_age   Always       -       124
184 End-to-End_Error        0x0032   100   100   099    Old_age   Always       -       0
187 Reported_Uncorrect      0x0032   094   094   000    Old_age   Always       -       6
188 Command_Timeout         0x0032   100   099   000    Old_age   Always       -       2
189 High_Fly_Writes         0x003a   062   062   000    Old_age   Always       -       38
190 Airflow_Temperature_Cel 0x0022   073   059   045    Old_age   Always       -       27 (Min/Max 24/27)
191 G-Sense_Error_Rate      0x0032   100   100   000    Old_age   Always       -       0
192 Power-Off_Retract_Count 0x0032   100   100   000    Old_age   Always       -       82
193 Load_Cycle_Count        0x0032   100   100   000    Old_age   Always       -       295
194 Temperature_Celsius     0x0022   027   041   000    Old_age   Always       -       27 (0 14 0 0 0)
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

sdd

ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  1 Raw_Read_Error_Rate     0x000f   113   094   006    Pre-fail  Always       -       51447952
  3 Spin_Up_Time            0x0003   093   093   000    Pre-fail  Always       -       0
  4 Start_Stop_Count        0x0032   100   100   020    Old_age   Always       -       124
  5 Reallocated_Sector_Ct   0x0033   100   100   010    Pre-fail  Always       -       0
  7 Seek_Error_Rate         0x000f   087   060   030    Pre-fail  Always       -       665278320
  9 Power_On_Hours          0x0032   026   026   000    Old_age   Always       -       65630
 10 Spin_Retry_Count        0x0013   100   100   097    Pre-fail  Always       -       0
 12 Power_Cycle_Count       0x0032   100   100   020    Old_age   Always       -       124
184 End-to-End_Error        0x0032   100   100   099    Old_age   Always       -       0
187 Reported_Uncorrect      0x0032   001   001   000    Old_age   Always       -       173
188 Command_Timeout         0x0032   100   099   000    Old_age   Always       -       1
189 High_Fly_Writes         0x003a   001   001   000    Old_age   Always       -       188
190 Airflow_Temperature_Cel 0x0022   073   055   045    Old_age   Always       -       27 (Min/Max 25/27)
191 G-Sense_Error_Rate      0x0032   100   100   000    Old_age   Always       -       0
192 Power-Off_Retract_Count 0x0032   100   100   000    Old_age   Always       -       82
193 Load_Cycle_Count        0x0032   100   100   000    Old_age   Always       -       296
194 Temperature_Celsius     0x0022   027   045   000    Old_age   Always       -       27 (0 14 0 0 0)
197 Current_Pending_Sector  0x0012   100   100   000    Old_age   Always       -       16
198 Offline_Uncorrectable   0x0010   100   100   000    Old_age   Offline      -       16
199 UDMA_CRC_Error_Count    0x003e   200   200   000    Old_age   Always       -       0
raid

1 Answers

  1. Best Answer

    Yes, there is a difference, but I'm not sure which is better.

    When you replace a disk in a RAID5, you need all other disks to be read-error free to reconstruct the new disk. With modern drives, that's no easy challenge. The read error chance on huge modern drives is almost such that you're guaranteed to run into one. That's why RAID5 is not recommended anymore. You can find a lot of info about this on the internet.

    By migrating to RAID6, you will also incur a write load on the old disks. This may even be advantageous: a drive can have 'current pending sectors' (see SMART). This means sectors it deemed damaged. It will reallocate them as soon as they receive a new write. If the sector is still not good enough to receive the write, the counter for 'reallocated sectors' will rise. The sector will then, transparently, reside elsewhere on the disk.

    My first step would be to look at the SMART status of your drives, and look at the above two parameters.

    But, I'm hesitant to recommend one way over the other if your drives indeed do have pending sectors. I think if they don't, I would first replace the disk in the RAID5, because that incurs a non-destructive operation on the other disks.

    And have/make backups...

    • 3