I get it -- there are a lot of threads about this problem, but this is not my area of expertise and there is a lot of information obfuscated in the varied responses for existing issues.

This weekend, one of my drives showed 6 read errors. Overnight, that jumped to over 7100. I don't know if it was the right move, but I performed a correcting parity check. It made sense at the time, but as I'm typing now, if the drive is having read errors, writing those incorrect values to parity probably isn't the appropriate action to take (please correct and educate me if I'm wrong in my thinking). Since the parity check completed, I also performed an extended SMART test. After restarting the system, I saw a notification that the 'Array has 0 disks with read errors'. That's not necessarily enough for me to move on without thinking something's lurking, waiting to crash everything.

Diagnostics attached. Any and all insight, protips, clarity and information are truly appreciated.

unraid-diagnostics-20261027-1446.zip

Edited 1 hour ago1 hr by leader-intron

Was disk 2 the drive showing the errors?

Its SMART history records recent uncorrectable read errors. However, the extended test completed without errors, and it currently has no pending or reallocated sectors. That is encouraging, but it does not establish that the problem is resolved.

The zero-error notification after restarting is expected because Unraid resets those array counters. The disk’s SMART history remains available.

A read error does not mean Unraid automatically writes bad data into parity. With usable parity and the remaining disks, Unraid attempts to reconstruct the unreadable data and write it back. A correcting parity check separately updates parity when it finds a mismatch. Your last check completed with two sync errors.

The temperatures also need attention. Disk 2 is at 55°C and has previously reached 74°C; disk 3 has reached 70°C. Those historical peaks exceed their reported 60°C operating limit, although we cannot tell whether they coincided with these errors.

Make sure important files have current backups, and check the drive cooling and airflow. Once temperatures are under control, run a non-correcting parity check with Write corrections to parity unchecked.

If read errors return, download fresh diagnostics before restarting and post them here. Recurring errors or a failed extended test would make disk replacement more likely.

@JorgeB I greatly appreciate you writing all that and explaining some stuff. I’ll check the cooling immediately and perform a non-correcting parity check also.

Thanks again for your eyes and help!

Welp, I'm back with troubling news. Disk 2 is showing more errors. See attached diagnostics. The parity check did not complete unfortunately and I only just got around to checking this morning. I'd be interested in understanding the important pieces of data within the archived logs, but first and foremost would be next steps to take. Again, I appreciate any and all insight.

unraid-diagnostics-20261028-0816.zip

Disk2 has failed again, and this time Unraid disabled it. Here is what the logs show:

After the non-correcting check started the disk itself reported these as uncorrectable media errors. There were tens of thousands of them over the next hours, and the disk also stopped responding several times and had to be reset, the connection to the disk stayed stable and there are no CRC errors, so this does not look like a cable or controller problem. Disk2 also stayed at or below 59°C during this boot, which is within its specification, so heat alone does not explain it. The SMART error count went from 13 to 89. This drive does not show pending or reallocated sectors, so those zero values, and the earlier passing extended test, do not mean the disk is healthy. Disk2 was disabled at about 18:37, and the parity check stopped at the same time. The log in the diagnostics ends before that point. A disk is disabled when a write to it fails, so most likely this happened when Unraid tried to write back data it had rebuilt for an unreadable sector. You can ignore the 43 sync errors from that check, because disk2 was failing while it ran.

Disk2 is now emulated. Its contents are rebuilt on the fly from parity and the other data disks, so you can still access them normally. But with a single parity disk, you have no redundancy left. If another disk has a problem before disk2 is replaced, you can lose data.

Next recommended steps:

1. Do not run any more parity checks, and keep array use to a minimum until the disk is replaced.

2. Back up anything important that does not have a current backup, starting with the data on disk2.

3. Browse the emulated disk2 and confirm your files are there as expected.

4. Replace disk2 with a new disk and rebuild onto it. The new disk can be the same size or larger, but not larger than parity (20TB). Stop the array, assign the new disk to the disk2 slot, then start the array to begin the rebuild: https://docs.unraid.net/unraid-os/using-unraid-to/manage-storage/array/replacing-disks-in-array/

5. Keep working on the cooling. Disk2 and disk3 run close to their limit even now, and both reached over 70°C at some point in the past.

I would not rebuild onto the current disk2 or reuse it for data. If anything unexpected comes up during the replacement, post new diagnostics before restarting.

52 minutes ago, JorgeB said:

Back up anything important that does not have a current backup, starting with the data on disk2.

You should not copy emulated disk2 data to other array disks. That would go against

54 minutes ago, JorgeB said:

keep array use to a minimum until the disk is replaced

Backup means to another system, or at least to an Unassigned Device.