RAID 6 Data Recovery — Two Drives Failed, Rebuild Failed or Array Offline
RAID 6 tolerates two drive failures. When a third drive drops, a rebuild stalls or the controller loses its configuration, the array goes offline and the business stops. Our RAID engineers have recovered RAID 6 arrays from Dell, HP, Supermicro, Synology and QNAP systems since 2002 — with emergency 24–48 hour handling when downtime is costing you money.
RAID 6 is the level IT teams choose precisely because it survives two simultaneous drive failures. That is also why a RAID 6 failure is so disorientating: by the time the array is offline, something has already gone wrong twice — and the safety margin everyone relied on has been used up. This page explains why RAID 6 arrays still fail, what to do in the first hour, the actions that turn a recoverable array into a permanent loss, and exactly how our lab reconstructs a RAID 6 from its member drives.
Why RAID 6 arrays fail despite double parity
RAID 6 stripes data across every member drive and stores two independent parity blocks (usually called P and Q) for each stripe, so any two drives can be missing and the data can still be calculated. The protection is real, but it protects against one specific event — drive loss — and RAID 6 arrays in the field fail for reasons that double parity was never designed to cover:
- Correlated drive failures. Arrays are usually built from drives of the same model, batch and age, running in the same enclosure at the same temperature. When one drive fails from wear, the others are often close behind. Two failures in a week is common; a third during the rebuild is the classic RAID 6 loss.
- Rebuild stress and unrecoverable read errors. A rebuild reads every sector of every surviving drive, often for days on large-capacity disks. A drive that was “fine” under normal load surfaces bad sectors under that sustained read, the controller drops it, and the rebuild aborts with the array now beyond its tolerance.
- The array was already degraded and nobody knew. A failed drive in a RAID 6 does not stop anything, so it can sit unnoticed for months. The second failure is then the last line of defence, and the third takes the array offline.
- Controller failure or lost configuration. The RAID controller (a Dell PERC, HP Smart Array, LSI/Broadcom MegaRAID, or the software RAID in a Synology or QNAP NAS) holds the array’s metadata. A controller fault, a firmware update, a flat cache battery or a replacement card that fails to import the “foreign” configuration can present eight healthy drives as no array at all.
- Human error during recovery attempts. Pulling the wrong drive, reinserting drives in a different order, accepting a “initialise disk” prompt, or creating a new array on the old drives to “see if it works”. These are the failures that cause most of the genuinely unrecoverable RAID 6 cases we see.
- Power events and firmware bugs. A sudden power loss mid-write can leave parity inconsistent across the stripe; a NAS firmware update can leave a Btrfs, ext4 or ZFS volume that will not mount even though the underlying RAID is intact.
How a RAID 6 failure usually shows up
- Two or more bays show amber or red status lights, and the controller reports the virtual disk as Failed, Offline or Foreign rather than Degraded.
- A rebuild started after a drive swap has stalled at a fixed percentage, or completed but the volume is still inaccessible.
- The server boots but the data volume is missing, shows as RAW or unformatted, or Windows offers to initialise the disk.
- A Synology or QNAP NAS reports the volume as Crashed or the storage pool as degraded beyond repair, or the unit boots to a blank configuration.
- Virtual machines, SQL Server or Exchange databases hosted on the array will not start, or start with corruption errors.
Two of these — a stalled rebuild and a controller that no longer recognises the array — are the most common triggers for the calls we take from IT managers. Both are highly recoverable if the drives are left alone from this point.
What NOT to do
Don’t force failed drives back online. Most controllers let you mark a “failed” drive as online again. If that drive dropped out days ago, its data is stale; forcing it online tells the controller to trust out-of-date stripes and can silently corrupt the whole volume.
Don’t start, resume or retry the rebuild. A rebuild writes to the replacement drive using data calculated from the survivors. If a survivor has bad sectors, or the drive order has changed, the rebuild writes wrong data — and it is writing over the very stripes a lab would use to reconstruct your array.
Don’t re-initialise, re-create or “clear” the array. Creating a new virtual disk on the old drives, even with identical settings, rewrites the metadata and may zero the first sectors of every member.
Don’t run chkdsk, fsck, DiskPart or a NAS “repair volume” on a degraded array. File-system repair tools assume the underlying block device is sound. On a broken RAID they “fix” damage by deleting the references to your files.
Don’t run RAID recovery software against the live drives. Consumer tools that scan drives in-place can work on a healthy set of disks, but they put hours of sustained reads through drives that are already marginal, and they cannot handle the mechanically failed members that caused the problem in the first place.
