In the first part of this article, I showed how I split the frontend, backend and database all into their own containers, and how each could be individually scaled using docker-compose.
If you’re already familiar with docker-compose and using haproxy for load balancing against container, you might have noticed there’s a limitation in my approach, as both the backend REST service was exposing it’s port 8080 externally, and so I don’t think HTTP requests from the frontend browser in the app were ever passing through the haproxy to be load balanced, only the requests to load the front end on port 80 were being load balanced.
I looked into how I could configure haproxy with multiple backends, listening on different ports, but eventually came to the conclusion that adding two different haproxy containers, one load balancing for port 80 and one for port 8080 was easy to do.
I’m not sure if this is the best way to approach this, but it certainly. works. Leave me a comment if you have any suggestions.
Here’s my final docker-compose.yml:
version: '2' services: mongodata: image: mongo:3.2 volumes: - /data/db entrypoint: /bin/bash mongo: image: mongo:3.2 depends_on: - mongodata volumes_from: - mongodata ports: - "27017" addressbook: image: addressbook depends_on: - mongo environment: - MONGODB_DB_NAME=addressbook ports: - "8080" links: - mongo web: image: docker-web-angularjs ports: - "80" lb-web: image: dockercloud/haproxy depends_on: - web environment: - STATS_PORT=1936 - STATS_AUTH="admin:your-password" links: - web volumes: - /var/run/docker.sock:/var/run/docker.sock ports: - 80:80 - 1936:1936 lb-addressbook: image: dockercloud/haproxy depends_on: - addressbook environment: - STATS_PORT=1937 - STATS_AUTH="admin:your-password" links: - addressbook volumes: - /var/run/docker.sock:/var/run/docker.sock ports: - 8080:80 - 1937:1937