Egregoros

@cjd Ubuntu's default partition setup allocates a 512MB /boot. You tell it "do what you want with the disk" and it'll do that. Then by default it turns on unattended-upgrades. Then it trickles new kernels to /boot but does not remove the old ones and updates the bootloader to boot the new kernel by default, which apparently occurs whether or not /boot is full. Then eventually the user reboots the machine. There are two options on the screen: new kernel, or new kernel plus safe mode. But the kernel it is configured to use does not represent a kernel that is actually on the disk, so it refuses to boot. Ubuntu renders itself impossible to boot. Ubuntu commits suicide if you don't intervene.

I do not know for certain whether this problem persists to the present day but it did occur for years. I don't know who is in charge of QA at Canonical but apparently no one is in charge of telling Ubuntu not to commit suicide.

@p @cjd about the boot partition, 512MB sounds a waste of space, but I think I must be confusing this with the EFI partition, because I made it of 16MB on /dev/sda2 because /dev/sda1 was suddenly "corrupted", and that amount of space was a lot.
Continuing about the topic, you *should* run sudo apt autoremove (or autoclean - I honestly don't remember the difference right now) - oh wait that doesn't remove old kernels I think...
it doesn't, because MX Linux comes with a tool for exactly that, remove old kernels from the boot partition... but isn't the partition where the BIOS checks first? I'm so confused right now.

Anyway, I will use this post to tell you I applied to a job at Canonical once and they made me have an interview with an AI that would skip questions if you wouldn't begin to reply after 3 seconds (each question) -it was so annoying that I let her keep asking to see what were the questions. Mostly about AppArmor and Snap.

@anokasion @cjd

> (can you have it on your Linux main one though?)

Yep! I don't do a separate /boot partition.

> another with the main system ("/") and another for my $HOME

Always good to have a separate /home partition. I usually do a root partition and another partition of the same size so I can update the system on one of them and then keep the other as the upgrade target/old root. So I wanna upgrade, I don't have to turn off the box to do the system update: I just rsync the current root and then chroot into it, do the upgrade, pop the chroot, set the other partition as the new root (bootloader, /etc/fstab), then I can either reboot and be in the new system immediately or I can chroot into it and start sshd and everyone that ssh's in gets the new server.

@p @Suzu @cjd @hakui Yeah, not as unusual in an enterprise setting for making sure a company actually pays for what they are using. But it's a really bad joke for personal stuff.

At least the derivatives aren't maintained by RH in the background like with CentOS before its inevitable death, and don't entirely depend on RH either. All three major ones kinda went their own ways: ABI compat only, full source rebuild, and whatever Larry Ellison is doing. Only one of them needs RH to publish full sources.