Don’t reorder or unlabel the drives. If drives have been pulled, note where each one came from. Drive order can be worked out in the lab, but it costs time you may not have.
Safe checks and what to record before you call
None of the following writes anything to the drives, and all of it shortens the diagnosis:
Record the controller and array details. Server or NAS model, RAID controller model and firmware, number of drives, drive model and capacity, and the RAID level as configured (some “RAID 6” arrays turn out to be RAID 60 or RAID 6 with a hot spare).
Export or photograph the controller log. The PERC, Smart Array or MegaRAID event log — or the NAS system log — shows the order in which drives dropped and whether a rebuild was attempted. Photographing the screen is fine.
Label every drive with its bay number before anything is removed, and photograph the front of the enclosure with the drives in place.
Note what lived on the array. File system (NTFS, ReFS, ext4, Btrfs, XFS, ZFS), whether it hosted VMware or Hyper-V virtual machines, SQL Server or Exchange databases, and which data is the priority if the recovery has to be staged.
Check the backup honestly. If a restorable backup exists, a recovery may still be the faster route to the most recent data, but knowing what you have changes the service level you need.
What our customers say
Recently lost all my data but data…
Recently lost all my data but data clinic retrieved it successfully. They answered all my questions when we'd call/email. Great communication and service.
Thank you Janice et al
Thank you Janice et al. There are a lot of rip off companies out there who take advantage of data-loss panic. I have my drive to one of them and they tried to charge me a crazy amount. This company is not one them. Very happy.
My phone had water damage none of the…
My phone had water damage none of the usual repair places managed to repaire the phone and said my data had been lost. However Data Clinic we able to get all my data recovered from the internal memory storage within a few days of them receiving my device. Highly recommend
Recovered External Drive
After an external hard drive failure on my home PC, Data Clinic successfully recovered my files and put them on a replacement drive. My experience with DC was excellent, all the staff I spoke to were efficient and helpful. I would 100% recommend Data Clinic.
My 1TB external drive drop on the floor…
My 1TB external drive drop on the floor at home and the drive heads got damaged and I take the drive to a company that I'm not going to mention they name but after they done the diagnostic test they to me to recovery the data would cost me £2400 and I told them no way then after I got back the drive from them they called me 2 days after and said they will drop the price from £2400 down to £1500 so that tells me that they wanted to scam or something and other companies prices was high and I didn't want to send my hard drive to different companies. Next day I googled and found Data clinic And the first phone call and the information they give me I was happy. The lady on the phone sent me a PDF File with a break down of all the prices for recovery and for hard drive repairs. They Recovered 90 to 95% of my files And the price I Paid Data clinic to repair my hard drive and to recovered the data and transfer to a new hard drive was cheaper than the first company that wanted to scam me . Data clinic price was lower than the £2400 and the £1500 . Data clinic services is on point and I would use them again if I have any issues and I would recommend Data clinic service 100%
Honeymoon pictures rescued
I dropped my "waterproof" phone in a swimming pool while abroad. Because I was roaming, nothing was backing up to the cloud, meaning the entirety of our honeymoon memories were trapped on the device. Thinking the phone was waterproof, I did everything you’re not supposed to do—I tried turning it on and plugging it in to charge. It overheated so badly that I was certain I had destroyed the phone and lost everything. Despite the massive damage I caused, the experts at Data Clinic managed to recover every single byte of information! The logistics were incredibly smooth. From my initial contact, they arranged a next-day courier and didn't take a penny in payment until they confirmed the device was physically in their hands. My only constructive feedback is regarding their communication during the actual repair. After that initial confirmation, things went completely quiet for about three weeks with no ETA. Handing over your phone and PIN to a third party requires a lot of trust, and the radio silence definitely made me worry. However, I really didn't need to panic. Out of the blue, they contacted me to say everything was recovered, and just 24 hours later, a memory stick arrived with all my data. Highly recommended, even if your phone seems completely beyond saving! Thanks Janice and Mohammed - you have been so pleasant and helpful. And thanks to the entire DataClinic team.
I cannot recommend this team enough
I cannot recommend this team enough. Chris and Janice and the team went above and beyond to help me retrieve data lost from my phone and as it was mostly charity work they even fixed my phone as a donation to the charity. I was so anxious and I am sure drove them mad with my questions and there were some issues backing up the data due to my settings but Janice was there calm and kind every step of the way reassuring me and they overcame every hurdle and everything was found. A 5 star personal, human and kind service which is rare but so talented and professional as well. Thank you all so much. Amazing customer service.
