mdadm --zero-superblock for all devices of a mdraid?

If you know the array component devices, you can simply issue something as

mdadm --zero-superblock /dev/sd[abcd]

to remove the superblock from multiple devices with a single command.

Please note that mdadm checks for superblock before zeroing anything, so specifying sd* in the command above should only touch component disks, without writing anything to the others. However, I strongly suggest against doing that: any issues in the superblock detection routing, or using --force, are going to overwrite good data on non-component disks (causing data loss). From the man page:

If the device contains a valid md superblock, the block is overwritten with zeros. With --force the block where the superblock would be is overwritten even if it doesn't appear to be valid.

For the reason above, if you don't know the array component devices I suggest listing them via

blkid -t TYPE="linux_raid_member" -o device

and then zeroing the specific disks with mdadm --zero-superblock. This can be scripted with a single-line bash command:

for dev in `blkid -t TYPE="linux_raid_member" -o device`; do echo "zeroing $dev"; mdadm --zero-superblock $dev; done

As always, triple-check any such commands to avoid data loss.

EDIT: to discover the array component devices without resorting to parsing of mdadm --detail output, you can use lsblk. For example, on a test array created via loop devices:

# lsblk -l -n -s /dev/md127
md127   9:127  0  255M  0 raid1
loop5   7:5    0  256M  0 loop
loop6   7:6    0  256M  0 loop

There isn't a single command (mdadm …) that will remove the superblock from all devices of a mdraid. However, it's fairly straightforward to write a script that then becomes a one-liner:

#!/bin/bash
md=$1
[ ! -b "$md" ] && { echo "ERROR: Not a block device: $md" >&2; exit 1; }

# Collect the slices
slices=($(mdadm --detail "$md" 2>/dev/null | awk '/active/{print $NF}'))
[ ${#slices} -eq 0 ] && { echo "ERROR: No slices found for $md" >&2; exit 1; }

# Sanity check
if [ -t 2 ]
then
    echo "Active MD RAID $md has slices: ${slices[*]}" >&2
    if [ -t 0 ]
    then
        read -p 'Stop and erase (Y/n)? ' yn
        [ ! "$yn" = 'Y' ] && exit 1
    fi
fi

# Stop the array
echo "Stopping: $md" >&2
if ! mdadm --stop "$md"
then
    echo "ERROR: Cannot stop: $md" >&2
    exit 1
fi

# Erase the slices
for slice in "${slices[@]}"
do
    echo "Zero: $slice" >&2
    mdadm --zero-superblock "$slice"
done

# All done
exit 0

Example usage, assuming the script is called mdzero and it's in the PATH, etc. and the unwanted RAID device is /dev/md1:

mdzero /dev/md1

Active MD RAID /dev/md1 has slices: /dev/dm-2 /dev/dm-3 /dev/dm-4
Stop and erase (Y/n)? Y
Stopping: /dev/md1
mdadm: stopped /dev/md1
Zero: /dev/dm-2
Zero: /dev/dm-3
Zero: /dev/dm-4

Unless you enter a (capital) "Y" in response to the "Stop and erase?" question - or the command is being run non-interactively - the tool will quit at that point. Essentially, you've got 20+ lines of sanity checking to ensure that the active mdadm --zero-superblock command is only run if and where absolutely intended

You can use blkid to do this with a little bit of manipulation.

First, get the raid array you care for.

mdadm --detail /dev/md0
...
UUID : 8c9c9aa9:0c574535:e9080607:7ac4c6d8

Convert the UUID Into a a proper UUID format and you can use blkid to pull out all disk members.

blkid -t UUID=8c9c9aa9-0c57-4535-e908-06077ac4c6d8 --output device
/dev/nvme1n1p3
/dev/nvme0n1p3

If you just want to blast all mdadm raid arrays you dont need to collect any other identifier:

blkid -t TYPE="linux_raid_member" --output device
/dev/nvme1n1p2
/dev/nvme1n1p3
/dev/nvme0n1p2
/dev/nvme0n1p3

To blast everything using this method you can even use wipefs. That could look something like..

blkid -t TYPE="linux_raid_member" --output device | xargs wipefs -n

Of course, this will zero all raid arrays of all disks without caring about the particular raid array in question. I added a -n to prevent people being careless with their copy/paste!