Hey there,

 

kinda need help because I maliciously misunderstood the process and just thought the doc was outdated for v7.

Situation:

sdf: 12 TB parity drive

sdb: 6 TB FAULTY drive (1 digit smart errors 187, 197, 198 ), formatted XFS

sdc: 6 TB good drive, formatted XFS

 

Plan: Format new 12 TB drive sdd with ZFS (to try around with that) and replace faulty sdb in the array.

 

Steps taken: Pre-clear sdd, stop array, unassign disk2 sdb, re-assign sdd to disk2, start array -> automatically starts re-build -> stop that, because sdd says "Unmountable: not mounted" due to no FS -> set sdd to ZFS and format in "Array Operations" so it can be mounted and the rebuild started: 2nd fuck up not connecting the dots to what I was doing. I thought it didn't matter because sdd wasn't mounted yet, but unraid formatted "disk 2" with its drive, not just sdd.

 

tl;dr: Stopped rebuild, formatted "new disk" so it can be mounted, restarted rebuild. Data gone.

 

So, had to leave house after starting rebuild, then came back. Rebuild at 50%, started to notice missing files. Wondered why disk 2 didn't show any "size: used" in the Overview, checked data and it is gone. I stopped the re-build and powered down the array. Found this topic and knew I effed it.

 

So what do I do now? 

Old sdb is still showing as active with SMART errors. Finish rebuild and copy all contents to sdd?

Re-insert sdb to array as disk 2, rebuild parity?, and re-swap to new sdd?

 

Any advice appreciated, not gonna touch with without more input. That's what I get for doing this quickly in the morning without properly reading the docs.

Edited March 16, 20251 yr by UnraidUser777
typo

Parity rebuild restores a drive exactly as it was, can't change filesystem without manually copying all the data somewhere else, formatting, then copying back.

If the array is now happy with the new (but empty) drive just mount the old drive in Unassigned Devices and copy data to the new drive hoping it lives long enough.

 

DiskX is all that matters, sdX can change at any time so it's not a proper way to refer to drives.

 

Post diagnostics and Main page screenshots if you need further input.

 

Edited March 15, 20251 yr by Kilrah

14 hours ago, UnraidUser777 said:

Finish rebuild and copy all contents to sdd?

That's what I would recommend, but would run an extended SMART test first, before overwriting the new disk, that way you would also have the option to use something like UFS explorer, if the old disk is really bad.

Thanks for the replies. Rebuild is in progress again.

 

The 8 errors each of 197 and 198 are the result of an extended SMART test after the one 187 popped up.

Thanks for the tipp with USF explorer. Was hoping to get away with an rsync run, but maybe should play it safe. Will look into it.

35 minutes ago, UnraidUser777 said:

The 8 errors each of 197 and 198 are the result of an extended SMART test after the one 187 popped up.

But you need to see if the test passed or failed, or post the SMART report.

Thanks for your help. Attached is the latest smart test report I have for the faulty drive from when I discovered the issue. It is currently offline, I can bring it back online when the re-build is done and test again.

 

 

ST6000VN001-2BB186_ZCT22T73-20250309-0910.txt

Test failed, so the disk really needs to be replaced, but you should still be able to copy most of the data from it, manually or by using, for example, ddrescue.

thank you, marked it as solved. I'll follow the guide you linked.

 

For everyone else: NerdPack/NerdTools is deprecated. Via cli downloaded the package from pkgs.org for slackware64-15.0 with wget on the running instance into /boot/extra, and used "installpkg ddrescue-1.26-x86_64-1.txz". The point of copying it into /boot/extra is so that it is automatically re-installed on the next boot.

Edited March 16, 20251 yr by UnraidUser777

18 hours ago, UnraidUser777 said:

NerdPack/NerdTools is deprecated.

Right, I'll need to update that post.