A very smooth experience from start to …
A very smooth experience from start to finish. I had a large music collection I thought I’d lost forever. I phoned them up and the process was clearly explained to me. They got my drive on the 4/8/26 and I got my data back on the 11/8/26. I got all of my data apart from one file that they reported was damaged so a fantastic result for me. Very reasonable price and very pleased with the quick service. The only negative was that the online payment system is not very mobile phone friendly and I had to call them to make the payment but apart from that i am super pleased I chose them. I would certainly recommend this company to everyone.
Brilliant service
Brilliant service. Janice and the lab team are outstanding. Good communication and updates throughout. Files on 2 discs were retrieved, 1 with new heads needed. Good pricing - transparent and competitive. They've been amazing. I wouldn't hesitate to use them again or to recommend them.
Communication was engaging
Communication was engaging. The services offered, potential cost and timescales were clear at the outset. The courier collection and delivery were managed well. A good experience and a positive outcome.
A very effective service
My iPhone completely bricked on holiday and I tried everything, even a new screen but I couldn’t get it to work. I thought I had lost all my holiday photos but Data Clinic managed to recover everything that was on the phone and I couldn’t be happier! They are very friendly and knowledgeable and the service is very in depth. They are honest and upfront about costs and will outline the whole process for you. I would definitely recommend instead of trying to take apart the phone yourself, it’s worth the money!
Highly recommend Data Clinic
Janice and the team at data clinic have been brilliant through the whole process of helping recover my drive. i was really impressed they could show me my drive contents virtually and get everything back to me very quickly. I would highly recommend! Josh
Not sure whether the array is recoverable?
Call our RAID team for a straight answer on service level, likely outcome and next steps — or use the online diagnostic tool for single failed drives.
What to send us
For a RAID 6 recovery we need every member drive, including the ones the controller has marked as failed, plus any hot spare that was in the enclosure. A drive the controller rejected is frequently still readable in the lab, and with a two-drive-tolerant array those “failed” members are often the difference between a complete recovery and a partial one. If the drives are still in the server or NAS and you would rather not remove them, send the whole unit — we will remove and label the drives ourselves. Include the RAID controller card where the array is hardware RAID, or at minimum the model and configuration details recorded above. Our guide on why we need all the RAID drives explains the reasoning in more detail.
We can arrange a courier collection from any UK address; for RAID, NAS and server jobs the assessment and collection charges are confirmed when you book. Business-critical jobs go on our Emergency service, and everything comes to our lab in Bury, Manchester, where all recovery work is carried out.
How professional RAID 6 recovery works
A RAID 6 recovery is never done on the original drives. The whole process is built around imaging first, so that nothing the lab does can make the situation worse, and then reconstructing the array virtually from those images.
- Booking-in and labelling. Each drive is logged against its bay number and serial, photographed, and given a case reference. For police, legal and regulated clients we document chain of custody from receipt to return.
- Individual drive assessment. Every member is tested separately on hardware imagers with write-blocking. We establish which drives are healthy, which have bad sectors, and which have mechanical or electronic faults — the controller’s opinion of a drive is a starting point, not a verdict.
- Repair of failed members. Drives with failed heads, seized motors or firmware faults are repaired in our cleanroom using matched donor parts, then imaged. Drives with read instability are imaged head-by-head at reduced speed with multiple passes, so weak areas are recovered rather than skipped.
- Sector-level imaging. Each drive is cloned to healthy media. The originals are then set aside and never written to.
- Parameter analysis. From the images we determine the array geometry: drive order, stripe (block) size, start offset, parity rotation and the P/Q layout the controller used. This is done analytically from the data itself, so it does not depend on the failed controller or its metadata.
- Virtual reconstruction. The images are assembled into a virtual RAID 6. With two missing or partial members, the dual parity is used to calculate the missing blocks; where a drive dropped out earlier than the others, we use the most recent consistent set of stripes rather than stale data.
- File-system and application recovery. The reconstructed volume is analysed as NTFS, ReFS, ext4, Btrfs, XFS, ZFS or VMFS, and the files, virtual machine disks (VMDK, VHDX) and databases (SQL Server, Exchange) are extracted and verified. Where a rebuild or repair tool ran before the drives reached us, this is the stage where its damage shows and is worked around.
- Verification and return. You receive a file listing to check before the data is copied to new media and returned by tracked courier. If your information-security policy sets requirements for how data is returned, tell us at booking and we will agree the method with you before dispatch.
Where the array hosted a hypervisor or database, we prioritise the extraction: the production database or the critical virtual machines first, the rest afterwards. That staging is what lets a business get back to work while the remaining data is still being processed.
