Hi All

I received a warning that one of my array disks had 76 read errors;

So I powered the server off, fitted a new 6Tb disk in a spare slot and successfully precleared it ready to replace this one with errors.

But, now the disk which had read errors is showing no errors...

I am guessing its reset because the machine has been rebooted... is it not showing these read errors now because it hasnt tried to read from the 'failed' sectors, or has the reboot cleared the fault so the disk doesnt need replacing?

If the recommendation is to replace it... what is the correct method? I am guessing;

1. Replace the failed disk with the new one

2. When the machine complains that the original disk is missing, assign the new one and confirm rebuild from parity disk

Is that correct?

Thanks

Read errors are frequently related to connection issues (power or SATA). The best way to test a disk’s health is to run the extended SMART test on it.

You are likely to get better informed feedback if you attach your system’s diagnostics (with everything in the one zip file) to your next post in this thread. It is always a good idea when asking questions to supply your diagnostics so we can see details of your system, how you have things configured, and the current syslog.

20 hours ago, itimpi said:

You are likely to get better informed feedback if you attach your system’s diagnostics (with everything in the one zip file) to your next post in this thread.

Sorry - my oversight... now attached :)

phoenix-diagnostics-20260114-1913.zip

Diags are after a reboot, so we can't see the errors, but in SMART they were logged as disk problem, so recommend running an extended SMART test on that disk.

On 1/15/2026 at 7:46 AM, JorgeB said:

...recommend running an extended SMART test on that disk.

Not sure why UnRAID reported read errors and then shows "completed without error"... unless I am missing something?

I have attached the SMART report too.

phoenix-smart-20260117-2032.zip

12 hours ago, SliMat said:

Not sure why UnRAID reported read errors and then shows "completed without error"

Because there were read errors before, but the SMART test passed, so the disk is OK for now, keep monitoring.

Hi all, I am re-opening this as the same disk has started reporting read errors again... so I am pretty sure its on its way out;

This machine is remote, so after the first time I was getting these errors I installed a new disk and ran Preclear on it. So, this disk is already in the machine ready to be added to the array. Is the correct process to just re-assign disk 4 to the new one, or do I need to unassign the existing disk 4 - reboot and then assign the new disk to start the rebuild? Its a while since I have had a disk fail, so just need to make sure I am following the correct process. Current diags attached :-)

Thanks

phoenix-diagnostics-20260301-0936.zip

Edited March 1Mar 1 by SliMat

I think this is the correct process to follow, but just want to make sure before starting...

It's logged as a disk error, so I would recommend replacing it.

Since that disk is likely failing, I would recommend not running the parity check, just replace the disk.

@SliMat once you have replaced the disk you might want to consider running the extended SMART test on the removed drive. If it fails that it is definitely on the way out, but if it passes it may be OK.

Thanks @itimpi @JorgeB

I have already prepared a new disk, so will just replace it. I would rather just dispose of the disk with warnings as I dont want to be in the position where I lose data by trying to save a few £'s :)

As I cant physically remove the old disk as the machine is in a data centre - can I check the correct steps to replace it?

I was going to stop the array, unassign disk 4 from the Main page, assign the new disk as disk 4 on the Main page and restart the array... is this correct and I think it should warn that the new disk 4 will be overwritten and then start the data rebuild once I accept the warning.

Is this correct - or is there anything I need to do differently?

Thanks

Edited March 1Mar 1 by SliMat

Correct