Linux

chown 777 everything

The only time was within a VM. I accidentally wrote

rm -rf ./* while my cwd was /

I use absolute paths with -rf now, to prevent the error again.

Every other breakage I had was with apt shitting itself. It has always been fixable just annoying.

I now use Fedora, to prevent the error again.

sudo apt upgrade -y

To this day I can’t figure out why it killed the GUI and all terminal commands on a Mint install…

I can't remember what I did to break it, but back when I was in high school I was tinkering right before class and rendered my laptop unbootable. I booted into an Arch Linux USB, chrooted into my install, found the config file I messed with, then reverted it. I booted back into my system and started the bell ringer assignment as quickly as I could. I had one minute left when the teacher walks by, looks at it, and says that I did a really good job. She never knew my laptop was unbootable just 2 minutes earlier.

I was running Fedora. Something like 27 or so. I needed drivers. I don't remember if it was AMD or Nvidia, but they were only available on RedHat.

So I downloaded the RedHat drivers for the GPU and forced it to install. It worked! It was great.

Then when I updated the distro to the next release... everything failed. It was dropping into grub, but no video was output. Ooof.

So I ended up enabling a terminal console and connecting to it via a serial port to debug. I had to completely uninstall that RPM and I was never happy that it was properly gone. So a few months later I ended up reinstalling the whole OS.

On the plus side, I learned a lot about grub and serial consoles. Worth it.

I mistyped my SU password when setup the OS...

Actually, I have a story that I'd consider an achievement even though it was extremely stupid and by all accounts should've bricked the system but didnt.

So I was on windows and wanted to install linux as a dual-boot on the main drive. The problem was that my mobo didnt like this particular and the only flash drive I've had, dropping it out mid-boot, before I got any usable terminal, so a usual install method wasn't an option. So I had this crazy idea to start a vmware vm in windows and pass the linux iso and the boot drive directly to it and try to install it live over the running system. Unfortunately, vmware guys thought of this and there's a check that disallows passing the boot drive to vms. So i created a bunch of .vmdks for another drive and fiddled with them in notepad until I somehow managed to trick vmware and at some point it started booting the same windows copy that I was sitting on. I quickly powered it off, added the linux iso and proceeded to install like I usually would. It did involve some partition shuffling, but, somehow, it went smoothly, linux installed, grub caught on, and even windows somehow survived, even though it was physically moved around on the disk. It serms that vmware later patched this out, because later in an attempt to re-create the trick of running the same copy of windows twice, but after updates to both windows and vmware, I was met with the same old error that boot drive is not allowed when trying to add that same virtual drive I had laying around.

I suppose it doesn't quite qualify as breaking the system in a funny or stupid way but it certainly was one of those stupid things that was easy to fix after a ton of trouble shooting, ignoring the issue for a while and trying to fix it again.

So i had an old pc where I had a failed hard drive which I replaced. Obviously I also accidentally unplugged my optical disc drive and plugged it back in. Now that failed drive was just a data drive so the system should have booted up no problem since the os was on a SSD but instead it got a kernel panic and got stuck at boot. Since it was late I left it at that and came back to that the next day where it would still not boot. So I unplugged the disc drive and looked up what it could be. Tried a ton of different possible solutions but every time I added that disc drive it would panic.

I eventually kind of gave up and just didn't use that disc drive at all and just had it as a paperweight in the system. Unplugged and all that. When my replacement SSDs for my old data drive and backup drive came in I tried again to get that optical drive working but to no avail. So I unplugged it again, got it all set up and ran into another issue where for some reason Linux couldn't properly use my backup SSD. So I investigated that as well and trough some miracle found a post on the forum from my Mainboard manufacturer... Turns out that particular Mainboard had a data retention chip on it that didn't like Linux.

So naturally I just plugged everything into the data ports that were not controlled by that chip and it all worked as intended.

Stupid dumb chip on a Mainboard, all I had to do was try the simple idea of unplugging and trying a different connector but instead I did all that other stuff first that didn't work and cost me so much of my time.

