Failing Disk Replacement

[Big Mike]

So... I have a 4TB disk, which is getting write errors, that needs to be replaced. I don't have a spare 4TB disk, but I do have two 8TB disks.

The 8TB disks will be the first >4TB disks in the array, so I need to replace the parity disk first. What i'm proposing to do is replace the parity disk first, rebuild the parity, then replace the failing disk and rebuild from the new parity.

 

Is this a safe thing to do?

 

My concern is that the parity disk will be rebuilt from an array which contains a disk with errors. Will this produce a parity disk which contains corrup data?

 

Or do I need to go buy a 4TB disk and replace the failing disk first?

 

Thanks for reading.

unraid-diagnostics-20240119-1006.zip

Edited January 19 by Big Mike
Added diagnostic data

[JorgeB]

See parity swap, it would be safer, use old parity for the disable disk, you can upgrade later.

[Big Mike]

So...

Stop the array.

Remove the failing disk.

Start the array.

Stop the array.

Put the existing parity disk as the replacement disk for the failing disk.

Put the new disk as the new parity disk.

Copy the parity over.

Restart the array when complete.

Done.

 

I can then upgrade another disk at a later date.

 

Thanks

[itimpi]

Which disk is disabled/failed?   Your diagnostics suggest it is parity2.  If that is the case then you can

• stop array
• assign new 8TB drive to parity2
• start array to rebuild parity2

The parity swap procedure is needed when it is data drive that has failed and you want to upgrade parity and re-use the old parity drive to replace the data drive as a single operation.

 

In fact if the failing drive IS parity2 then you could replace parity1 with a 8TB drive at the same time as replacing parity2 and rebuild both parity drives simultaneously (keeping the old parity1 intact just in case something goes wrong with a data drive while rebuilding the parity drives).

[Big Mike]

The data disk which is getting errors is: WDC_WD40EFRX-68W_WD-WCC4E1YTKARZ (sdo). The array only has one parity disk.

 

I have another array (mirror) which has a SSD which is failing according to SMART tests, but that's another issue.

[itimpi]

16 minutes ago, Big Mike said:

The data disk which is getting errors is: WDC_WD40EFRX-68W_WD-WCC4E1YTKARZ (sdo). The array only has one parity disk

Strange then - the diagnostics do not show that disk11 has failed.  In them it is marked as OK.

[Big Mike]

The disk has not failed, it just get's write errors, so i'm assuming it's on it's way out.

[itimpi]

3 minutes ago, Big Mike said:

The disk has not failed, it just get's write errors, so i'm assuming it's on it's way out.

OK.   Normally a write error makes Unraid disable the drive.   I only see read errors in the syslog.  It might be worth running an extended SMART test on it (either before or after it has been replaced) as failing that is the best indication that a drive needs replacing.

[Big Mike]

Will do, thanks

[Big Mike]

I got disks replaced at the weekend without issue. Thanks for the advice.