💾 Archived View for ebc.li › posts › why-not-automate-systems.gmi captured on 2022-01-08 at 13:50:44. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2021-11-30)
-=-=-=-=-=-=-
2021-04-28
I want to spend as little as effort as possible, so I can get on with using the thing, instead of configuring and debugging for a day straight beforehand.
No matter how great your tools are, some software will just not cooperate. And trying to work around that is not fun.
When I tried automating my server previously, I had times where I just gave up and wrote what I did in a separate file to "read later on". I never did, because...
I have this thing of very frequently re-formatting and starting from scratch on a lot of my machines. I don't think I had an OS install that went for more than a year since, like, 2015.
And if I start from scratch for basically everything already, why would I _need_ automation?
Currently, I use hand-crafted docker-compose files, written directly over SSH, without any meaningful version control (or anything similar).
docker-compose manages to have the right amount of "declarative configuration" vs "quickly getting things working", especially if the thing I am looking to install already has a Docker image ready.
If the thing doesn't have a Docker image, I'll first fall back to distro packages and creating Docker images using those. Building from source is the last thing I'll try, because for most things it's not that fun to debug.
I copy paste parts of my configuration from old dotfiles repositories manually. This also helps me get rid of the "cruft" that accumulates of old and unused software.
So, do you know of a tool that might be able to work better with how I do things? Do reach out if you do. This kind of automation / infrastructure as code / whatever you call it thing is still a neat topic to me and I still want to learn more even if I might not use it right now.
🐺 · CC BY-SA 4.0 · me@ecmelberk.com