My crippled kernel count is around 6, how about yours?
The “starting over” part is what made it take so long for linux to “stick” with me.
Once it became “restore from an earlier image”, it was a game changer!
I could be weird for this but the starting over part actually contributed to me continuing to use linux tbh. Trying out a new distro, figuring out how to use it, and building a new user interface each time I killed my system kept me engaged with linux beyond its utility. It functioned essentially as a way to learn about computers and as a creative outlet. I don’t fuck around and find out as much as I used to but I still swap distro every year or so.
It was similar for me, but not quite the same. The thing I hated was starting from scratch. I’m very much not a distro hopper. Back in the day, I enjoyed the challenge of trying to troubleshoot issues and get the system working again, and that kept me interested, but eventually, I’d hit a problem I couldn’t resolve, and I’d have to start again from scratch, and at that point, I’d just go back to Windows.
Now, I still get to do the same thing. If I break it, I get to learn how I broke it and try and fix it, and I find that process compelling. But because I’m using btrfs restore points now, I don’t get to the point where I have to start again from scratch. So I can work at solving it to the limit of my abilities, with confidence that if I can’t work it out, it’s not a huge issue.
“Starting over” is how we learnt Windows in the 90’s too
Giving our computer ghonorrea by downloading Napster mp3s
I’d just re-install Windows over the top of the fucked up install normally. It was a bit easier to recover from, and a bit harder to fuck up
Tell me more
Timeshift was a gamechanger
Timeshift itself borked my shit up. I had to reinstall all registered packages to fix its fuckups…
sudo aptitude reinstall '~i'Edit: Sure it took a long while, about as long as a full OS reinstall, but never once was there any issue with the kernel.
While only once, timeshift destroyed my bootloader. Don’t update and reboot before a meeting, kids
My test of Timeshift was pretty simple and straightforward.
-
Fresh install Linux Mint
-
Install most of the main software I wanted.
-
Do a Timeshift backup.
-
Install some extra software I didn’t necessarily need, but might want to use someday.
-
Restore the backup from step 3.
Results: Everything from step 4 was still registered as installed, but almost nothing from step 4 actually worked.
So I brute force reinstalled everything in place, and haven’t used Timeshift since. I’m perfectly comfortable using the terminal, and at worst a live boot media, to fix any issues that might come up.
Timeshift is a good piece of software doing a tired trick.
The new hotness is copy on write file systems and snapshots. I can snapshot, instantly, then do a system update and revert to the previous snapshot also instantly.
Instead of using symlinks files, like Timeshift, the filesystem is keeping track of things at the block level.
If you update a block it writes a new copy of the block (copy on write). The old copy is still there and will be overwritten unless it is part of a snapshot. Since the block is already written, snapshots don’t require any data to be copied so they’re instant.
Once you finish the system update, all of the overwritten blocks are still there (part of the snapshot) and reverting is also just a filesystem operation, theres no mass data to be copied and so it is also instant.
It does use disk space, as allocated blocks AND snapshotted blocks are stored. It uses less than Timeshift though, since Timeshift copies the entire file when it changes
ZFS and btrfs are the ones to use.
Didn’t quite follow what you were saying completely. Are you suggesting a new program over time shift or change the file system type like ZFS and Btrfs? I’m using Ubuntu and not sure if I seen those before.
-
Uhm, zero? With ten years of using Linux? What did you do to fuck up the damn kernel? o_O
It can be done if you mess with the initramfs.
The kernel starts everything else by unpacking an archive containing a minimal environment to set stuff up for later. Such as loading needed kernel modules, decrypting your drive, etc. It then launches, by default, the /init program (mines a shell script).
That program is PID 1. If it dies, your kernel will panic.
After it finishes setup, it execs your actual /sbin/init. These means it dies, and that program (systemd, openrc, dinit, runit, etc) becomes PID 1. If an issue happens, both could fail to execute and the kernel will loop forever.
Thank you for explanation :) I suspected something like that - mess up with some internals, you do have a chance to bring the thing down. Which is why I always have a bootable usb around before doing anything risky
OpenSUSE Tumbleweed helps because you can create a btrfs snapshot at any moment and then roll back to it if you get in trouble. And it does this automatically whenever you update the packages.
I wanted to give OpenSuse Tumbleweed a go yesterday, but the live USB got stuck at “Loading basic drivers” so I couldn’t even get to being able to install it.
It’s the same as learning anything, really. A big part of learning to draw is making thousands of bad drawings. A big part of learning DIY skills is not being afraid to cut a hole in the wall. Plan to screw up. Take your time, be patient with yourself, and read ahead so none of the potential screw-ups hurt you. Don’t be afraid to look foolish, reality is absurd, it’s fine.
