One drive has failed in my RAID 5 array, leaving it degraded and without redundancy. I have roughly 40 TB of data, and the remaining drives are already aging. I'm planning to build a new array, probably RAID 6, and eventually move everything over. Should I replace the failed drive and rebuild the existing RAID 5 first, or start copying the data to the new array immediately and retire the old one afterward? The surviving array may be under heavy load either way, and the drive replacement will take some time to arrive. I also tried installing a replacement drive, but the enclosure did not show activity or automatically start a rebuild, so I'm wondering whether the bay or RAID configuration may need attention. I do have backups, but I want to minimize the chance of losing the array during the migration.
4 Answers
If inserting the replacement disk did nothing, check the RAID management interface. Many controllers require you to mark the disk as a replacement, hot spare, or online member before rebuilding starts. Also test the slot and backplane, because the enclosure bay itself may have failed. Do not assume the original disk was the only problem until the hardware logs and SMART data confirm it.
Replace the failed drive and rebuild the array before migrating. Right now the RAID 5 has no remaining redundancy, so another disk failure during the data move could take down the whole array. Once it is healthy again, copy the data to the new RAID 6 setup and retire the old array. Make sure your backups are current before doing anything.
A rebuild does put a lot of stress and read activity on every surviving disk, so there is still some risk. I would avoid unnecessary work, but leaving the array degraded while moving 40 TB is risky too.
RAID is not a backup, so confirm that the important data exists somewhere else before stressing the array. For the next system, RAID 6 is a safer choice for a large disk set, but it still will not protect against accidental deletion, corruption, controller problems, or other non-disk failures.
The answer depends on the condition of the remaining drives and the reliability of your backups. Large RAID 5 arrays have a meaningful chance of encountering an unrecoverable read error during a rebuild, which can prevent recovery. If the old disks are healthy and you have verified backups, rebuild first; if the array is showing additional errors or the disks are failing, prioritize copying irreplaceable data to the new array while you still can.
A replacement drive not lighting up could mean the bay is bad, the drive is unsupported or dead, or the controller has not been told to use it. Check the controller status before forcing any rebuild.