Kernel-install generates wrong UUID

So the last update as of 2/4/25 broke my system. I’m not really clear on what that implies for the “official” update date, but that’s when I tried the update. It would no longer bring up the desktop window. It also did not appear to be on the network, i.e. it seemed to be hung up. The strange thing was that it DID get through the initial boot process.

I brought up the install flash, got timeshift working, (don’t forget pacman -S archlinux-keyring !) and restored my system.

Sadly that does NOT set up the boot directories properly even though /efi is a specific restore directory in timeshift. That was disappointing :frowning:

After struggling with the documentation and various web-pages and cursing the fact that I used systemd-boot instead of grub, I finally got the system working.

I will now provide the info as to what I did for future travellers.

first get the right disks mounted as per various web pages about restoring boot on your arch system.

sudo mount /dev/nvme0n1p2 /mnt ← my / partition
sudo mount /dev/nvme0n1p1 /mnt/efi ← why this isn’t already part of / i don’t understand

Do the magic chroot thing

sudo arch-chroot /mnt

Everything from here on out is not that complicated, although it was difficult finding a source that provided the steps in a clear fashion.

bootctl install ← this is in the systemd-boot arch wiki

and here’s the tricky part which is NOT mentioned on the systemd-boot page and absolutely should be. If you don’t do this, nothing will work. Why! oh why! is this not in the systemd-boot page ???

kernel-install ← don’t bother reading the manpage . it’s incomprehensible.

so I reboot and… it doesn’t %$#%!^ work.

I look at the .conf file in

/efi/loader/entries

examine it and … IT HAD THE WRONG UUID FOR THE BOOT DEVICE.

Not only was it wrong, but the uuid it generated was nonexistent.

That’s the bad news. The good news is that I simply substituted the correct uuid which I found by using

lsblk -f ← do this in the regular filesystem, not in the chroot filesystem.

rebooted and … it worked !

yay me :smiley:

meanwhile, I highly recommend that you be very careful about doing any updates right now. Something is severely borked.

1 Like

On EndeavourOS we use a little more convinience:
https://discovery.endeavouros.com/installation/dracut/2022/12/

If you modified the entry in /efi/loader/entries it will likely break again on the next update. This is because the entries are generated by kernel-install.

You need to modify /etc/kernel/cmdline

Then you run sudo reinstall-kernels

1 Like

Thank you !!

This means that /etc/kernel/cmdline was generated incorrectly.

That’s still a mystery.

1 Like

How would that have changed what I did ?

sudo reinstall-kernels

??

i was only hinting on what dalto then continued…