This was the way. Then you find Debian.
I switch distro once I start feeling that my current installation is too bloated and requires a heavy cleaning
Which is why I switched to nixos, so that I can’t bloat my system up with packages I eventually forget about
NixOS is so incredibly stable it’s crazy. Even if my entire computer implodes I can just download my couple config files off github and get exactly the same system on a different computer.
Well, you still need to backup and restore your persistent drive, but that’s trivial too.
Yeah, I use Impermanence and all my important things and dotfiles are synced between my devices. Other stuff is just games and stuff I can reinstall anytime.
I’m going to try Nix as my desktop OS. The only thing stopping me up until now is I like running the same OS that I run on servers (Debian). Do you think there’s a good use case for Nix on servers?
Yeah NixOS is great for servers, since you’re able to configure everything through the NixOS configs. Like if you want nginx you just add services.nginx.enable = true and similarly set the different virtualHosts and everything. That way your nginx configuration is stored in the same place as your system configuration, which can all be backed up with Git, and you can see everything running on your system and their configuration by just looking through your NixOS config.
I reinstalled Linux when it crashes, or used Timeshift for years, but at this time I learned totally nothing.
Then I tried Arch manual installation, and it changes my mind.
The fresh feeling of a reinstall lasts for about a week.
Do a snapshot and roll back. Actually faster and easier.
This was in the long long ago, grasshopper. We did bare metal installations back in the day.
I am 5 years in, and that is still what I do most of the times
When I decided to switch to Fedora, I wanted a safety net. I had a 500GB SSD, so I bought an additional 2TB SSD, so I could make full disk image backups and be able to store 3 of them (I used full disk encryption, so my disk image backups were the full 500GB). And I dutifully made backups, either monthly, before I made a big change, or before a major update. Been doing this for nearly two years now and I haven’t used a single backup image even once. It’s almost disappointing, in a perverse sort of way. I was looking forward to having to learn stuff by fixing things that break, but nothing ever does!
Cool you did backups
BTRFS is your friend guys and gals ☺️.
Yup. Being able to run my home and root(s) in separate subvolumes, and simply booting into a specific root with a kernel parameter… 😌
Snapshots 4 life
Mhm, pretty much. I haven’t reinstalled since I started using snapshots.
I switched to BTRFS recently, but found myself even more fucked when my system stopped working suddenly and I didn’t know how to fix it without reformatting and installing grub again. Actually lost even more than I would have otherwise just because I wasn’t knowledgeable enough to get any form of recovery to work. That first EndeavourOS install didn’t last 2 months sadly.
Yep, everyone goes through that the first 2 or 3 installs, until you learn how CoW FSes work. It’s not like anything else and it takes a while to master it, but once you learn how to use it, you don’t reinstall ever again, just roll back snapshots 😉.
I’ve lost so much data to btrfs disk mirrors. Zfs is my friend now.
Meeh, anything that is CoW and has snapshots will do the job, ZFS or BTRFS, whatever rocks your boat 🤷.
I agree cow + snapshot is pretty useful. I would just never use btrfs for data I care about. There is a reason no one sane runs it in production. Your computer and data do what you want 😊🙂😊.
It’s being used in Meta 🤨… in production.
Cool I had no idea. I like zstd from them. I don’t really want to argue if it works for you that’s great. I’ve seen so many problems with corruption that I wouldn’t recommend it. I guess I’ll give it another try in a VM some day. I really tried to move to it before migrating back to zfs land. I do recall the send and receive working pretty flawlessly. Also was a huge fan of duperemove.
Do you know if it has support for something like zvols yet?
Yes, it did have problems a few years ago, especially regarding RAIDs, but it’s improved a lot since then. RAID5 still sucks though 😁… but I read the problem is finally been worked on (haven’t checked code, I read about it in a sub on reddit).
No, it doesn’t have something like zvol, it has the regular subvolumes (pools in ZFS) and you can assign quotas, the same as in ZFS. But, to represent itself as separate block device, no. And I don’t think this is something that’s planned, though I could be wrong (as I said, I haven’t looked at their git in ages).
Lol this is still me after 20 years of using linux
I haven’t properly dotfilesed all of my rice yet, so I’m just hoping l don’t break something until I get that sorted.
Then there’s the cloud: “Oh, crap. I have a typo in a config file. I guess I’ll destroy the machine and set up a whole new one!”
I did this without having my distro broken. It was like “oh shiny, let me try this distro”
Timeshift makes all the difference, no more panicking after breaking something.
As a noob, I wonder what would be the right way to do it?
