mdadm raid is broken - please give me advice

Question

I created a mdadam raid array like this under KDE Neon,

sudo mdadm --create --verbose /dev/md0 --level=0 --raid-devices=3 /dev/sdf /dev/sdg /dev/sdh

I take a look at gparted and am surprised to see that sdg, and sdh have a partition with linux file system linux-raid. But sdf doesn't have any partition.

I created a second raid array as well, so I had these,

/dev/md0
/dev/md1

This is completely data that I don't care about as it is 100% a backup drive, but I do need to build a stable place to backup my data.

After a day or so I restarting my machine and ended up with two raid arrays on my machine (I am using webmin to look),

/dev/md126
/dev/md127

Neither of these are legit because they both show no devices in the array. Plus I created md0 and md1, not md126 and md127.

When I run these commands,

sudo mdadm --examine /dev/md126
sudo mdadm --examine /dev/md127

I get no result.

I recreated md0 and md1, but now when I try to mount,

sudo mount /dev/md0 /media/petermc/Security

I get this,

mount: /media/petermc/Security: wrong fs type, bad option, bad superblock on /dev/md0, missing codepage or helper program, or other error.

This,

sudo mdadm --examine /dev/md0
sudo mdadm --examine /dev/md1

Returns,

mdadm: No md superblock detected on /dev/md0
mdadm: No md superblock detected on /dev/md1

About the only other thing I can think of that's important at the moment is that these drives are mostly external drives, plus I moved one of the drives to a different USB port.

Open to any advice available. I am contemplating abandoning this approach, and going back to using the drives separately.

Answer 1

I have pieced together information from google, and have done the setup below in a virtual machine against a fresh KDE neon. Subsequently I have applied this to my server, with good results so far.

When testing in the virtual machine I had some boot issues, so had to revise my setup until this was sorted. Quite simply the issues I had above were because I missed out some important parts of the setup. Most notibly update-initramfs.

My info is based on this with tweaks as required to work with KDE neon,

https://www.tecmint.com/create-raid0-in-linux/

After a clean KDE neon installation I created 3 drives, sdb, sdc, sdd.

First is to create partitions,

e.g. repeat the following on each drive

fdisk /dev/sdb

n (new partition)
P (Primary)
1 (number 1)
Enter (default value)
P (print)

L list
t chose the partition
fd (linux raid auto)
P (print)
w (write)

Create the raid array,

mdadm -C /dev/md0 -l raid0 -n 3 /dev/sdb /dev/sdc /dev/sdd

Check it,

cat /proc/mdstat
mdadm --detail /dev/md0

Create filesystem,

mkfs.ext4 /dev/md0

Mount it,

mkdir /mnt/raid0
mount /dev/md0 /mnt/raid0/

Verify the mount point,

df -h

Add a file to the file system to check it is working,

touch /mnt/raid0/tecmint.txt
echo "Hi everyone how you doing ?" > /mnt/raid0/tecmint.txt
cat /mnt/raid0/tecmint.txt
ls -l /mnt/raid0/

Add an entry into fstab to mount after boot

nano /etc/fstab

The entry in fstab, I created it like this, but upon boot I get an error that defaults is an invalid option,

/dev/md0                /mnt/raid0              ext4    defaults         0 0

This is what worked for me instead based on reading the man pages. This is equivalent to defaults. But for whatever reason defaults didn't work,

/dev/md0                /mnt/raid0              ext4 rw,suid,dev,exec,auto,nouser,async 0 0

Check the mount,

mount -av

Write the raid config to the mdadm config file (this part also differs from the link above, I had to switch to root to do this, and the folder is different),

sudo -i
sudo mdadm --detail --scan >> /etc/mdadm/mdadm.conf

This is when I started to get weird things on boot. The raid array /dev/md0 was not found when booting up which was causing boot delays while it was attempting to initialise. After boot it started to appear as /dev/md126 (which is what I was seeing in my original setup mentioned in my question).

This part is crucial to avoid this,

sudo update-initramfs -u

It's described here https://ubuntuforums.org/showthread.php?t=1764861

As soon as I did that, the boot process works correctly. The raid starts up as /dev/md0 and the mount works. All good.