My first home server would get lost on the network every week, at different times and without any apparent reason. I performed hard resets by unplugging and plugging it back in.

After several months, I decided to connect a screen to it, and I initially thought it had hung up, but it hadn't. After some investigation, I discovered that every time my router obtained a new dynamic IP address, the server lost its network connection, requiring a reset. I wrote a script to check the network connection every minute, and if it's lost again, it will be reset.

I managed a CentOS system where someone accidentally deleted everything from /usr, so no lib64, and no bin. I didn't have a way to get proper files at the time, so I hooked the drive up to my Arch system, made sure glibc matched, and copied yum and other tools from Arch.

Booted the system, reinstalled a whole lot of yum packages, and... the thing still worked.

That's almost equivalent to a reinstall, though. As a broke college student, I had a laptop with a loose drive, that would fall out very easily. I set it up to load a few crucial things into a ramdisk at boot, so that I could browse the web and take notes even if the drive was disconnected, and it would still load images and things. I could pull the cover off and push the drive back in place to save files, but doing that every time I had class got really tiring, so I wanted it to run a little like a live system.

I feel seen here, I was building a Ubuntu server and messed up the firewall settings not being able to get an internet connection, hours of trying to get back to where I was I gave up and plan to just start from scratch next time.

Is there a way of taking system snapshots with Linux?

Jumping from the default kernel with zfs to the xanmod kernel using a manually compiled version of zfs. I don't rememeber a whole lot but it was quite... interesting. Next would be a suddenly vanished efi partition and my f* mainboard refusing to boot ZBM.

Bonus: my currently still unfixed problem is a very weird freezing/stuttering of the whole OS and the only (useless) "lead" I have is workqueue: fill_page_cache_func hogged CPU for >10000us 4 times, consider switching to WQ_UNBOUND

Fast data transmission via TCP over a lossy link.

I recently managed to recover from a corrupted libstdc .so. Turns out I shouldn't have bothered because the it was a Pi and, of course, the SD card had shit the bed, but I was pretty happy with myself for like 30 minutes.

Accidentally put grub on the wrong partition on the device, which it was not happy with. Was able to copy some files over, manually boot the OS, and reconfigure grub to be in the right partition, took me about 2 hours? Then I did it again on a different machine, and speedran it lol