I'm facing a worst-case scenario with my array and need advice on minimizing data loss.

Current situation:

• Running dual parity

• Lost three disks total

• Disks 1 and 2 were cleared and failing during rebuild (persistent read errors)

• While troubleshooting cables and power, disk 3 failed

• Due to how my shares are organized, I can afford to lose disks 1 and 2, but disk 3 contains critical data

My questions:

1. Can I emulate or bypass disk 3 to cut my losses on disks 1 and 2, or is the array unrecoverable at this point?

2. I'm planning to use Seagate Rescue Data Recovery for disk 3. Should I wait for them to recover the data before attempting to bring the array back online?
   

Any guidance would be appreciated. This has been a steep learning curve.

Are you really sure that disk3 has actually failed rather than simply being marked as disabled? I guess the same question applies to the other problem drives.

Are you able to get SMART reports for the drives?

Hi @itimpi, thanks for the quick response.

Unfortunately, disk 3 has definitely failed. During my troubleshooting, I accidentally used the wrong PSU cable and the SATA cable caught fire with visible smoke, quickly powered off but when I inspected disk 3, the PCB shows burn damage. I'm afraid to power it on at this point to verify if it still spins up or to pull SMART data.

Disk 1 and 2's data is already gone, which I've made peace with. My main concern now is whether there's any way to proceed with the array while I send disk 3 out for professional data recovery, or if I should just wait until the data is recovered before doing anything.

I do not think that there is a possibility to bring back the array in a state where recovery of the failed disks is possible. You might want to wait and see if @JorgeB concurs as the best expert on recovering from various combinations of disk failure.

If it is just a case of getting your array back into a usable state with just the contents of the remaining drives then you can use Tools->New Config to do this. In such a case easiest to start with the option to initially retain all assignments. When you then return to the Main tab all drives will be marked with blue icons (I mention this because many people forget to hit the Apply option in the New Config tool and then you do not get the blue icons), and you can remove the drives that have failed. If you now start the array Unraid will recognise the existing data drives and bring them online with their data intact, and will start rebuilding parity based on those drives.

Thanks for confirming, @itimpi . That's what I suspected.

I've been working with Seagate on disk 3, and it's covered under warranty with data rescue service available. They're estimating a 45-day turnaround, hopefully sooner. I'm fairly confident they can recover the data after swapping the PCB. My understanding is that Unraid doesn't encrypt data by default, so once they mount the drive, they should be able to access everything—is that correct?

My plan to bring the array back online:

1. Tools → New Config, then hit Apply

2. Confirm all drives show blue icons on Main tab

3. Unassign the 3 failed disks

4. Start array and let parity rebuild complete

5. Rebuild my ISO collection for disks 1 and 2 from scratch

6. Once Seagate returns disk 3 data, copy it back to the array

7. If the external drive they send back is shuckable, add it as a new disk and expand the array

Does this approach look solid, or am I missing anything?

The steps look good. If in doubt at any stage stop and ask here in the forum.

BTW: Make sure you have a backup strategy for anything important that keeps a copy off the server. Parity is not a substitute for backup.

4 hours ago, itimpi said:

The steps look good. If in doubt at any stage stop and ask here in the forum.

BTW: Make sure you have a backup strategy for anything important that keeps a copy off the server. Parity is not a substitute for backup.

Thank you! And yes absolutely agreed about parity not being a backup strategy. thankfully the most absolutely important docs and pics are in the cloud.

I’ve recently completed a deep-dive troubleshooting session following a system crash/PSU failure to verify the remaining hardware. I’ve confirmed that my PSU, SAS cables, and power splitters are fully functional by testing them in a separate, known-good server.

However, I’m seeing a strange brand-specific pattern: All of my Western Digital (WD/HGST) disks are completely unresponsive—they will not spin up at all. In contrast, my two Seagate drives survived; one is currently in data rescue, and the other is online and healthy.

Is there a specific 'circuit breaker' behavior or protection circuit (like TVS diodes) on WD/HGST boards that is more sensitive than Seagate's? I am also wondering if this could be a 3.3V Power Disable (PWDIS) reset loop, though I’ve tested them across different power environments.

Has anyone experienced a total 'WD-only' blackout after a power event while other brands survived? Are these drives likely toast, or is there a known protection state I can reset/bypass?"

27 minutes ago, withthePH said:

3.3V Power

Have you tried connecting them to power with a MOLEX-SATA adapter? MOLEX has no 3.3V pin so can't exhibit the 3.3V problem.

7 hours ago, trurl said:

Have you tried connecting them to power with a MOLEX-SATA adapter? MOLEX has no 3.3V pin so can't exhibit the 3.3V problem.

I was already running the 3.3v SATA Power Cable Mod on the SATA extension cables, per this guide, so my guess is that all the WD drives are toast then... I had my old Lantech with Power Adapter (without 3.3v line) and SATA to USB and tried that but nothing, just silence when power is delivered. Interesting that one Seagate IronWolf burned and the other Enterprise Seagate survived. Expensive lesson learned. Don't cheap out on PSU, make sure to validate and use that PSU's specific cables, and validate backups before troubleshooting esp when two disks already disabled.

New question! I got new eight 6TB disks pre-clearing, I’m wondering what the best steps are to take. I’m currently manually copying the sole surviving 12TB Seagate data to my other Unraid server and plan to use that as the new Parity. I assume I need to create a new configuration with 12TB as the Parity and the rest as array disks. Do the previous Shares get preserved? I am guessing no. So then, do I recreate them as how they were setup before and my dockers will be happy again? No VMs. I really appreciate any help you can provide to make this transition as smooth as possible!

40 minutes ago, withthePH said:

Do the previous Shares get preserved?

User Shares are simply the combined top level folders on array and pools.

If you create a user share in the webUI, Unraid creates top level folder(s) named for the share on array and/or pools as needed in conformance with settings for the share.

Conversely, if you create a top level folder on array or pools, it is automatically part of a user share named for the folder. Any share you don't make settings for has default settings.

If the folders exist, the shares exist. If the folders don't exist, the shares don't exist. But any settings you made previously for shares are part of your configuration on flash, and will be applied if the folders exist again.

On 3/13/2026 at 10:19 PM, trurl said:

If the folders exist, the shares exist. If the folders don't exist, the shares don't exist. But any settings you made previously for shares are part of your configuration on flash, and will be applied if the folders exist again.

awesome, this is great news. since the most important share (appdata) is on my NVMe and that cache pool is still healthy, it sounds like the transition will be smoother than I feared.

one quick follow-up: what happens to the 'Include Disks' settings for my old shares now that those physical disks (like Disk 1, 2, 3) no longer exist in the new array? will Unraid automatically ignore the missing disk IDs, or should I manually 'Clean' those settings to avoid errors?"

Edited March 15Mar 15 by withthePH

I am realizing something.. how does this sound to keep it clean? I'm planning to disable Docker/VMs, and then manually remove the old .cfg files for the failed shares from the flash drive's /config/shares/ directory before doing the New Config. I'll keep the appdata and system configs so they stay mapped to my NVMe. then I'll just recreate the data shares fresh in the UI. Does that sound like a solid 'clean slate' approach?"

That will work.