14.15 Replacing a Failed Segment in a Software RAID

For some RAIDs, you can replace a failed segment by removing the segment from the RAID, replacing the failed disk, and then adding a segment to the RAID to replace the failed one. The following table provides recommended actions for each RAID type.

Table 14-4 Recommended Actions on Segment Failure in a Software RAID

Software RAID

Remove Segments

Recommended Action

RAID 0

Not allowed

If one or more disks in a RAID 0 fails, all data is lost on that device. You must re-create the RAID and restore its data from a backup copy.

RAID 1

All but one mirrored segment

The last segment of a RAID 1 device is the primary device that contains the data.

You can use the NSSMU Partitions option to remove mirrored segments to free up the related space on those segments, but the data on the last remaining segment is no longer mirrored.

If a disk fails in one of the segments, all of its data is lost on that segment. The remaining mirrors continue to operate normally. Remove the segment, replace the failed disk, and then add the segment as an element in the mirror. The data is synchronized over time until it is fully mirrored on the new segment.

RAID 5

One data segment at a time

You can temporarily remove one segment at a time in a RAID 5 device to replace a failed data disk.

For example, you can replace a single failed data disk while the system is still operational. Use the NSSMU Partitions option to remove the failed segment, replace the failed disk, add a segment to the RAID 5, and then restripe the RAID. The parity is used during restriping to restore the missing data. Read and write performance is degraded on the failed segment until the data is recovered because of parity-related reads and calculations.

If multiple data segments fail concurrently, all data is lost on that device. You must delete the RAID 5 and re-create it with good disks. Recover its data from a backup copy.

For example, if a second segment fails before the restriping is completed for the first drive replacement, this is considered a two-drive failure. You must recover data from a backup copy.

RAID 0+1

All but one mirrored segment; cannot remove disks from the underlying RAID 0

The last segment of a RAID 0+1 is a RAID 0 device that contains the original data.

If one or more disk fails in one of the mirrored segments, all data is lost on that segment. The remaining mirrors continue to operate normally. Remove the failed RAID 0 from the RAID 0+1. Delete the RAID 0, replace the failed disks, re-create the RAID 0, and then add the RAID 0 segment as an element in the mirror. The data is synchronized over time until it is fully mirrored on the repaired RAID 0 segment.

RAID 5+1

All but one mirrored segment; can safely remove one segment at a time per segment

The last segment of a RAID 5+1 is a RAID 5 device that contains the original data.

If a single data segment fails in a RAID 5 that is an element in a RAID 5+1, repair the RAID 5 while it is operational, as you would with any RAID 5.

If multiple disks in a mirrored segment fail concurrently, all data is lost on that segment. The remaining mirrors continue to operate normally. Remove the failed RAID 5 from the RAID 5+1. Delete the RAID 5, replace the failed disks, re-create the RAID 5, and then add the repaired RAID 5 as a segment in the RAID 5+1. The data is synchronized over time until it is fully mirrored on the repaired RAID 5 segment.

A Segment Fails in a RAID 0

If a segment fails in a RAID 0, you must delete the software RAID 0 device, create a new RAID 0 device, then copy your data to the RAID from backup media. For information, see Section 14.16, Deleting a Software RAID Device.

A Segment Fails in a RAID 1

  1. From the command console, enter nssmu.

  2. From the NSSMU main menu, select Software RAIDs.

  3. Remove the bad segment.

    1. Select the software RAID 1 device that you want to manage.

    2. Press Enter to show its member segments. The bad segment should show a status of Bad - Unavailable Partition.

    3. Select the bad segment, then press Delete.

  4. Expand the RAID with a replacement segment.

    1. Select the software RAID 1 device that you want to manage.

    2. Press F3 to increase the size of the RAID.

    3. From the list of available devices, select the device you want to use for the new segment. The segment size defaults to the size of existing partitions in the RAID 1.

    4. Select OK twice.

  5. The data begins mirroring automatically and continues until the segment is 100% mirrored.

A Single Data Segment Fails in a RAID 5

To replace a single failed data segment in a software RAID 5:

  1. From the command console, enter nssmu.

  2. From the NSSMU main menu, select Software RAIDs.

  3. Remove the bad segment.

    1. Select the software RAID 5 device that you want to manage.

    2. Press Enter to show its member segments. The bad segment should show a status of Bad - Unavailable Partition.

    3. Select the bad segment, then press Delete.

  4. Expand the RAID with a replacement segment.

    1. Select the software RAID 5 device that you want to manage.

    2. Press F3 to increase the size of the RAID.

    3. From the list of available devices, select the device you want to use for the new segment. The partition size defaults to the size of existing partitions in the RAID 5.

    4. Select OK twice.

  5. The restriping should begin automatically. If it does not, from the Software RAIDs page, select the RAID 5 device, then press F6 to restripe.

Multiple Segments Fail in a RAID 5

If two or more segments fail concurrently in a RAID 5 or if the parity partition fails, you must delete the software RAID 5 device, create a new RAID 5 device, then copy your data to the RAID from backup media. For information, see Section 14.16, Deleting a Software RAID Device.