Docker setup for Atavism Server

[arturelevated]

Hey. I prepared a setup to run Atavism Server in a docker container. Repo here:

https://github.com/ArturMoczulski/atavism-server-docker/

Hope this helps some of you guys. I think this setup is much easier than manual installation or VirtualBox.

[Kindred]

On 5/4/2024 at 4:14 AM, arturelevated said:

Hey. I prepared a setup to run Atavism Server in a docker container. Repo here:

https://github.com/ArturMoczulski/atavism-server-docker/

Hope this helps some of you guys. I think this setup is much easier than manual installation or VirtualBox.

I would love to combine our efforts, I see a need for this to be a truly distributed service and not a single container. I wanted to split up the `auth.sh` and `world.sh`, as they launch about I think it was 8+ different services I wanted to containerize using compose, then move to Kubernetes, Azure Container Apps, or AWS. 

https://github.com/delconis/atavism-server-docker

[arturelevated]

Fantastic. I just checked out the repo. I will look deeper tomorrow and see what you got exactly.

First thought is I wanted to separate each plugin into separate container so that logging is easier to inspect cause right now with all in one its complete chaos. And I see you already have it on your roadmap there.

I think that before moving to orchestration more important goal would be to get it to place where it provides smooth development experience locally first, like clarity of log inspection.

Also, I've setup a pretty good workflow for VS code to trigger `agis.jar` build and linking on File save with VS code. I think if we combine it with appropriate docker restart we could achieve hot reload for the Atavism server on file save. I think it's essential for solid dev velocity. Extracting JAR from eclipse, copying files and manually restarting Atavism server all the time is a killer.

One thing that worries me though is the overhead of containerizing all the services. I currently work on a different project with about 13 different containerized docker services and it proves to be quite heavy for engineers with slower machines plus a big disk space hog - docker tends to do that from my experience, although I'm defo not a docker expert so maybe can be optimized some way.