Deployment on a Docker Swarm

As for the multi Docker host environment, a Docker Swarm requires a key value store to gather the nodes / containers configurations and states.

Creation of a key-value store

Note: if you still have the key-value store from the previous chapter do not re-create it and go directly to the creation of the Swarm.

Several steps are needed to run the key-value store

• Create dedicated Docker host with Machine) docker-machine create -d virtualbox consul
• Switch to context of the newly created machine eval "$(docker-machine env consul)"
• Run container based on Consul image docker run -d -p "8500:8500" -h "consul" progrium/consul -server -bootstrap

Creation of the Swarm

Additional options need to be provided to docker-machine in order to define a Swarm.

Creation of the Swarm master

$ docker-machine create \
-d virtualbox \
--swarm \
--swarm-master \
--swarm-discovery="consul://$(docker-machine ip consul):8500" \
--engine-opt="cluster-store=consul://$(docker-machine ip consul):8500" \
--engine-opt="cluster-advertise=eth1:2376" \
demo0

Creation of the Swarm agent

$ docker-machine create \
 -d virtualbox \
--swarm \
--swarm-discovery="consul://$(docker-machine ip consul):8500" \
--engine-opt="cluster-store=consul://$(docker-machine ip consul):8500" \
--engine-opt="cluster-advertise=eth1:2376" \
demo1

List the nodes

We have created 3 Docker hosts (key-store, Swarm master, Swarm agent)

$ docker-machine ls

NAME     ACTIVE   DRIVER         STATE     URL                         SWARM
consul   *        virtualbox     Running   tcp://192.168.99.100:2376
demo0    -        virtualbox     Running   tcp://192.168.99.101:2376   demo0 (master)
demo1    -        virtualbox     Running   tcp://192.168.99.102:2376   demo1

Create a DNS load balancer

In order to load balance the traffic towards several instances of our app service, we will add a new service. This one uses the DNS round-robin capability of Docker engine (version 1.11) for containers with the same network alias.

Note: to present the DNS round-robin feature, we do not use the load balancer of the previous chapter (dockercloud/haproxy).

The following Dockerfile uses nginx:1.9 official image and add a custom nginx.conf configuration file.

FROM nginx:1.9

# forward request and error logs to docker log collector
RUN ln -sf /dev/stdout /var/log/nginx/access.log
RUN ln -sf /dev/stderr /var/log/nginx/error.log

COPY nginx.conf /etc/nginx/nginx.conf

EXPOSE 80

CMD ["nginx", "-g", "daemon off;"]

The following nginx.conf file define a proxy_pass directive towards http://apps for each request received on port 80.

apps is the value we will set as the app service network alias.

user nginx;
worker_processes 2;
events {
  worker_connections 1024;
}
http {
  access_log /var/log/nginx/access.log;
  error_log /var/log/nginx/error.log;

  # 127.0.0.11 is the address of the Docker embedded DNS server
  resolver 127.0.0.11 valid=1s;
  server {
    listen 80;
    # apps is the name of the network alias in Docker
    set $alias "apps";

    location / {
      proxy_pass http://$alias;
    }
  }
}

Let's build and publish the image of this load-balancer to Docker Cloud:

# Create image
$ docker build -t lucj/lb-dns .

# Publish image
$ docker push -t lucj/lb-dns

The image can now be used in our Docker Compose file.

Update our docker-compose file

The new version of the docker-compose.yml file is the following one

version: '3'
services:
  mongo:
    image: mongo:3.2
    networks:
      - backend
    volumes:
      - mongo-data:/data/db
    expose:
      - "27017"
    environment:
      - "constraint:node==demo0"
  lbapp:
    image: lucj/lb-dns
    networks:
      - backend
    ports:
      - "8000:80"
    environment:
      - "constraint:node==demo0"
  app:
    image: lucj/message-app
    expose:
      - "80"
    environment:
      - MONGO_URL=mongodb://mongo/messageApp
      - "constraint:node==demo1"
    networks:
      backend:
        aliases:
          - apps
    depends_on:
      - lbapp
volumes:
  mongo-data:
networks:
  backend:
    driver: overlay

There are several important updates here

• usage of the lb-dns image for the load balancer service
• constraints to choose the nodes on which each service will run (needed in our example to illustrate the DNS round robin)
• creation of a new user-defined overlay network to enable each container to communicate with each other through their name
• for each service, definition of the network used
• definition of network alias for the app service (crucial item as this is the one that will enable nginx to proxy requests)

Deployment and scaling of the application

In order to run the application in this Swarm, we will issue the following commands

• switch to the swarm master context eval $(docker-machine env --swarm demo0)
• run the new compose file docker-compose up
• increase the number of app service instances docker-compose scale app=5

Our application is then available through http://192.168.99.101:8000/message

192.168.99.101 is the IP of the Swarm master. 8000 is the port exported by the load balancer to the outside.