I wanted something simple to avoid loosing time in configuring and customizing this blog. I just have to focus on writing some new stuff to share 🙂

The backend is a simple docker compose with 2 services:

  • blog to serve static content with Nginx
  • blog-build as a “sidecar” container constantly watch for updates in blog/ directory and generate static content

This is docker-compose.yml content:

version: "3.9"

services:
  blog:
    container_name: blog
    image: nginx
    volumes:
      - /srv/docker-compose/blog/_site/:/var/www/html
      - /srv/docker-compose/nginx.conf.template:/etc/nginx/templates/nginx.conf.template
    ports:
     - "8001:8001"
    environment:
      NGINX_HOST: blog.ralex.fr
      NGINX_PORT: 8001
    restart: always

  blog-build:
    container_name: blog-build
    image: jekyll/jekyll:latest
    entrypoint: jekyll
    command: build --watch --incremental
    environment:
      JEKYLL_ROOTLESS: 1 
    volumes:
      - /srv/docker-compose/blog:/srv/jekyll
    restart: always

There is a special feature of the official Nginx docker image that is used to customize hostname and port with nginx.conf.template file:

server {
    listen       ${NGINX_PORT};
    server_name  ${NGINX_HOST};
    root /var/www/html;

    location / {
        try_files $uri $uri/ $uri.html =404;
    }
}

You can read more about it here.

Any modification to _config.yml will need a restart with:

docker-compose restart blog-build

But any addition to _posts/ directory will be automatically published. It is a good idea to put published: false in the header of a new article while it is not yet available for public reading.

Another way is to put a date in the future in the header of your article. You will see this kind of message in your logs until the post is ready to be published:

Skipping: _posts/2023-12-21-how-to-serve-this-blog.md has a future date

Then, it’s up to you to choose to version this code in Git or, at least, back it up! 😉