An idea for a future MineOS?


#1

With IncludeOS (http://www.includeos.org/) you could really make MineOS standalone for virtual machine environments, without the hassle of making an entire OS.


#2

For the most part, all the work in making an OS is already done for me by the hard-working folks at Turnkey Linux. Docker, too, has this sort of idea where pretty much only the deployment of the app does the dev need to worry about.

One thing about includeos (and likewise about docker) which keep these models from being explored further is that the needs of Minecraft admins vary so widely. Sometimes, users want MySQL. Sometimes (or even, maybe most times) admins don’t. Should an includeOS/docker instance include databases? Does includeOS/Docker make it easy to step into the filesystem? (I know for docker, many people don’t think it does).

I would definitely like to see a more streamlined process and–more importantly–a more self-contained deployment method, but I struggle to come up with one that doesn’t step away from the overt simplicity of “install any OS you want, and put the MineOS webui atop it”.

Realistically, one answer would be for the webui itself to be aware that it is container-ized, and then to add functionality within the UI that allows for easy portability. An example of this is discourse, the engine that drives these forums. Discourse itself is in a docker container for which I’m only vaguely aware of all the components within–postgres, nginx, etc. However, if I needed to move my entire forum elsewhere, the Discourse UI allows me to export the entire forum and all the relevant files.

This is on the order of a few hundred meg at the top end, and sometimes much less. Exporting Minecraft servers could be several gig. And downloading 14 gig of Minecraft worlds via HTTPS is not a great design move.

Docker-compose is nice, and actually might make some headway with the docker-compose example recently posted here, but ultimately, one of the most reasonable ways to address this is somewhat of a rewrite with the way MineOS works from the get-go, e.g., not relying on the underlying systems’ users and filesystem/binaries.

I actually started to explore this with mineos-ruby (which is the most stable approach at creating servers yet), but it lacks the UI which makes it the least usable one yet. Ideally, such solid back-end scripts should be written into an ESXi model, where basically satellite hosts could be created and they would be maintained remotely; I wasn’t sure if this will ever get any traction though, considering the target audience of MineOS, so dev halted. I still think I might move this forward one day, though.

I’m going off on a tangent. TL;DR IncludeOS is great for the task of low-footprint, minimalist OS deployments, but the end result is mostly just a more compact version of the MineOS one provided via MineOS Turnkey. It would most certainly be a cool thing to explore though, if you wanted; and I’d gladly lend my knowledge and efforts to making it happen if it’s spearheaded by anybody else.