Re-build BTRFS array

The /dev/sdd has bad sectors, it does not mean that the RAID is broken.

Your RAID1 system is running MDADM. BTRFS is the file system.

First you need to identify the /dev/sdd faulty disc by serial number. Don't eject the disc now. You may shutdown your NAS server and find the faulty disc (unless you already have this information) by serial number. Reboot without ejecting the faulty disc. The worst thing that could happen would be to remove the wrong disc.

I know how to replace a faulty disc in command line, but not using OMV so I wait for someone else to answer. The disc can be replaced very easily, don't panic.

You can check your device serial numbers in OMV - I don't remember exactly where (not at home currently) but in the Storage tab either under Disks or under SMART -> Devices.

I warn you: it's better to wait for someone explain you how this can be done in OMV, as I don't know OMV very much (see my profile: beginner). You are not in a hurry, a disc can keep bad sectors during months before failing.

I fetched this from Claude AI and verified that it was Okay and made some additions:

To replace a faulty disk in an mdadm RAID array (adapt it to your case):

Please make sure that you understand what you are doing and read carefully.

0. BEFORE EVERYTHING: BACKUP

Make sure you have a backup of your NAS.

An RAID array is not a backup and reconstruction might fail.
So you need a backup.

1. IDENTIFY THE FAULTY DISK

Run: cat /proc/mdstat

Or: mdadm --detail /dev/md0

Note which disk is marked as faulty (e.g., /dev/sdd).

This is not your case as your disc only has bad sectors.

Note the serial ID of the faulty disc.

2. MARK THE DISK AS FAILED (if not already auto-detected)

mdadm /dev/md0 --fail /dev/sdd

3. REMOVE THE FAULTY DISK FROM THE ARRAY

mdadm /dev/md0 --remove /dev/sdd

4. PHYSICALLY REPLACE THE DISK

Power down if hot-swap isn't supported, swap the drive, power back on.

Note the serial ID of your new disc.

WARNING! WARNING! WARNING!

4.1 After powering up find your new disc
It may not be /dev/sdd and might have change naming.

If you don't understand what you are doing, this can lead to a catastrophe.

5. PARTITION THE NEW DISK TO MATCH THE OTHERS

Hereafter we consider that naming of discs has changed, so adapt to your situation

Copy the partition table from a healthy disk (e.g., /dev/sdc) to the new one (/dev/sdb):

sfdisk -d /dev/sdc | sfdisk /dev/sdb

6. ADD THE NEW DISK TO THE ARRAY

mdadm /dev/md0 --add /dev/sdb

mdadm will automatically start rebuilding.

7. MONITOR THE REBUILD

watch cat /proc/mdstat

TIPS:

- Check SMART data on the new disk first: smartctl -a /dev/sdb

- Don't replace more than one disk at a time on RAID5/6 beyond redundancy limits

- Update /etc/mdadm/mdadm.conf after replacement

- Run update-initramfs -u (Debian/Ubuntu) to keep boot config in sync

- On RAID1 boot drives, reinstall the bootloader on the new disk

You may run this command to install grub on the new disc:

grub-install /dev/sdb

update-grub

These steps are not necessary, and some don't apply to OP's situation at all.

A RAID member (disk that is part of an array) can be failed & removed in one single step from the OMV GUI. When that is done, just shut the server down and replace the disk with a good one.

It isn't necessary to copy anything from the failed drive - mdadm will automatically start rebuilding the array using the new disk and rewrite any data and/or parity information.

One question for LkS45 though: RAID 1 implies two drives as a mirror of each other, but your screenshot is showing four drives. Are you running two separate RAID 1 arrays, or just using two additional drives individually along with one RAID 1?

cubemin: the array has not failed (we don't know), only bad sectors. There is no hurry.

I looked at my own server with RAID1 to see how this could be done in the OMV interface.

First, you need the openmediavault-md plugin. Under my system it was not installed by default, so I had to install it using OMV interface or a command line:

apt-get install openmediavault-md

Then the procedure is not straightforward. My RAID system is synchronized and I see no way to remove a disc from the web interface when the array is synchronized (which seems to be your case, you only have bad sectors), so I let other members of the forum answer and guide you.

You probably need to follow cubemin advice : shutdown, replace faulty disc and reboot.

I only issue A BIG WARNING. When you reboot your NAS with a new disc, the naming of discs might change. For example /dev/sdd might become /dev/sda and so on. This is one reason why people might loose a disc when using only command lines. Use OMV interface to identify disc naming based on disc serial ID.

Before everything, make a backup. Good luck.

Dear Macom, how to fail a disc from GUI?

Thanks. It is not very straightforward to select the RAID and click "Remove", as in command line you don't remove the RAID, you only fail a disc.

Quote from R2-D2

It is not very straightforward to select the RAID and click "Remove", as in command line you don't remove the RAID, you only fail a disc.

Good point, but then OMV wouldn't just let you remove the entire RAID while it's in use. That's what the Delete button (the trash can symbol) is for, and you'll notice it being greyed out.

Got it, thanks. This is the "Prohibited" circular icon.