Hi all,

I suspect I’ve made a mistake during a disk failure and could use some guidance to minimise data loss. I have two data disks and one parity drive. Below is the notification history and some screenshots I captured, which I hope help paint a clearer picture:

The start of the month parity check was fine:
09/01/2025

Notice [TOWER] - Parity-Check finished (0 errors)

Duration: 7 hours, 19 minutes, 42 seconds. Average speed: 151.6 MB/s

Then I started getting errors.

09/22/2025 at 3:01 AM

Alert [TOWER] - Parity disk in error state (disk dsbl)

ST4000VN008-2DR166_ZGY6SXTF (sdd)

Warning [TOWER] - array has errors

Array has 2 disks with read errors

I logged in to investigate, but the server was barely responsive. The GUI eventually froze, and I couldn’t retrieve diagnostics. Screenshot 2025-09-22 08:18 AM.

I pressed the power button to shut down the server, it took a while, and I believe it performed a clean shutdown. Though it hung at the diagnostics collection stage, I eventually had to do a hard power off. Screenshot 2025-09-22 11:26am.

I attempted to retrieve logs from the USB drive, but my only other computer didn’t recognise it.

Upon restarting the server, the parity drive was disabled and marked with a red cross. After a quick browse through the forums, I thought the correct action was to remove and re-add the parity drive to trigger a rebuild.

09/22/2025at 7:04 PM

Notice [TOWER] - array turned good

Array has 0 disks with read errors

Thinking the issue lay with the parity drive, I ran a full SMART test overnight. It completed successfully. I then ran short SMART tests on the other two disks, both passed. Confident things were stable, I initiated a parity rebuild before heading to work.

09/23/2025 at 7:38 AM

Notice [TOWER] - Parity disk, parity-sync in progress

ST4000VN008-2DR166_ZGY6SXTF (sdd)

Notice [TOWER] - Parity-Sync started

Size: 4 TB

09/23/2025 at 10:28 AM

Warning [TOWER] - array has errors

Array has 1 disk with read errors

09/23/2025 at 2:56 PM

Notice [TOWER] - Parity disk returned to normal operation

ST4000VN008-2DR166_ZGY6SXTF (sdd)

Notice [TOWER] - Parity-Sync finished (541656284 errors)

Duration: 7 hours, 17 minutes, 33 seconds. Average speed: 152.4 MB/s

09/23/2025 at 3:25 PM

Alert [TOWER] - Disk 1 in error state (disk dsbl)

ST4000VN008-2DR166_ZGY715JZ (sdc)

This is the current state, Screenshot 2025-09-23 205210.

Fresh diagnostics are attached. How can I determine what data has been lost, and what can be recovered?

Thanks in advance for your help.

tower-diagnostics-20250922-2035.zip

There are issues with multiple disks; check/replace all cables, especially if using any power splitters, and then post new diags after array start.

Thanks for taking a look.

For additional context, the server is an HP ProLiant MicroServer Gen8, so cable replacement options are limited.

I’ve given the system a thorough clean and reseated all the cables to ensure proper contact. I did the same with the HDD backplane, removed all drives and reseated them. I even moved Disk 1 to a different bay, just to be thorough.

Here’s the current state after startup. Do I need to do something to try enable the disk?

Fresh diagnostics.

tower-diagnostics-20250924-0953.zip

Cheers.

No disk-related errors so far. Since emulated disk1 is mounting and assuming contents look correct, you can rebuild on top:

• stop array

• unassign disk1

• start array

• stop array

• reassign disk1

• start array to begin the rebuild

Any more errors, post new diags before rebooting.

Thanks for the quick reply.

I guess my concern is the initial parity drive error:

09/22/2025 at 3:01 AM

Alert [TOWER] - Parity disk in error state (disk dsbl)

ST4000VN008-2DR166_ZGY6SXTF (sdd)

Warning [TOWER] - array has errors

Array has 2 disks with read errors

Followed by the sync with all the errors.

09/23/2025 at 2:56 PM

Notice [TOWER] - Parity disk returned to normal operation

ST4000VN008-2DR166_ZGY6SXTF (sdd)

Notice [TOWER] - Parity-Sync finished (541656284 errors)

Duration: 7 hours, 17 minutes, 33 seconds. Average speed: 152.4 MB/s

Can I trust that the data on the parity drive is reliable enough to justify overwriting Disk 1? Instead, I’m considering rebuilding to a new disk (due to arrive in a few days), so I can attempt data recovery from the current Disk 1 (not sure how yet).

My critical data is backed up elsewhere, but there is some nice-to-have data I'd like to minimise the risk of losing.

2 minutes ago, dominicmck said:

Can I trust that the data on the parity drive is reliable enough to justify overwriting Disk 1

You need to look at the contents of the 'emulated' disk1 to see if it looks OK as that is what will be rebuilt to a replacement disk.

SMART for parity looks fine, and since you are having issues with multiple disks, they are likely not the problem.

Hi all. Sorry for the delay, but I believe I'm finally back to a good place where I trust my data is healthy! Though it took some time to get there. Thanks as always for your support.

I ordered a new drive (which took a while to arrive), ran a pre-clear (unbelievably slow), and rebuilt onto the new disk (also slower than expected). Finally, I used BLAKE2 hashing to verify the rebuilt data against the old drive, all matched perfectly. I'm now looking into the Dynamix File Integrity plugin to give me peace of mind more quickly in future, though I still have a bit of research to do there.

One final question, people seem to speak highly of ZFS for error correction, as opposed to XFS which only offers error detection. I’ll need to do some research before considering the switch, are there any good resources people can recommend to help me understand whether it’s worth it?

17 hours ago, dominicmck said:

people seem to speak highly of ZFS for error correction, as opposed to XFS which only offers error detection.

XFS doesn't checksum the data; Btrfs and ZFS do, but note that for array devices, they can only detect corruption, not fix it. With redundant pools, they can detect and fix it.