Comment 0 for bug 2004485

The new autoinstall installer which deprecated the previously used debian-installer supports the creation of an raid array during setup (in my case raid1).
Also as part of the installer the /dev/md0 device is registered as Physical Volume in LVM, a Volume Group is being created and a Logical Volume is created which will be mounted at "/" for the root filesystem.

Creating such an array works fine and results in a flawlessly working system. After the installation I verified that the array is in sync by issuing "cat /proc/mdstat".

To simulate a failure of one disk I erased one of the disks completely and rebooted the system.

The expected behaviour would be: The system boot's normally since there is still one disk available.

The actual behaviour: The system is stuck during boot displaying the following relevant messages:

"Begin: Mounting root file system ... Begin: Running /scripts/local-top ... Volume Group "VG" not found"
"Cannot process volume group VG"
"Gave up waiting for root file system device"

What helped resolving the issue (before removing wiping one of the disks):
  echo "sleep 60" > /target/etc/initramfs-tools/scripts/init-premount/init-mdfix
  chmod 744 /etc/initramfs-tools/scripts/init-premount/init-mdfix
  update-grub
  update-initramfs -u

I assume the system is trying to find the root file system but this is not yet available since the raid array is taking too long to initialize after one of the disks has been wiped.

After applying the above fix and wiping one of the system booted normally after waiting for 60 seconds. This seems to be enough for the raid to initialize in a degraded state.