Run updates without me having to worry that “whoops, an update was fucked, and the system is not unbootable anymore. Enjoy the next 6 hours of begging on forums for someone to help you figure out what happened, before being told that the easiest solution is to just wipe your drive and do a fresh install, while you get berated by strangers for not having the entirety of the Linux kernel source code committed to memory.”
As someone who has hundreds of installed programs with tweaks on top of tweaks and hundreds of thousands of files, I always find the suggestion to “just reinstall” beyond laughable.
Windows recovery fails in plenty of circumstances, it’s not a magic bullet. Snapshots are like you can do with btrfs, but that’s not exactly how Windows recovery works.
I have an uncle who will assume anything that takes over 20 minutes has crashed so managed to break his Windows box by continually hard resetting as it was trying to apply a large upgrade.
Even in the most stable distros I’ve had this issue. We had a RHEL 9 server acting as a graphana kiosk and it failed after an update. Something dbus related. I’d love to know why, as it’s been the only failure we ever had but nonetheless it shakes confidence. Windows 11 updates trashed three servers, one to the point we had a to fly an engineer out. My hope is that immutable distros fix this.
You might be suffering from the opposite of survivorship bias: When you work in IT you end up having to fix the strangest shit that reoccurs on certain categories of hardware.
I know for a fact that RHEL 7 just did not like certain appliances by vendors that used it (back in the day). They would regularly break themselves until the vendor put out an update that switched it to a Debian-based custom thing.
Also, all the (thousands of) appliances that use Windows are utter shit so it’s not really a high bar. The vendor just needs to hire people that actually know what they’re doing and if they do they won’t use Windows on an appliance!
Moving to ublue/silver blue has really been a treat for avoiding this. Oh update borked my system time to boot to last update and wait on that one. I personally really want to get a CI/CD running next for my updates to make sure my specific build and collection of software just works the way I want it too.
I had to literally give up on a windows install that worked itself into an update hole, run the update, cant log in, undo the update, it tries to update at night. Endless cycle, no possible fix.
I don’t want to berate you, but just know with enough practice, you’ll be able to fix that linux install. Windows wont let you fix it.
Run updates without me having to worry that “whoops, an update was fucked, and the system is not unbootable anymore. Enjoy the next 6 hours of begging on forums for someone to help you figure out what happened, before being told that the easiest solution is to just wipe your drive and do a fresh install, while you get berated by strangers for not having the entirety of the Linux kernel source code committed to memory.”
Just to provide another data point: I’ve had bad Windows updates render my machine unbootable too.
And then you’re left searching for bullshit error messages and potentially unable to fix the problem regardless of your level of expertise.
sfc /scannow didn’t work? Well too bad, cuz now you gotta reinstall your OS
And Microsoft support that’s in fact clueless fanboys.
… No you just use Windows built-in rollback feature. Which I think even auto-recovers these days of it detects a failure to boot after an update.
Hah! Can someone here chime in and tell me when the slow AF (as in, it can take hours) rollback feature actually worked‽
Who TF is that patient‽ You can reinstall Windows and all your apps in half the time required.
I think it recovered my PC for me twice, and it took about ~10 minutes each time at most. Good luck reinstalling everything in that time lol.
As someone who has hundreds of installed programs with tweaks on top of tweaks and hundreds of thousands of files, I always find the suggestion to “just reinstall” beyond laughable.
Windows recovery fails in plenty of circumstances, it’s not a magic bullet. Snapshots are like you can do with btrfs, but that’s not exactly how Windows recovery works.
Of course not, but it works 9/10 times for most people. Enough so that most people never have to deal with a faulty Windows update.
I have an uncle who will assume anything that takes over 20 minutes has crashed so managed to break his Windows box by continually hard resetting as it was trying to apply a large upgrade.
Spoken like someone who doesn’t do stable releases
Even in the most stable distros I’ve had this issue. We had a RHEL 9 server acting as a graphana kiosk and it failed after an update. Something dbus related. I’d love to know why, as it’s been the only failure we ever had but nonetheless it shakes confidence. Windows 11 updates trashed three servers, one to the point we had a to fly an engineer out. My hope is that immutable distros fix this.
You might be suffering from the opposite of survivorship bias: When you work in IT you end up having to fix the strangest shit that reoccurs on certain categories of hardware.
I know for a fact that RHEL 7 just did not like certain appliances by vendors that used it (back in the day). They would regularly break themselves until the vendor put out an update that switched it to a Debian-based custom thing.
Also, all the (thousands of) appliances that use Windows are utter shit so it’s not really a high bar. The vendor just needs to hire people that actually know what they’re doing and if they do they won’t use Windows on an appliance!
That’s why I make a btrfs snapshot of my system before every upgrade. Rolling back from a rescue image takes only a minute.
What a great idea! They should automate something like that! Maybe they could call it System Restore?
I never claimed to have invented the technique.
They’re just pointing out that Windows does this too.
Moving to ublue/silver blue has really been a treat for avoiding this. Oh update borked my system time to boot to last update and wait on that one. I personally really want to get a CI/CD running next for my updates to make sure my specific build and collection of software just works the way I want it too.
I had to literally give up on a windows install that worked itself into an update hole, run the update, cant log in, undo the update, it tries to update at night. Endless cycle, no possible fix.
I don’t want to berate you, but just know with enough practice, you’ll be able to fix that linux install. Windows wont let you fix it.