I have set up a new Unraid array. I am now building parity. While parity is still building, disk 3 shows errors. Building parity continues. Attaching diagnostics.

Is the disk faulty? If so, what's the best way to replace it?

tower-diagnostics-20260219-0823.zip

Disk3 is bad

197 Current_Pending_Sector  -O---K   098   098   000    -    26128
# 1  Short offline       Completed: read failure       90%     32112         31238768520

So bad in fact, that it makes me wonder if you don't have other old, bad disks in your new array.

Do any other disks show SMART warning (👎) on the DASHBOARD page?

Would you suggest I stop parity and try to copy the files over to another disk? Or let it finish first?

No point in building parity when one of the disks is unreadable. Good luck on trying to copy anything from it. You may need to resort to ddrescue.

https://forums.unraid.net/topic/192095-plugin-ddrescue/

4 minutes ago, trurl said:

Do any other disks show SMART warning (👎) on the DASHBOARD page?

33 minutes ago, trurl said:

  4 minutes ago, trurl said:

Do any other disks show SMART warning (👎) on the DASHBOARD page?

No. At least not yet...

I have stopped parity built and now trying to copy the files out. Fingers crossed that this allows me to save some of the data.

After this is done, I will do a thorough smart check on all disks. And once completed, I will rebuild parity.

This may well take a few days. Fingers crossed!

Shows the risk of building parity "fresh". I downsized one disk from 28TB to this faulty 16TB drive. Thought the quickest way to do this was to rebuild the parity from scratch. Should have looked for other ways.

What a drama...

I copied things from the disk and succeeded to recover 99% of all files. But you won't believe what happened then... I dropped the new disk with all the recovered content from around 1m to the floor. And now this disk no longer mounts (for obvious reasons).

Any idea whether something can be rescued now beyond spending a fortune on some forensics company?

26 minutes ago, steve1977 said:

And now this disk no longer mounts

Can you plug it in and get a SMART report?

Maybe you can copy from the original disk again?

Unfortunately, I moved it from the original disk (not copied) :-(

I can plug it in. When connected in usb enclosure, it shows properly, but cannot mount (though "mount button" shows up). When I put it inside the actual server case, it also shows, but doesn't allow to mount, but gives me the option to "format" it. I can try to get a SMART report later (once back home).

2 minutes ago, steve1977 said:

Unfortunately, I moved it from the original disk (not copied) :-(

That is more likely to fail with a bad disk, since it must write to the bad disk to remove the files. And copy would be faster since it doesn't have to write to the bad disk to remove files. And, obviously, the source data is gone now. Lesson learned I hope.

6 minutes ago, steve1977 said:

it also shows, but doesn't allow to mount

Maybe check filesystem will allow it to be repaired so it is mountable.

I checked and cannot run SMART. The log gives me below. I will try "check file system" next.

=== START OF READ SMART DATA SECTION ===
Read SMART Log Directory failed: scsi error data protection error

Read SMART Error Log failed: scsi error data protection error

Tried the repair. Got error below:

root@Tower:~# xfs_repair -v /dev/sdc

Phase 1 - find and verify superblock...

superblock read failed, offset 0, size 524288, ag 0, rval -1

fatal error -- Input/output error

The "fallen" disk was not part of the array, but UD. The linked guide asked me to run the array in maintenance mode. Shall I add the disk to the array as "new config" and then start in maintenance mode? Don't want to make things worse, so checking with you first. Thanks for your help!

50 minutes ago, steve1977 said:

Read SMART Log Directory failed: scsi error data protection error

Read SMART Error Log failed: scsi error data protection error

This is typically a bad disk, but you can try replacing/swapping cables just to be sure.

Other disks work in this slot, so very unlikely a cable issue.

Still any hope to add the disk to the array with "new config" and run some of the check things from the link above (https://docs.unraid.net/unraid-os/using-unraid-to/manage-storage/file-systems/#checking-a-file-system).

Or already reached the stage where I need to move on?

A new config won't really help with a failing disk, if parity is not yet built, you can try to clone it with ddrescue to see if it can recover some data.

https://forums.unraid.net/topic/46802-faq-for-unraid-v6/page/2/#findComment-525075

3 hours ago, steve1977 said:

xfs_repair -v /dev/sdc

This is the wrong command.

For disks assigned to the array, you must use the md device or you will invalidate parity.

For disks not assigned to the array, you must specify the partition, so

xfs_repair -v /dev/sdc1

As you can see, more than one way to get the command wrong, so it is always best to use the webUI. It will get it right.

Thanks. As I cannot mount the disk, I don't have sdc1. See below. Any thoughts?

root@Tower:~# xfs_repair -v /dev/sdc1

/dev/sdc1: No such file or directory

/dev/sdc1: No such file or directory

fatal error -- couldn't initialize XFS library

"Mount" is the wrong terminology. You can't even run xfs_repair on a mounted disk. That is why you start in Maintenance mode if trying to repair an array disk. If it is unmounted as an Unassigned Device, that is as it should be. You can still use the webUI to check filesystem on an Unassigned Device.

Maybe sdc is the wrong device, I was just taking your previous command line as an example. Or maybe partition 1 doesn't exist.

7 hours ago, JorgeB said:

you can try to clone it with ddrescue to see if it can recover some data.

https://forums.unraid.net/topic/46802-faq-for-unraid-v6/page/2/#findComment-525075

Or maybe you could use UFS Explorer to try to recover the deleted data from the original disk.

Mmh... I don't see a partition for the "fallen disk" and that's why there is probably also no sdc1? It shows "format" button. So, maybe it's at a stage that I cannot even run an xfsrepair?

UFS Explorer indeed seems a good choice. From the description, it may help with both the corrupt disk as well as the "fallen" disk.

Any experience whether the professional version will do better compared to the standard version. Price difference is 10X. Also, any idea whether I can run it from my VM. I read somewhere that this is not possible?

So, what I will do:

New disk: Wait your guidance whether anything I can still try with xfsrepair? If not, give it a shot with UFS explorer.

Old disk: Use UFS Explorer. Either standard or professional. Preferably from VM

44 minutes ago, steve1977 said:

professional version will do better compared to the standard version

Standard should be fine and trial version will let you see what it can recover before you buy.

45 minutes ago, steve1977 said:

don't see a partition for the "fallen disk" and that's why there is probably also no sdc1

What do you get from command line with this?

lsblk

20 minutes ago, trurl said:

What do you get from command line with this?

lsblk

No sdc1, just the info below (plus of course all the other drives):

sdc 8:32 0 25.5T 0 disk

20 hours ago, steve1977 said:

dropped the new disk with all the recovered content from around 1m to the floor

I drop my phone all the time without a problem.

But hard drives are extremely precise mechanical devices, and it doesn't seem to be working now. See what you can get from the original with UFS Explorer.

There may still be some hope. I SSH into the server. Although the fallen disk is not even mounted, I can see it and it at least gives me the folder structure. No files though.

Let me try to mount it as second vdisk and see whether UFS can do any magic with it. Otherwise, may go back to the old disk.

2 minutes ago, steve1977 said:

I can see it

How? What exactly are you looking at?