Snapper and kernel mismatch

Snapper only snapshot / subvolume, leaving /boot ext4 partition.

Let’s say I wanna create snapshot called “backup”, when I take this snapshot, the system is currently using (let’s say) Kernel 1.

20 days later, my system got kernel updates 10 times and now is running with Kernel 10 and Kernel 1-5 has been cleaned up for both on / subvolume and /boot ext4 partition.

I got a problem and need to rollback on to the snapshot that I took before (“backup”); when I rollback, my / subvolume will use Kernel 1 but on /boot ext4 partition it still uses Kernel 10 so it will triggers Emergency Mode and can’t continue.

Wanna use Kernel 1 ? Sorry, that Kernel didn’t even exist anymore on /boot ext4 partition due to cleaning purpose.

So if that happens, what should I do?

I believe Fedora’s motto is: “if you want bulletproof rollbacks that include the kernel, use Silverblue. If you use Workstation, we prioritize boot compatibility over snapshot ease-of-use”.

That said, you can look into some practical options:
Don’t use a separate /boot partition. Keep only /boot/efi (FAT32). Let /boot be a normal folder inside your Btrfs root (snapshots now include kernels).

Manual: boot live USB > chroot > reinstall kernel - dracut regenerate/force
This is more reactive not preventive..

It’s not ideal and there are pro’s and cons.

1 Like