Moral of the story, when in doubt try and put stuff on different connectors and see if that fixes it. Might just be a dead connector for all you know. Or an incompatible chip on the Mainboard.

FWIW I bought that Mainboard long before I switched to Linux and didn't plan at all to switch at the time. But that's a different story.

I was testing a custom initramfs that would load a full root into a ramdisk, and when I was going to shut down I tried to run rm -rf --no-preserve-root / to see what would happen, since I was on a ramdisk anyway. The computer would not boot after that because it nuked the UEFI options.

Not really a "braking my linux setup", but still fun as hell! Back in university, a friend of mine got a new notebook at a time... we spent the night at the university hacking and they wanted to set the notebook up in the evening. They got to the point where they had to setup luks via the cryptsetup CLI. But they got stuck, it just wouldn't work. They tried for HOURS to debug why cryptsetup didn't let them setup LUKS on the drive.

At some point, in the middle of the night (literally something like 2 in the morning) they suddenly JUMPED from their seat and screamed "TYPE UPPERCASE 'YES' - FUCK!!!"

They debugged for about six hours and the conclusion was that cryptsetup asks "If you are sure you want to overwrite, type uppercase 'yes'". ... and they typed lowercase. For six hours. Literally.

The room was on the floor, holding their stomach laughing.

Just yesterday I overwrote some pacnew files and borked user authentication for myself. Very rough time

and a need to find another PC to flash an archiso to a flash drive ('cause ofc I didn’t have one at the time).

you can do that from your phone using etchdroid

i don't remember ever breaking my system in a terrible way, but when i started using linux (with linux mint) i uninstalled ca-certificates and i think that uninstalled the whole DE

I somehow locked myself out of sudo when trying to give my user permission to read serial devices.

Had to reinstall.

Wanted a cool bootscreen on my Nixos machine - commented out the bootloader to troubleshoot, why my meme-boot-picture wouldn't show - after rebooting, it loaded straight into the BIOS and finally realized what I had done... Was able to fix it thankfully

I wanted to move my Arch VM to bare metal, so I copied out all the important bits. Then I wanted to move that copy to a new drive so I could boot into it.

I THOUGHT I'd MV all the files in the Arch install's etc directory using sudo MV /etc ...

I also (somehow) mashed my install's etc with Arch's and bungled both, with no live CD to help.

I learned a thing or two about absolute file paths...

Before I understood how to properly build and test mesa (graphics driver), I compiled it and then procedeed to manually symlink the files in the lib and lib32 directories. When I pressed enter on that ln command, the UI immediately crashed and X would no longer start after rebooting the computer. Reinstalling mesa from a virtual terminal wouldn't fix it so I just reinstalled the system. Good times :)

I had rEFInd and GRUB installed entirely by accident, and a botched update for Arch hosed my entire EFI setup making it impossible to boot Linux or Windows w/o a LiveCD. Thankfully it self repaired once I nuked rEFInd. I ended up going back to Ubuntu, but I hate snaps. I still would recommend Arch for most Linux users who want the power windows.

Somehow I found ways to remove and break the GUI multiple times in multiple ways in multiple distros.

Different scenarios, different times, different issues trying to "fix". My usual fix after this was always to copy what I think I still had important and then move on with a reinstall.

Recently I have been playing with ZorinOS and broke it in the same way by fidgeting with pipewire. Distro hoped to Fedora Silverblue due to the immutable filesystem. I wonder if I will break this one in a way I cannot revert it easily with rpm-ostree. I almost feel challenged.

I deleted the entire taskbar.

Renaming a mount point while mounted was a fun experience in losing data back in the big box Redhat 5.0 days.

Can't say I have any interesting stories. Most of mine are just the head-scratching "I don't know why that didn't work; guess I need to reinstall" kind of story. Like enabling encrypted LVM on install and suddenly nothing is visible to UEFI. Or trying to switch desktop environments using tasksel and now I have a blank screen on next reboot. That lame kind of stuff.

My coworker though... he was mindlessly copy/pasting commands and did the classic rm -rf $UNSETVARIABLE while in / and nuked months of migrated data on his newly built system. He hadn't even set up backups yet. Management was upset but lenient.