This is a lot safer on Linux than Windows, this year. A lot of engineering has gone into making updates resilient.
And Linux hasn’t done the Windows 10 to Windows 11 - black screen for a couple hours, hope you know not to touch it - that we sometimes see.
Linux now has a stronger default permissions model, so it’s a lot harder for user error to break the machine in serious ways, even if they do reboot during a sensitive update.
Linux does do the black screen and hope you don’t touch it, at least OpenSUSE and Fedora do. And that’s a good thing. The “reboot to update is bad” meme needs to die but I digress. I’m skeptical that Linux is more resilient than Windows when it comes to updating but even if it is, Windows automatically rolls back failed updates while Linux will boot you into broken system and expect you to know what to do. Regular people can’t deal with this, even if the answer is a simple as selecting a different entry from the GRUB.
while Linux will boot you into broken system and expect you to know what to do.
But…
even if the answer is a simple as selecting a different entry from the GRUB.
Okay. Yeah. It’s often that simple.
I take your point, but I’ve had my Windows blow itself to hell way more than my Linux has, and putting Linux on relatives machines has been by far the least hassle of the big three, for me.
Anything’s possible. But, they try to make that hard. The system always keeps 2 versions around, the newest one and the previous one, so if you screw up the newest one you can always boot into the previous one. And Bazzite, at least, uses BTRFS which uses copy-on-write, so it’s much harder to corrupt the filesystem. I think the /boot partition is still ext4 though, so it’s possible that if you time it just right you could theoretically mess up your boot partition. Then you’d need to use a rescue USB drive to fix it.
The /boot partition is FAT32 due to RedHat’s stupidity but that’s neither here nor there. The point is that regular users don’t know how to boot into a previous version of the OS. Yes, I know you just have to select it on GRUB but a black screen with a list of kernels qualifies as broken for regular people.
I’d agree that in the current state it’s pretty useless. But, I don’t think it would take too much to make it usable. If the GRUB menu had some basic information on it like: what version is it, when was it installed, has it booted successfully, etc. then I think that would be enough for most people to figure out. Although, I do think that the current Bazzite timeout is way too short.
BTW, on my system /boot is ext4, /boot/efi is FAT32 and the rest mounted at /sysroot is BTRFS.
BTW, on my system /boot is ext4, /boot/efi is FAT32 and the rest mounted at /sysroot is BTRFS.
Your installation is probably quite old. It used to be like that but now the default is mounting the ESP to /boot. The old way makes way more sense to me, btw.
Sort of. In my experience with Windows it gets really annoying. They tell you there’s an update and that you have to restart. If you put it off for long enough and just hibernate your computer, Windows will eventually boot your computer even if it’s “off” to install the update.
With Bazzite and other atomic distros it’s more like it lets you know that there’s an update available for you, and that the next time you reboot you’ll get that update. I personally haven’t ever had it bug me to reboot, but maybe it does that eventually. I don’t know of any Linux distros that would have the nerve to boot your system when it’s off to install an update.
Also, if you don’t want that update, you can “pin” your current deployment and you don’t have to update. Next time you boot you can choose the “pinned” deployment rather than the new one. Normally you wouldn’t want to do that because you’d be missing out on security updates, but if you’ve heard that the newest drivers are unstable, you can definitely choose not to update – or at least not to boot into the updated version.
Also worth mentioning that there are always 2 boot entries, the newest one and the previous one. So, if the newest one does get installed and there’s some issue, you can reboot and choose the previous one. Theoretically you can also roll back to an earlier one from months ago, but I haven’t ever done that.
Until everything breaks because the average user hasn’t bothered updating.
With atomic distros, that updating happens in the background, you don’t have to do anything. It’s like MacOS or Android.
Until everything breaks because the average user held down the power button mid-update because the computer wouldn’t shut down.
This is a lot safer on Linux than Windows, this year. A lot of engineering has gone into making updates resilient.
And Linux hasn’t done the Windows 10 to Windows 11 - black screen for a couple hours, hope you know not to touch it - that we sometimes see.
Linux now has a stronger default permissions model, so it’s a lot harder for user error to break the machine in serious ways, even if they do reboot during a sensitive update.
Linux does do the black screen and hope you don’t touch it, at least OpenSUSE and Fedora do. And that’s a good thing. The “reboot to update is bad” meme needs to die but I digress. I’m skeptical that Linux is more resilient than Windows when it comes to updating but even if it is, Windows automatically rolls back failed updates while Linux will boot you into broken system and expect you to know what to do. Regular people can’t deal with this, even if the answer is a simple as selecting a different entry from the GRUB.
But…
Okay. Yeah. It’s often that simple.
I take your point, but I’ve had my Windows blow itself to hell way more than my Linux has, and putting Linux on relatives machines has been by far the least hassle of the big three, for me.
But that’s just my anecdotal experience.
Anything’s possible. But, they try to make that hard. The system always keeps 2 versions around, the newest one and the previous one, so if you screw up the newest one you can always boot into the previous one. And Bazzite, at least, uses BTRFS which uses copy-on-write, so it’s much harder to corrupt the filesystem. I think the /boot partition is still ext4 though, so it’s possible that if you time it just right you could theoretically mess up your boot partition. Then you’d need to use a rescue USB drive to fix it.
The /boot partition is FAT32 due to RedHat’s stupidity but that’s neither here nor there. The point is that regular users don’t know how to boot into a previous version of the OS. Yes, I know you just have to select it on GRUB but a black screen with a list of kernels qualifies as broken for regular people.
I’d agree that in the current state it’s pretty useless. But, I don’t think it would take too much to make it usable. If the GRUB menu had some basic information on it like: what version is it, when was it installed, has it booted successfully, etc. then I think that would be enough for most people to figure out. Although, I do think that the current Bazzite timeout is way too short.
BTW, on my system /boot is ext4, /boot/efi is FAT32 and the rest mounted at /sysroot is BTRFS.
Yep. That’s what I’m saying, Linux isn’t ready.
Your installation is probably quite old. It used to be like that but now the default is mounting the ESP to /boot. The old way makes way more sense to me, btw.
Nope, it’s maybe a month old.
So like forced updates in Windows?
Sort of. In my experience with Windows it gets really annoying. They tell you there’s an update and that you have to restart. If you put it off for long enough and just hibernate your computer, Windows will eventually boot your computer even if it’s “off” to install the update.
With Bazzite and other atomic distros it’s more like it lets you know that there’s an update available for you, and that the next time you reboot you’ll get that update. I personally haven’t ever had it bug me to reboot, but maybe it does that eventually. I don’t know of any Linux distros that would have the nerve to boot your system when it’s off to install an update.
Also, if you don’t want that update, you can “pin” your current deployment and you don’t have to update. Next time you boot you can choose the “pinned” deployment rather than the new one. Normally you wouldn’t want to do that because you’d be missing out on security updates, but if you’ve heard that the newest drivers are unstable, you can definitely choose not to update – or at least not to boot into the updated version.
Also worth mentioning that there are always 2 boot entries, the newest one and the previous one. So, if the newest one does get installed and there’s some issue, you can reboot and choose the previous one. Theoretically you can also roll back to an earlier one from months ago, but I haven’t ever done that.