Building the deployment tool I wish I had | Hacker News Reader

Building the deployment tool I wish I had

ruuda.nl

44 pointsruuda5 comments

The current batch of "building X bespoke just for me" is really lovely to see. Many of these tools are very much a bicycle built for one, and I'm not test-driving them myself, and _that's okay_. Not everything has to be a profitable enterprise--and seeing folks vibe up something that patches over a hole in their workflow or life is a nice upside of all the AI tooling.

Note that I have no idea if this is vibed! It might not be! I just like the general theme of lots of pretty specific tooling.

I've written my own DNS server and my own "I can't believe it's not docker compose and kind of ansible", among other more and less esoteric but consistently dubiously motivated enterprises

They all inevitably ended up with me using them for a good 1-2 years, while also gaining an appreciation for whatever I thought was over-complicated about the OSS version (not ansible tho. But maybe pyinfra.). And I learned soooo much. I can't tell you how helpful it's been both here and there while designing architecture for actual work ("this is basically just DNS") and especially in interviews, both just having something to talk about, and also having a story arc to it showing lessons learned etc. plus it's nerdy which is always helpful branding as long as you done come off too weird or boastful along the way.

I actually really like this!

This is unique and cool. And it’s a n actual new take on something and I could totally see it working on some use cases I personally have atm.

Even stuff like k8’s the whole thing other than like kube-api if I had time would just be the same binary as kubelet lol

It would solve a ton of things

There's definitely a much smaller, cleaner kubernetes hiding inside the current monstrosity.

LB, Services, pods, DNS gets you like 99% of the way there

What's wrong with OpenTofu and cloudinit?

Reply on news.ycombinator.com