We give children largess to fail because they have everything to learn. Then, as adults, we don’t give ourselves permission to fail. But why should we be any better than children at new things? Many adults have forgotten how fraught the process of learning new skills is and when they fail they get scared and frustrated and quit. That’s just how learning feels. Kids cry a lot. Puttering around on a spare computer is an extremely safe way to become reacquainted with that feeling and that will serve you well even if you decide you don’t like Linux and never touch it again. Worst case you fucked up an old laptop that was collecting dust. That is way better than cutting a hole in the wall and hitting a pipe.
So this is why I’m bad at drawing. I have 954 more drawings to go!
It’s even better if your only internet connection is that computer you broke.
Ah yes, reminds me of messing with my 1st pfSense firewall… I learned how good their recovery process was that evening
Just bricked my Proxmox install an hour ago and I have the pleasure to learn their recovery process sucks. (At least for my case)
Ah, yeah, you’ve just reminded me, I must move my stuff off proxmox when I get a chance.
I tried that proxmox backup thing when I first set it up, good god what a complex mess… backup & recovery needs to be as simple and as smooth as possible.
This is the nightmare of my last 2 weeks. Well 11 days.
Great incentive to learn even faster
And enforces the value of installing documentation and source packages 😅
I’m not sure I’ve ever actually killed a system, I’ve booted from UEFI shell manually just to recover systems. Back when I was using arch id just chroot into the system from a flash drive and fix whatever ¯\_(ツ)_/¯
This is the way!
No no no! When you break something in Linux systems you fix it. Starting over and reinstalling everything is what you do when you mess up on Windows.
Funny I did not expect so many people that resist starting over. Next time I’ll give fixing stuff a shot :)
It is more about being lazy.
In most cases, where you havn’t destroyed your filesystem, you can just boot another Linux from a USB stick, mount your filesystems to /mnt, chroot into it, and then investigate and fix there.
See the Archlinux wiki, even if you do not use Archlinux, it is great: https://wiki.archlinux.org/title/Chroot
Generally yes. My exception was the time i accidentally nuked python in it’s entirety…
Well, that could have been fixed by booting from an usb stick, chrooting into you real system and either downloading and (re)installing the python package this way, or, if your package manager depends on python, download the package in the Live Linux and extracting the python package into your system, and then reinstalling it, so the package management overwrites your “manual installation”.
Could be tedious, but less so that having to reinstall everything IMO.
Fair, unfortunately it was a work machine that i needed operational again asap.
Luckily i image my machine monthly, so it was fairly straightforward to roll back.
Recently I accidently deleted the contents of /boot/ on my first arch install. The lesson that followed was something I would have rather saved for later ^^
Pretty much everytime I try to do fancy stuff with the bootloader I get pretty close to nuking systems. Worst was my 1st UEFI system where I was trying to get rEFInd to show multiple OS to boot from… eventually gave up and went back to the warm embrace of GRUB
If you take the plunge and switch to systemd-boot it’s worth it. It’s the only boot manager I’ve tried in the last decade that feels like an upgrade from GRUB.
🤔 Maybe I’ll try that next time… I kinda feel loyal to Grub, it’s been my friend for sooo looong.
I just had 8 titles in boot menu all for the same OS. 🤌😅 I know exactly what I’m doing. It’s a dual boot system.
I tried to use dd with too much hubris once. I had to restore from backups (which ironically, I had made with dd). I’m usually overly cautious, but I was in a hurry.
I did this one a few weeks ago lmao. You think once would be enough. But I am a truly special being.
I haven’t majorly fucked up any recent systems (almost botched the steam deck once or twice but nothing that required a reinstall), but god 10 years ago I probably reset my arch dual boot like five times lmao
Maybe 1 or 2 back when things were less stable, but any time I have used Linux in the past 7 years or so, and particularly since I started using Debian as my primary OS, I haven’t had any problems outside of trying to get some windows applications to emulate correctly, and one time when I echo’d into sources.list with > instead of >>. Anything else is just stuff I had to learn, like my boot folder filling up with old images that have to be cleaned out occasionally.
If you want shit to just work when you want and stay out the way when you aren’t using it. Debian of whatever source is what they call stability. I’ve done rolling, and bleeding edge. It’s all a constant pain. Becomes a job to maintain or bug track or check logs. I’ll never go back.
That was my thought as well.
Back when I was new to Linux, I tried a lot of different distros in virtualization for shorter periods of time, and of course ran into the issues that come with the cutting edge stuff.
Last year I wanted to install a distribution to my laptop properly as a test before putting it onto my desktop, and I came to that same conclusion because at the end of the day I couldn’t justify using bleeding edge, because I couldn’t really even name anything I NEEDED from it. Yes, it is fun to have cool, new things, and it can be a lot of fun to play around with in a VM or something, but I don’t actually need any of that stuff for what I do on a computer day to day right this second.
After that, the answer was pretty clear for me as to what distribution to use.
i broke debian on my plex server and said fuck it and migrated to endeavor because im more familiar with arch
Both, to the point it doesn’t boot, and just tweaking enough bugs that it’s easier to jist start over.
Reply fail?
1
