Linux

Oh, sweet!

In that case, I highly recommend taking a look at some more real-world examples. My original link is just something that makes self-hosting and small jobs more or less thoughtless for me.

Imagine all those config management tools built into your OS, and that's NixOS in a nutshell. There's obviously WAY more it can do if you look into creating your own derivations, or getting into the new-ish concept of Flakes.

Again, though, nixops is the thing that makes me continue to use it, besides just already knowing how to throw together a config in nix's syntax. The nixops tool basically allows you to federate all your systems, tag them, group them, and do anything under the sun with each machine (or several in batches). It's hard to get across in a simple text blurb.

In my case (SaaS), imagine having 10 devs that all want their own dev environment that mirrors production within our VPN, then you need a beta and production environment for each client that licenses the app. Each environment has a couple databases, a few different APIs, some background scraper-type applications, and front-ends for everything. Some of that stuff can live on one machine, some needs to be alone and redundant. You can see how very quickly there's a lot of machines to keep track of.

Now I need to update a couple config pieces to match a new feature in the app itself. Well, all I gotta do is sort out the config, then run a couple nixops command to push to all the dev environments. When ready, do the same for beta, then do it for prod when the fat lady sings.

Being all within one ecosystem, focused on security hardening, is what I really like about it. Hopefully that wasn't too stream-of-consciousness for ya, lmao.

ETA: links, also note that nixops is undergoing some serious changes in the past year. NixOS itself also undergoes changes fairly regularly in syntax as vulnerabilities are addressed and improvements made.