RAID 6 is meant to be the safe choice. It can lose two drives and keep running, so when a server or NAS reports that the array has failed, or a volume drops offline overnight, the first reaction is often disbelief. The second is usually to reach for the controller’s force-online option. Don’t do that yet. A failed RAID 6 array is often recoverable, but the first hour matters, and a few well-meant steps can overwrite exactly what a recovery needs. Here is what to do, and what to leave alone.
How RAID 6 fails
RAID 6 stripes your data across all the drives and stores two independent sets of parity. With any two drives missing, the array can still calculate the data. Lose a third while those two are still out, and the volume goes offline. That sounds unlikely, but in practice it tends to happen in a few familiar ways:
- Nobody saw the first failure. The array runs degraded for weeks because alert emails stopped working or the amber light is in a closet nobody visits.
- A rebuild pushed weak drives too far. Rebuilding reads every sector of every surviving drive. Drives of the same age and batch often wear out together, and the extra load can expose bad sectors or finish off a marginal drive.
- Something other than the drives failed. A bad controller, cable, backplane or power event can knock several healthy drives out at once.
- The configuration was lost. After a firmware update or controller swap, the array can show up as foreign, missing or uninitialized even though the data is still on the drives.
What to do first
- Stop writing to the array. Pause backup jobs, databases, virtual machines and users’ access. If the volume is still mounted and readable, copy your most critical data to separate storage now, and only that data.
- Record everything. Before shutting down, photograph or export what the management screen shows: the controller or NAS model, RAID level, stripe size, drive states and the event log. Note the order in which the drives dropped, if the log shows it.
- Shut down cleanly if you can. A proper shutdown is better than cutting power. If drives are clicking or the system is hanging, don’t wait around; power it off.
- Label each drive with its slot number before anything is removed, using the bay number on the chassis. Drive order matters when the array is rebuilt.
- Call your IT provider or a recovery engineer before trying a fix. If several people have access to the server, tell them it’s off limits.
What not to do
- Don’t force a failed drive back online. A drive that dropped out days or weeks ago holds old data. Bring it back and the controller may treat that stale data as current, mixing old and new blocks and corrupting the file system.
- Don’t start or restart a rebuild. If a surviving drive has weak sectors, a rebuild can knock it out too. Parity written from bad data will also overwrite good data.
- Don’t create a new array or initialize anything. Dell PERC, HPE Smart Array and Broadcom MegaRAID controllers, along with most NAS systems, may offer this as the obvious next step. It can overwrite metadata or wipe the contents.
- Don’t change drive positions or test the drives one at a time in another machine, which may write to them.
- Don’t run CHKDSK, fsck or a NAS repair tool on a volume that has just come back from a failure. These tools assume the array underneath is sound. If it isn’t, they “fix” the file system by damaging it.
- Don’t replace the controller and hope. A new controller, or the wrong firmware, can misread the existing setup.
Can the data be recovered?
Often, yes. RAID 6 needs all but two of its drives to rebuild the data, so the question becomes whether enough of the failed drives can still be read. A drive that the controller rejected is not always dead. It may have been dropped for a timeout or a handful of bad sectors, and it may still give up most of its data when it’s handled carefully.
A recovery engineer doesn’t repair the array in place. Each drive is copied to healthy storage first, the array’s layout is worked out from the data on those copies, and the volume is rebuilt virtually. The original drives aren’t written to. That also means drives that dropped out early can be identified and kept out of the rebuild, so stale data doesn’t get mixed in.
What hurts the odds is anything that wrote to the array after the failure: a forced rebuild, a re-created array or a repair tool that ran for hours. That’s why stopping early counts for so much. For more detail, see our RAID data recovery and server data recovery pages, or our NAS data recovery page if the array lives on a Synology or QNAP unit.
Once you’re back up
- Remember that RAID is not a backup. It covers drive failure, not ransomware, deletion or a bad controller, so keep a separate backup and test restoring from it.
- Make sure alerts reach someone who will act, and send a test alert to prove it.
- Run the controller’s scheduled consistency checks or NAS data scrubbing, which can find weak sectors before a rebuild does.
- Replace a failed drive promptly, and think about retiring the rest of an aging set together.
Need help with a failed RAID 6 array?
Leave the array powered down and call Data Retrieval at (213) 986-4855. An engineer will evaluate the drives and give you a fixed quote, and recovery only starts once you approve it. We work on-site across Los Angeles, the Inland Empire and Ventura County, and we accept mail-in recoveries from anywhere in the US.