Whenever I want to host something on Computer Science House’s OKD cluster (like this website!), I always forget what steps need to be done to acomplish that. I figured it would be a good idea to document this process on my shiny new blog in the hopes that it might help others.

While some of the steps are specific to node apps, most need to be done for anything else.

Creating the Dockerfile

The whole point of OKD (which is a distribution of Kubernetes) is to manage a whole bunch of containers. Therefore, anything that runs on OKD must be in a container.

The first step of creating a container is writing a Dockerfile which specifies a sequence of steps (or commands) that need to be executed to create the container.

In broad strokes, a Dockerfile might look something like this:

1. Specify a base image (another Dockerfile that is already preconfigured to do something like run a Node app or a Python app)
2. Do some setup (for instance setting environment variables and copying files from the host to the container)
3. start our application

In the case of lonniegerol.com, this is what that looks like:

# Base Image
FROM docker.io/node:19 as builder

# Setup

#1
WORKDIR /app

#2
COPY ["package.json", "package-lock.json*", "./"]

#3
RUN npm install

#4
COPY . /app/

#5
ENV NODE_ENV=production

#6
RUN npm run build

#7
COPY build/ /app/build/

# Starting the Application
CMD [ "node", "build/index.js" ]

I also have a .dockerignore file since I don’t want generated directories, like node_modules/ and generated files like npm-debug.log to be copied over to the container

node_modules
npm-debug.log

Components of the Dockerfile

Base Image

• The app I’m hosting is a node app, so i’m using the node base image

Setup

1. Setting node’s environment to production so that things like the commit hash that are only shown in dev are hidden
2. Setting the current working directory in the container
3. Copying both package.json and package-lock.json into the container’s working directory
4. Installing node packages
5. Copying all files (that aren’t in the .dockerignore file) into the container’s working directory
6. Compiling all .svelte files into html files that are ready to be served
7. Copying the built site into the containers working directory

Starting the Application

• Starting the node server and serving my website

Building and running the container

Once the a Dockerfile has been created you need to have your container engine of choice, like Docker or Podman build and then run the container. I suck at remembering shell commands so I usually have a shell script that accomplishes this for me:

podman build -t lontronix/lonniegerol.com .
podman run -i -t -p 4200:4200 \
	-e PORT='4200' \
	lontronix/lonniegerol.com

In the above script, the -t flag in the build command allows you to set a tag for your build so that you don’t need to refer to it using its randomly generated identifier. The -i and -t flags make it easier to quit the container by using ctrl + c and the -p flag lets you forward traffic from a port in the container to a port on the host machine.

Welcome to my blog 👋🏻

I plan on posting bits of technical knowledge that I learn and think are worth sharing.

Huge thanks to Josh Collinsworth for writing a FANTASTIC blog post that helped give me a better idea of how to work with SvelteKit.