I’ve used minio briefly, and I’ve never used any other self hosted object storage. In the context of spinning it up with docker, it’s pretty easy. The difficult part in my project was that I wanted some buckets predefined. The docker image doesn’t provide this functionality directly, so I had to spin up an adjacent container with the minio cli that would create the buckets automatically every time I spun up minio.
But for your use case you would manage bucket creation manually, from the UI. It seems straight forward enough, and I don’t have complaints. I think it would work for your use case, but I can’t say its any worse or better than alternatives.
I think containers get seen as overhead unfairly sometimes. Yes, its not running on bare metal, so theres a layer of abstraction, but I think in practice the performance is nearly identical. Plus, since AIO does things out of the box for you (like a redis cache for instance) it ends up being more performant than a standalone nextcloud instance that isnt configured properly.
That is to say, I use AIO without issues.
Im using nextcloud news and the associated app. I like it because it lets me play podcasts in a player built into the android app. I havent found an up to date rss reader for freshrss that does the same (read you is beautiful, but doesnt have this feature.) And I have nextcloud already up so its easy to start with.
Theres also many plugins for freshrss, including one for rss-bridge that turns urls into rss feeds. I use this for youtube subscriptions. You could also use rss bridge independently, which is what I use for nextcloud news.
I really think the learning curve will be less than you think. Please consider at least reading the installation instructions. Here’s the page for linuxserver.io’s maintained plex docker container. I’ve linked to the usage section, where you can copy the compose file to deploy it. https://github.com/linuxserver/docker-plex?tab=readme-ov-file#usage
If you use docker, it doesnt matter the distro. And to use docker, you dont really need to understand how/why it works. As long as you can take an example compose file and spin it up (docker compose up) it’ll be less complicated in the long run than managing plex on the host machine (or most software for that matter, which is why containerization is so popular.)
It seemed nice at first, but one major issue: GPU passthrough was a nightmare. It cant be done in the UI and I didnt understand fully how it worked. There are many different tutorials not by promox that are outdated or may not work. It was frustrating enough I jumped to NixOS. Other hiccups included having to go to the terminal to passthrough drives for openmediavault, but that one was kind of straightforward atleast, and it worked first time.
In hindsight, I didnt actually need to virtualize everything at that level, so I never really had a good use case for it anyway. I use containers over entire VMs.
not sure if this fits your usecase, but nixos-mailserver
Ive been playing with the nixified.ai project, which packages two web interfaces for LLMs and image generation. Im also looking into Tabby.ml for code assistant as well. I haven’t gotten deep, but these all look like promising options for utilitizing a server’s hardware but offering the functionality across the network.
I’ve done project rewrites. This minimizes the problem solving to mostly just syntax, sometimes a new paradigm if the framework is different enough. But in my experience a rewrites goes so much faster than I expect it, since theres a very clear goal to achieve while rewriting. If someone has an existing project to rewrite, I recommend it. If not, you could implement some project in a framework your comfortable with, and then do a rewrite in the new thing.
Just gonna throw out HJSON as another alternative: https://hjson.github.io/
I thinks a great idea but I have never seen it used in the wild, unforunately.
I’ve been using OpenMediaVault inside Proxmox.
I’ve been very happy with OMV, for the short time I’ve been playing with it. Its FOSS and the web interface makes it very clear all the layers of abstraction you can use to manage a NAS. I highly recommend it.
And proxmox is good too, also FOSS (proxmox VE). I also has another slick web interface to manage stuff. I like the web interfaces because, albiet intimidating, it exposes alot of options available to me, which give me opportunities to research and understand how it works.
But I’m still working on getting everything with it set up, so take my suggestion with a grain of salt!
Theres so many. Check out the awesome list: https://github.com/awesome-selfhosted/awesome-selfhosted
I think your stategy should be one service at a time. Do everything in docker, and start by tackling a simpler service. For example, you should try paperless-ngx. Absolute game changer. I didnt realize how much managing ny own directory structure sucked until I used this. Then, grow your service list more and more!
For more “traditional” or “statistical” modeling (not NN) 100% start with sklearn. It has a plethora of algorithms, and their docs read like a book. You can learn a whole bunch of new methods and techniques from there too. In tandum, you should familiarize yourself with matplotlib, which is the plotting library it uses under the hood (and is by far the most popular plotting library.)
For deep learning, I’d say PyTorch? Tensorflow used to be standard but its fallen out of favor compared to PyTorch. I don’t use either so I’m nit sure.
Actually cd isnt a program. Your present working directory is managed entirely by the shell. If you type “type cat” in the terminal it will tell you its a program, but “type cd” says its a shell builtin. So yes, cd depends on the shell and zsh has some awesome quality of life features. This is not something you can do in bash.
Futhark: a functional language that can be compiled to run in parallel on cpu or gpu. (No need to write cuda directly) https://futhark-lang.org
The guy behind the youtube channel Context Free (about programming languages) made this site that tracks language popularity based on github/stack overflow: https://tjpalmer.github.io/languish
This may not be helpful but I’ve used EJS as a template engine before. It’s a simple engine and been around for a while.
I used Zola for a while, but at the end of the day there wasnt enough themes available that fit what I was looking for. I ended up messing with the templating engine to get what I needed.
I suggest OP choose Hugo over Zola, in the hopes that they find a theme that suits them best and for the most part prevents them from having to touch templating to begin with.