A disk in your array failed, you slotted in a replacement, and the rebuild won’t complete — it stalls, errors out, or drops the whole array. This is one of the most dangerous moments in the life of a RAID, and what you do in the next few minutes matters enormously. The instinct is to try the rebuild again. That instinct is usually wrong.
A failed rebuild means the array is fragile. Retrying stresses already-weak disks and can cause more failures. Power it down, don’t force it, and image the disks instead.
To understand the danger, you need to know what a rebuild actually does. When you replace a failed disk, the array reads every remaining disk in full, intensively, to reconstruct the missing data onto the new drive. That’s a heavy, sustained workload on disks that have often been running together for years.
And here’s the problem: those remaining disks are frequently aged and carrying weak sectors of their own. Under the strain of a rebuild, a second disk can hit an unreadable sector or fail outright — and on an array with no redundancy left during the rebuild (RAID 5 especially), that’s enough to make the rebuild fail and the array drop. Disks bought together tend to wear out together, which is why one failure so often triggers another.
When a rebuild fails, the temptation is to restart it — maybe it’ll work this time. Don’t. A failed rebuild tells you the array is in a fragile, degraded state, and each retry runs the same punishing read across the same weak disks, risking yet another failure and pushing you further from recovery.
Even more damaging is trying to “fix” it by recreating or reinitialising the array — a step some controllers offer that can overwrite the array’s configuration and make the data far harder, sometimes impossible, to recover. The array is trying to tell you it can’t safely rebuild; forcing it rarely ends well.
When a rebuild fails, the safe sequence is short and deliberate:
Don’t retry the rebuild. Power the array down to take the strain off the remaining disks and prevent further failures.
Never reinitialise, recreate or reconfigure the array to ‘fix’ it — that can overwrite the very information needed to recover the data.
Label the disks with their slot positions and keep even the ‘failed’ one — it often still holds readable data that helps the recovery.
Have each disk imaged and the array rebuilt virtually from the copies — safely, without stressing the fragile originals.
The reason a failed rebuild is a job for specialists is simple: rebuilding on live, degraded disks is the very thing that’s failing. A lab does it differently — every disk (including any that “failed”) is imaged to healthy storage first, using tools that read gently around weak sectors, and the array is then reconstructed virtually from those images. All the risky work happens on copies, so the fragile disks are read once and never rebuilt.
This is why a failed rebuild is usually recoverable when handled properly, and often unrecoverable when forced — the difference is whether the weak disks were protected or pushed.
Two lessons come out of every failed rebuild. First, a RAID is not a backup — it’s uptime protection, and a failed rebuild is exactly the scenario where a separate backup saves you. Second, because disks age together, it’s worth replacing ageing arrays proactively and monitoring disk health, rather than waiting for the first failure to expose how worn the rest are.
Redundancy buys you a window to act; it doesn’t remove the need to back up and maintain the array.
What people ask us most about failed RAID rebuilds.
No. A failed rebuild means the array is fragile, and retrying runs the same intensive read across already-weak disks, risking another failure and further loss. Power the array down instead, don’t recreate or reinitialise it, keep the disks in order, and have them imaged and reconstructed by a specialist. Repeated rebuild attempts are one of the main reasons recoverable arrays become unrecoverable.
Because a rebuild forces a heavy, sustained read across every remaining disk to reconstruct the replacement — and those disks are often aged and carrying weak sectors, having run together for years. Under that strain a second disk frequently fails or hits an unreadable sector, which drops the rebuild. It’s especially risky on arrays with no redundancy left during the rebuild.
Usually yes, if you stop in time. Recovery works by imaging every disk — including any that failed — to healthy storage, then reconstructing the array virtually from those copies, so nothing stresses the fragile originals. The main thing that turns a recoverable array into a lost one is forcing repeated rebuilds or recreating the array, so the key is to power down and not force it.
Don’t retry or recreate the array — that’s what loses data. Power it down, keep the disks in order, and send it in. We image and reconstruct safely from copies. Post it in from anywhere in the UK, or drop it to us in Belfast.