The Kubernetes suggestion is spot on too. It's almost a must-have these days for anything beyond a basic deployment.
And the RabbitMQ rationale is super important. Transparency about those kinds of choices builds trust and encourages community contributions. Took me a while to figure out why some projects went with certain dependencies, and clear documentation would have saved me a lot of time.
Edit: looks like I managed to break my instance without even trying. I can't edit my first task that I created.
https://demo.kaneo.app/dashboard/workspace/nyxfpvfgkk412xg4n...
Steps to reproduce:
1. Log in as [email protected]
2. Click on workspace Microsoft Blue
3. Click on project DataCenters
4. Click on the first task, DAT-1 Build data centers around the world
5. Observe the screen just says "Task edit" with nothing else there.It's the webpage that should be friendlier to people who just want to toy and explore their app before committing to get it running in production.
Though of course it’s nice if you pre-add it, it’s nothing people can’t do themselves.
Source: Senior IT Engineer who likes K8s as a concept, but is incredibly disappointed with its unnecessary complexity and everyone's insistence on its use without understanding its role or purpose.
As an author of a DevOps tool [1] based around Compose files, I beg to differ! It should be pretty easy to adapt the config from the README to use in medium-scale production setups [2] (the only thing that comes to mind is a reverse proxy perhaps).
I think I’ll add Kaneo to the Lunni Marketplace after some more testing!
[1]: https://lunni.dev/
[2]: Read: small to medium sized businesses, startups without a lot of funding etc.
The Compose file sets up a volume for it, so the data is persisted correctly.
[1]: If everything’s set up correctly, there shouldn’t be concurrent writes to the SQLite db, and it should scale pretty well then. (The RabbitMQ + SQLite stack is a bit weird, I agree.)
With documentation on why it was chosen and the function it serves, I (the IT Engineer) am better able to communicate this back to leaders pitching a fit over Broadcom's name on there as well as brief on alternatives. Not saying it's right, just a frustrating reality.
Personal feedback:
* Docker Compose is sufficient for a Quick Start, but not production installation. Needs substantially more documentation about production-ready setups at the very least, especially security considerations.
* Would also add additional Quick Start guides for Kubernetes.
* Documentation should also cover rationale behind prerequisite choices. RabbitMQ is Broadcom's domain, and there's a lot of sour grapes out there who want to steer clear of them, MPL-licensed or not. If alternatives won't be suggested or offered, then at least explain why a given piece was chosen so other builders can explore contributing alternatives.
All in all, the essentials seem to be coming along nicely. Just take the time to document, document, document now, so you're not treading technical debt later should its use take off.