RAID 5 is designed to survive the failure of one drive. When a disk dies, the array keeps running in a “degraded” state, and once you replace the disk it rebuilds the missing data from parity. In practice, the rebuild is the most dangerous moment in the life of a RAID 5 array. Here’s why rebuilds fail and how to avoid turning a recoverable situation into a lost one.
Why RAID 5 rebuilds fail
- A second drive fails during the rebuild. Drives in an array are usually the same model, the same age and from the same batch, so they wear out at similar times. A rebuild reads every sector of every remaining disk for hours, which is a heavy workload for drives that are already old.
- Unreadable sectors on a “healthy” drive. If another disk has a bad sector the rebuild needs, many controllers stop the rebuild or mark that disk as failed too.
- The wrong disk was replaced. Pulling a working drive instead of the failed one takes the array offline immediately.
- Controller or configuration problems. A failed controller, a reset configuration or disks put back in the wrong order can all make the array unreadable.
Mistakes that make recovery harder
Never initialise or re-create the array to “get it working again”. Initialising can overwrite the RAID metadata and, on some controllers, start wiping the disks.
- Forcing a failed drive back online without understanding why it dropped out.
- Running file system repair on a degraded or partially rebuilt array.
- Swapping disks between slots, or labelling them only after removal.
- Restoring a backup onto the same disks before you know the backup is complete.
What to do if your rebuild has failed
- Stop. Don’t start another rebuild, and power the system down if it is behaving erratically.
- Label every disk with its slot number before removing anything.
- Record the details: controller or NAS model, RAID level, number of disks, and any error messages or logs.
- Don’t write anything to the array. That includes reinstalling the operating system on it.
How a RAID recovery works
In the lab, each disk is first imaged individually. Any drive with physical faults is repaired in the clean room before it is copied. The array is then rebuilt virtually from the images: we work out the disk order, stripe size and parity layout, and reconstruct the volume without ever writing to your original disks. That’s why it matters so much that nothing has been written to them.
We recover RAID 5, RAID 0 and RAID 1 arrays, servers and NAS devices from Synology, QNAP and others. If your array has failed, call 020 7394 2529 before trying anything else. Our emergency service is available 24/7 when the business is down.