RAID 50 Online but Files Are Corrupt After Rebuild
A controller can report RAID 50 online after rebuild while files remain corrupt. Online means the controller assembled a virtual disk; it does not prove that it used the correct member state, reconstructed every stripe correctly, or produced a consistent filesystem.
Stop writes. Keep the removed member and rebuild target, and do not run another rebuild, consistency check, or filesystem repair on the only copy.
What this condition means
The rebuild may have read through bad sectors, used a stale member, completed from a group with inconsistent parity, or reproduced damage that existed before the member was replaced.
Corruption can begin inside a lower RAID 5 group, in the stripe across groups, or in the filesystem. Those layers must be checked in that order.
What to record
- Removed member, target, bay, serial number, size, and lower group
- Rebuild start, completion status, and all errors
- When corruption first appeared and which files are affected
- Expected capacity and current virtual-disk status
- Consistency checks, restores, repairs, or writes performed after rebuild
What not to do
- Discard the removed member
- Repeat the rebuild or parity synchronization
- Run filesystem repair against the production array
- Overwrite corrupt files with a restore
- Assume online status proves the reconstruction is correct
How ADR evaluates it
ADR protects the removed member and target, then determines the best source state for the affected RAID 5 group.
The lower groups, upper stripe, filesystem, and priority files are validated before useful data is recovered to separate storage.
What happens next
Keeping both the old and rebuilt states can materially improve the recovery options. ADR can compare them without relying on the controller’s online status as proof.