A typical RAID 6 recovery scenario
The following is a composite of the RAID 6 cases we see most often, not a single client’s job.
An eight-drive RAID 6 in a rack server at a manufacturing SME holds the ERP database and the file shares. One drive failed in the spring and was never replaced. Over a bank-holiday weekend a second drive fails; the controller carries on, degraded. On the Tuesday the IT contractor swaps the first failed drive and starts a rebuild. Twenty hours in, a third drive throws read errors under the rebuild load and the controller drops it. The virtual disk goes offline, the ERP system stops, and the last full backup is ten days old.
In the lab, the drive that dropped during the rebuild images at 99.9% with a small number of unreadable sectors. The spring failure has a failed head stack and is repaired and imaged in the cleanroom. The second failed drive has a firmware fault and images fully after repair. With those images the array geometry is confirmed, the partially completed rebuild is identified and excluded, and the volume is reconstructed with the dual parity filling the gaps left by the bad sectors. The ERP database is extracted and verified first and returned on the emergency service; the file shares follow.
Cost, turnaround and service levels
RAID 6 recoveries are quoted individually, because the cost depends on how many drives need physical repair and how much reconstruction work the array needs. We operate transparent pricing: after assessment you receive a fixed quote before any recovery work starts, and you decide whether to proceed. If you choose not to proceed after diagnosis, your drives are returned for a £20+VAT handling fee.
- Standard service — 5–20 working days.
- Priority service — 2–5 working days, for arrays where the business is working around the loss but cannot for long.
- Emergency service — as fast as possible, usually 24–48 hours from the drives arriving. Emergency jobs jump the queue and are worked through to completion, including evenings and weekends where needed. Full details on our emergency data recovery page.
You deal with one engineer-led team from booking to return. If you need a written statement of work for your insurer, board or auditor, ask at booking. For the wider picture of how we handle servers and arrays of every level, see our RAID data recovery service page; for Synology, QNAP and other NAS units running RAID 6, see NAS data recovery; and for a single failed disk outside an array, our hard drive data recovery service applies.
Frequently asked questions
Two drives have failed in my RAID 6. Is the data still there?
Almost always, yes. RAID 6 is designed to keep working with two drives missing, so the data on the surviving drives is complete. The risk is what happens next: a third drive failing under rebuild load, or a rebuild or repair attempt writing over good stripes. Power the array down and get the drives assessed before anything else is tried.
Three drives have failed in my RAID 6. Is it hopeless?
No. In most three-drive cases at least one of the “failed” drives is readable in a lab, either because the controller dropped it for slow responses rather than a genuine fault or because its bad sectors can be worked around during imaging. Once one of the three is recovered, the dual parity covers the other two.
The rebuild failed part-way through. Has it destroyed my data?
Usually not. A stalled rebuild has written to the replacement drive only, and the lab reconstructs the array from the original members and excludes the partial rebuild. The danger is repeated attempts or forcing drives online afterwards, which is why we ask you to stop at the first failure.
My RAID controller has failed. Do I need the same controller to recover the data?
No. We reconstruct the array from the drive images by analysing the data layout directly, so the recovery does not depend on the original controller or its metadata. Sending the controller helps with diagnosis, but a failed or unavailable card does not prevent recovery.
Can you recover the virtual machines and databases on the array, not just files?
Yes. Once the RAID 6 is reconstructed we extract and verify VMware VMDK and Hyper-V VHDX virtual disks, SQL Server and Exchange databases, and the file systems inside them. We prioritise the production systems you need first so the business can restart while the rest is completed.
How quickly can a RAID 6 be recovered?
On the Emergency service the target is 24–48 hours from the drives reaching our lab. Mechanically failed drives that need cleanroom repair can extend that, and we tell you on the initial call rather than after collection. Priority (2–5 working days) and Standard (5–20 working days) services are available where the pressure is lower.
RAID 6 array down right now?
Power it down and leave the drives alone — the state they arrive in decides the outcome. You get a fixed quote before any recovery work starts. Leave your details and a RAID engineer will call you back.
✓ Thank you — we have your details
One of our data recovery specialists will call you back shortly. If it’s urgent, call us free now on 0800 151 2207.
Prefer to talk it through?
About Data Clinic — A UK data recovery specialist since 2002, recovering data from RAID arrays, servers, NAS units, hard drives, SSDs and phones at our lab in Bury, Manchester. Fully GDPR-compliant and ICO registered, trusted by organisations including the BBC, Sony, Tesco, HSBC and Williams F1, and rated 4.8/5 on Trustpilot. Freephone 0800 151 2207.