I’ve never had so much fun self-hosting. A decade or so ago I was hosting things on Linode and running all kinds of servers for myself but with the rise of cloud services, I favored just giving everything to Google. I noticed how popular this community was on Reddit/Lemmy and now it’s my new addiction.
I’m a software engineer and have plenty of experience deploying to AWS/GCP so my head has been buried in the sand with these cloud providers. Now that I’m looking around there are things like NextCloud, Pihole, and Portainer all set up with Cloudflare Zero Trust… I feel like I’m living the dream of having the convenience to deploy my own services with proper authentication and it’s so much fun.
Reviving old hardware to act as local infra is so badass it feels great turning on old machines that were collecting dust. I’m now trying to convince my brother to participate in doing hard-drive swaps on a monthly basis so I have some backup redundancy off-site without needing to back up to the cloud.
Sorry if this feels ranty but I just can’t get over how awesome this is and I feel like a kid again. Cheers to this awesome community!
EDIT: Just also found Fission and OpenFaaS, selfhosted serverless functions, I’m jumping with joy right now!
Is there a specific part that you’re having trouble with? Is it more how it works under the hood, or more about using it to spin up containers? I can try to answer any questions and post some how tos for you.
I think I just need a general overview. Something about the concept isn’t clicking for me, and it makes it hard for me to learn how to use it when I fundamentally don’t get it. Is there a really good “Introduction to Docker and the tools people use with it” that I haven’t found?
I think the real benefits of Docker don’t become unquestionably obvious untill you’ve ever tried to manage more than one installation of some kind of server software in the same machine and inevitably learn the hard way that this comes with a lot of problems and downsides.
Say you install with apt, and the app needs a dependency that breaks your setup. You use docker to utilize your os, but containerize dependencies. You can also better organize which containers use your computer’s network, and which use a virtual network where you can redirect an incoming port to avoid conflicts.
Containers are like VMs, but for an application instead of a whole OS, though you can put multiple apps in one container. Good for if they need to share files.
For a more visual approach, look into Portainer. It gives you an admin page you can open in your browser to manage docker containers.
I actually have Portainer set up and running, and I even spun up a few simple containers in it. Unfortunately I did so by following a guide to complete a specific task. I completed the task successfully, but now I have a Portainer install that I don’t understand in the slightest, and don’t know how to update it or any of the containers in it, or really do anything that wasn’t covered in the guide I followed (which I now cannot find). I found a YouTube video that tries to explain Portainer, but I don’t know the terminology of Docker enough to understand what they are saying, and I haven’t found a Docker video simple enough to bring me up to speed.
The easiest way to think about docker is to consider it a type of virtual machine, like something you’d use VirtualBox for.
So let’s say you run Windows, but want to try out Linux. You’d could install Ubuntu in a VirtualBox VM, and then install software that works on Ubuntu in that VM, and it’s separate from Windows.
Docker is similar to this in that a docker container for a piece off software often includes an entire operating system within it, complete with all of the correct versions of drivers that the software needs to function. This is all in a sandbox/container that does not really interact with the host operating system.
As to why this is convenient: Let’s say that you have a computer running Ubuntu natively/bare metal. It has a certain version of python installed that you need to run the applications you use. But there’s some new software you want to try that uses a later version of python that will break your other apps if you upgrade.
The developer of that software you want to try makes a docker version available. There’s a docker-compose.yml file that specifies things like the port the application will be available on, the time zone your computer is in, the location of the docker files on dockerhub, etc. You can modify this file if you like, and when you are done, you type
docker compose up -din the terminal (in the same directory as the docker-compose.yml file).Docker will then read the compose file, download the required files from the repository, extract them, set up the network and the web server and configure everything else specified in the compose file. Then you open a browser, type in the address of the machine the compose file is on, followed by the port number in the compose file (ex: http://192.168.1.100:5000), and boom, there’s your software.
You can use the new software with the newer version of python at the same time as the old stuff installed directly on your machine.
You can leave it running all the time, or bring it down by typing
docker compose down. Need to upgrade to a new version? Bring the container down, typedocker compose pull, which tells docker to pull the latest version from the repository, thendocker compose up -dto bring the updated version back up again.Portainer is just a GUI that runs docker commands “under the hood”.