Comment 40 for bug 1489855

After some additional tests, it seems I was a bit too hasty to declare that there exists a regression, as, even though a similar error is displayed as the one from the original bug, the persistent partition appears to be mounted regardless.

For the record however, `/var/log/boot.log` displays the following error while mounting the persistent partition:

--------------------------------------------------------
ln: /tmp/mountroot-fail-hooks.d//scripts/init-premount/lvm2: No such file or directory
mount: mounting /cow on /root/cow failed: No such file or directory
adduser: The user `ubuntu' already exists.
[FAILED] Failed unmounting /cdrom.
--------------------------------------------------------

Also, the introduction of `/casper/vmlinuz$casper_flavour` in `grub.cfg` does add a new hurdle to utilities, such as Rufus, that attempt to add the required 'persistent' keyword to the kernel options, because, of course, the new `$casper_flavour` variable makes it even more difficult to insert the keyword in a location that is not going to cause trouble.

I sure wish persistence on Linux didn't require developers of boot utilities to treat every single release as its own special case, and/or end up to the boot process to throwing error messages that seem to indicate that there exists and issue.

As such, I would strongly encourage to test at least UEFI boot in the following manner:
- One FAT32 partition where the whole content of the Ubuntu ISO has been extracted (and grub.cfg patched to enabled persistence)
- One ext3 or ext4 casper-rw partition following the FAT32 partition (since this has been the longtime recommended way of enabling persistence for Ubuntu)

The above is a a very reasonable way to expect persistence to be achieved for 20.04 LTS, so if it does throw errors, as it currently appears to do, I would assert that there are still some improvements that could be made.