6 hours ago, trurl said:

format before preclear is pointless since preclear writes zeros to the entire disk, wiping out whatever filesystem you formatted. Preclear might be a good test of the disk. No point in formatting it after preclear. And you absolutely MUST NOT format the disk after you put it back in the array for rebuild.

Preclearing seems to be going extremely slow. I am at 40 mb/s, and 5% complete after around 4 hours 40 min. When I initially set up the server, at this point the preclear was about 6x fast, around 250mb/s (falling below 200mb/s towards the end of the preclear cycle).

Additionally in the logs this error constantly pops up:

Feb 19 20:35:27 Babel kernel: I/O error, dev sdc, sector 1384611712 op 0x1:(WRITE) flags 0x104000 phys_seg 168 prio class 2

(and over and over again for many sectors)

In the disk log specifically, I get this:

Feb 19 15:57:22 preclear_disk_ZRT2F5W4_1023928: blockdev: ioctl error on BLKRRPART: Input/output error

The preclear keeps running though, so I'll let it run overnight unless anyone recommends otherwise. But at this rate it will probably take a week to complete one cycle! (I skipped the pre-read).

If this helps, originally the rebuild failed after ~8 hours, attempt 2 failed after 2-3 hours, and subsequent attempts failed within minutes.

Diagnostics attached in case that would help

babel-diagnostics-20260219-2049.zip

Still not getting a complete smart report for that disk. Can you plug it in to a different port? Or maybe change cables? Or both?

1 hour ago, trurl said:

Still not getting a complete smart report for that disk. Can you plug it in to a different port? Or maybe change cables? Or both?

I have already changed the cables twice (first time was with extra Aliexpress junk cables, second time with proper Startech cables).

In terms of SATA ports on the motherboard itself, I have changed ports before already without any improvement.

If by port you mean the actual slot the hard drive is plugged in to (the bays), I have moved the drives around multiple times. Moving "disk 1" to a 3rd slot didn't fix the issue. Swapped the slots between parity and the other disk (now parity is in the 3rd slot, "disk 1" in the 1st slot), and the parity drive still works fine and is able to perform a short SMART test without any problems. So the bay isn't the issue, nor is the SATA cable (both the new cables and the Aliexpress junk)

I've found a relatively cheap (by today's standards) 4TB drive which is enough to test a rebuild, will have an answer in ~24 hours to test if a rebuild successfully works (assuming it is still in stock tomorrow)

Forgot to add: SMART test on the affected disk 1 still doesn't work. Diagnostics attached

babel-diagnostics-20260220-0015.zip

Edited February 20Feb 20 by Wazzy
forgot diagnostics

7 hours ago, Wazzy said:

found a relatively cheap (by today's standards) 4TB drive which is enough to test a rebuild

You can't replace a 12TB disk with a 4TB disk.

2 hours ago, trurl said:

You can't replace a 12TB disk with a 4TB disk.

I confused myself with expanding the array with another smaller drive vs. replacing the drive.

I have ordered a 12TB drive (WD Red Pro) to figure out what's going on.

Good news: The new drive arrived last week, and I was able to rebuild successfully with the new drive, so it's confirmed the old drive died. It is a helium drive so I suspect a slow leak was the cause? Anyways When I tried preclearing the old drive, it showed as zeroing at over 3000 mb/s 😅

Few concerns:

1. This error showed up during the rebuild: Notice - Array health report [FAIL], Array has 3 disks (including parity & pools) - see screenshot. From this post apparently this is intended, but I'm bringing it up along with diagnostics just in case (and because of concern #2)

2. I ran an extended SMART test, and there was a very loud noise coming from the new hard drive when the SMART test was near the end. It lasted about 10 minutes, and stopped once the Extended SMART test was complete. I have uploaded a video with the noise.

Diagnostics also attached, however the server seems to be running smoothly without any issues.

Thank you for your help!

babel-diagnostics-20260305-2202.zip

Looks good.

Your appdata and another share

appdata                           shareUseCache="only"    # Share exists on cache, disk1
b----------l                      shareUseCache="no"      # Share exists on disk1, cache

appdata has files on the array, and they won't be moved to cache because there is no Secondary so no Mover action. Nothing can move open files anyway so you would also have to disable Docker to allow those to be moved.

The other share has files on cache which won't be moved to the array because there is no Secondary so no Mover action.

See if you can fix those and post new diagnostics.

babel-diagnostics-20260308-1847.zip

On 3/5/2026 at 11:38 PM, trurl said:

Looks good.

Your appdata and another share

appdata                           shareUseCache="only"    # Share exists on cache, disk1
b----------l                      shareUseCache="no"      # Share exists on disk1, cache

appdata has files on the array, and they won't be moved to cache because there is no Secondary so no Mover action. Nothing can move open files anyway so you would also have to disable Docker to allow those to be moved.

The other share has files on cache which won't be moved to the array because there is no Secondary so no Mover action.

See if you can fix those and post new diagnostics.

Should be fixed - Appdata only exists on cache, and that second share exists only on the disk.

Diagnostics attached

Edited March 8Mar 8 by Wazzy

2 hours ago, Wazzy said:

Appdata only exists on cache, and that second share exists only on the disk

👍

For anyone reading this later: My drive was slowly failing until it finally failed during one of the rebuilds. Drive was RMA'ed and the replacement works perfectly fine