2

In "What is a container?" the docker docs show the following nifty image:

docker

For my scenario I have only App A and App B; let's designate them Alice and Bob. How do Alice and Bob communicate, generally and specifically? Using the KISS principle, concretely I'm using MySQL and BaseX from the docker hub.

The notion being for an xquery file residing on the host, which is then sent to the BaseX container for execution. This specific query is actually SQL, so the MySQL container will need to be queried for the result.

Does this necessitate port forwarding or other networking approaches? Volumes? Does this require an additional compose file composed of, in turn, MySQL and BaseX?

No doubt there are many possible approaches. What's the most pragmatic, from a learning perspective? Conceptually light, with a gradual learning curve.

One approach seems to utilize bridge networking: simple

Taking advantage of the fact that the containers can see each other.

This doesn't seem to me related to RESTful web services.


Looking at related questions, this might touch on a reverse proxy.

Nicholas Saunders
  • 375
  • 3
  • 12
  • 22

1 Answers1

1

I'm still fairly new to the Docker game, too, but I've got a few little projects that I do exactly what you are asking about. Through my own experience with trial & error (mostly error...) I would suggest using a docker-compose.yml with your network set to static internal IPs so each container will have predictable, repeatable results and IPs that you can configure your containers to communicate across.

I'm not familiar with BaseX but I have a sample situation using docker-compose version: "3.7" here that could easily run multiple apps and/or be added to or repurposed to fit your hypothetical.

This example docker-compose has the following:

  • A separate container for populating my app-volume volume with the contents of a public GitHub repo.

  • An Apache Web Server container with PHP-FPM enabled for a "selective backend" that gets built from a Dockerfile in the ~/dockerbuild directory.

  • An nginx web server container used as a reverse proxy for all things PHP to be passed to Apache (but serves & caches pretty much everything else).

  • A MariaDB container in the mix for the app DB that I populate with a mysqldump file after the app gets built.

This is a slimmed down version of my docker-compose.yml, so note that you can add a variety of things like healthcheck: and command: to fit your needs.

You execute with docker-compose up -d --build to both build and start your app in one command (but you'll probably want to create your own ba(.sh) file that you execute afterwards, too for those important finishing touches).

Take notice of static-network. And while I am using named volumes in the example, that's just a preference of mine.

version: "3.7"

services:

github_repo_clone: image: debian:latest container_name: github_repo_clone networks: static-network: ipv4_address: 172.20.0.254

command: bash -c "
    apt-get update && 
    apt-get install git -y &&
    rm -rf examplerepo &&
    git clone https://github.com/exampleuser/examplerepo.git &&
    rm -rf /var/www/vhosts/exampleuser/examplerepo &&
    mkdir -p /var/www/vhosts/exampleuser/examplerepo &&
    cp -rapv examplerepo/* /var/www/vhosts/exampleuser/examplerepo &&
    rm -rf examplerepo &&
    tail -f /dev/null
  "
volumes:
 - app-volume:/var/www/vhosts

apache_php-fpm_backend: container_name: debian-build-demo0 build: ~/dockerbuild ports: - "8080:8080" networks: static-network: ipv4_address: 172.20.0.2 volumes: - app-volume:/var/www/vhosts

nginx_rp: image: nginx:latest container_name: nginx-build-demo0 ports: - "80:80" - "443:443" networks: static-network: ipv4_address: 172.20.0.3 volumes: - app-volume:/var/www/vhosts

mariadb: restart: always image: mariadb:latest container_name: mariadb-build-demo0 ports: - "3306:3306" networks: static-network: ipv4_address: 172.20.0.10 environment: - MYSQL_ROOT_PASSWORD=4007p@$$w04d - MYSQL_USER=dbusernamegoeshere - MYSQL_PASSWORD=@p@$$w04d - MYSQL_DATABASE=dbnamegoeshere volumes: - app-volume:/var/www/vhosts healthcheck: test: ["CMD", "mysqladmin" ,"ping", "-h", "localhost"] timeout: 10s retries: 5

volumes: app-volume:

external: true

networks: static-network: driver: bridge ipam: driver: default config: - subnet: 172.20.0.0/16

My clean up command when I take the app down is kind of ugly, but not terribly difficult to read and goes something like this: docker-compose down -v ; docker rmi -f dockerbuild_apache_php-fpm_backend:latest mariadb:latest nginx:latest debian:latest ; docker rmi -f $(docker images -q --filter "dangling=true") ; docker rmi -f $(docker images -q --filter label=stage=intermediate) ; docker volume prune -f;

I think each "independent app" could (and probably should) have its own docker-compose and/or Dockerfiles, but you may find synergies with certain Dockerfiles that you can create custom command: instructions for (such as the git_repo_clone example) to be used & shared across multiple "apps".

CoolElectricity
  • 195
  • 1
  • 9