This example is built around a web-based voting application that collects, tallies, and returns the results of votes (for cats and dogs, or other choices you specify). The voting app includes several services, each one running in its own container. We’ll deploy the app as a stack to introduce some new concepts surfaced in Compose Version 3, and also use swarm mode, which is cluster management and orchestration capability built into Docker Engine.
If you haven’t yet downloaded Docker or installed it, go to Get Docker and grab Docker for your platform. You can follow along and run this example using Docker for Mac, Docker for Windows or Docker for Linux.
Once you have Docker installed, you can run docker run hello-world
or other commands described in the Get Started with Docker
tutorial to verify your installation.
If you are totally new to Docker, you might continue through
the full Get Started with Docker tutorial
first, then come back.
In this tutorial, you’ll learn how to:
docker machine
to create multiple virtual local hosts or
dockerized cloud serversdocker
commands to set up and run a swarm with manager and worker nodesvote
app services across the two nodes by feeding our example docker-stack.yml
file to
the docker stack deploy
commandvisualizer
to explore and understand the runtime app and servicesdocker-stack.yml
and redeploy the app using a different
vote
image to implement a poll on different choicesA service is a bit of executable code designed to accomplish
a specific task. A service can run in one or more
containers. Defining a service configuration for your app
(above and beyond docker run
commands in a Dockerfile) enables you to
deploy the app to a swarm and manage it as a distributed,
multi-container application.
The voting app you are about to deploy is composed of several services, each based on an image:
Service | Description | Base Image |
---|---|---|
vote |
Presents the voting interface via port 5000 . Viewable at <manager-IP>:5000 |
Based on a Python image, dockersamples/examplevotingapp_vote |
result |
Displays the voting results via port 5001. Viewable at <manager-IP>:5001 |
Based on a Node.js image, dockersamples/examplevotingapp_result |
visualizer |
A web app that shows a map of the deployment of the various services across the available nodes via port 8080 . Viewable at <manager-IP>:8080 |
Based on a Node.js image, dockersamples/visualizer |
redis |
Collects raw voting data and stores it in a key/value queue | Based on a redis image, redis:alpine |
db |
A PostgreSQL service which provides permanent storage on a host volume | Based on a postgres image, postgres:9.4 |
worker |
A background service that transfers votes from the queue to permanent storage | Based on a .NET image, dockersamples/examplevotingapp_worker |
Each service will run in its own container. Using swarm mode, we can also scale the application to deploy replicas of containerized services distributed across multiple nodes.
Here is an example of one of the services fully defined:
vote:
image: dockersamples/examplevotingapp_vote:before
ports:
- 5000:80
networks:
- frontend
depends_on:
- redis
deploy:
replicas: 2
update_config:
parallelism: 2
restart_policy:
condition: on-failure
The image
key defines which image the service will use. The vote
service
uses dockersamples/examplevotingapp_vote:before
.
The depends_on
key allows you to specify that a service is only
deployed after another service. In our example, vote
only deploys
after redis
.
The deploy
key specifies aspects of a swarm deployment, as described below in
Compose Version 3 features and
compatibility.
In addition to defining a set of build and run commands in a Dockerfile, you can define services in a Compose file, along with details about how and where those services will run. You can use Compose files to kick off multiple Dockerfiles, or use Compose files independently of Dockerfiles.
In the Getting Started with Docker tutorial, you wrote a Dockerfile for the whalesay app then used it to build a single image and run it as a single container.
For this tutorial, the images are pre-built, and we will use docker-stack.yml
(a Version 3 Compose file) instead of a Dockerfile
to run the images. When we deploy, each image will run as a service in a
container (or in multiple containers, for those that have replicas defined to
scale the app).
To follow along with the example, you need only have Docker running and the copy
of docker-stack.yml
we provide
here.
This file defines all the services shown in the table
above, their base images, configuration details
such as ports, networks, volumes, application dependencies, and the swarm
configuration.
version: "3"
services:
redis:
image: redis:alpine
ports:
- "6379"
networks:
- frontend
deploy:
replicas: 2
update_config:
parallelism: 2
delay: 10s
restart_policy:
condition: on-failure
db:
image: postgres:9.4
volumes:
- db-data:/var/lib/postgresql/data
networks:
- backend
deploy:
placement:
constraints: [node.role == manager]
vote:
image: dockersamples/examplevotingapp_vote:before
ports:
- 5000:80
networks:
- frontend
depends_on:
- redis
deploy:
replicas: 2
update_config:
parallelism: 2
restart_policy:
condition: on-failure
result:
image: dockersamples/examplevotingapp_result:before
ports:
- 5001:80
networks:
- backend
depends_on:
- db
deploy:
replicas: 2
update_config:
parallelism: 2
delay: 10s
restart_policy:
condition: on-failure
worker:
image: dockersamples/examplevotingapp_worker
networks:
- frontend
- backend
deploy:
mode: replicated
replicas: 1
labels: [APP=VOTING]
restart_policy:
condition: on-failure
delay: 10s
max_attempts: 3
window: 120s
visualizer:
image: dockersamples/visualizer:stable
ports:
- "8080:8080"
stop_grace_period: 1m30s
volumes:
- "/var/run/docker.sock:/var/run/docker.sock"
deploy:
placement:
constraints: [node.role == manager]
networks:
frontend:
backend:
volumes:
db-data:
To deploy the voting application, we will run the docker stack deploy
command
with this docker-stack.yml
file to pull the referenced images and launch the
services in a swarm as configured in the .yml
.
Note that at the top of the docker-stack.yml
file, the version is indicated as
version: "3"
. The voting app example relies on Compose version 3, which is
designed to be cross-compatible with Compose and Docker Engine swarm mode.
Before we get started, let’s take a look at some aspects of Compose files and deployment options that are new in Compose Version 3, and that we want to highlight in this walkthrough.
docker-stack.yml
is a new type of Compose file
only compatible with Compose Version 3.
The deploy key allows you to specify various properties of a swarm deployment.
For example, the voting app configuration uses this to create replicas of the
vote
and result
services (2 containers of each will be deployed to the
swarm).
The voting app also uses the deploy
key to constrain some services to run only
on a manager node.
docker stack deploy
is the command we will use to deploy with docker-stack.yml
.
This command supports only version: "3"
Compose files.
It does not support the build
key supported in standard Compose files, which
builds based on a Dockerfile. You need to use pre-built images
with docker stack deploy
.
It can take the place of running docker compose up
to run
Version 3 compatible applications.
Taken together, these new options can help when you want to configure an app to run its component functions across multiple servers, and use swarm mode for load balancing and performance. Rather than thinking about running individual containers, we can start to model deployments as application stacks and services.
For more on what’s new in Compose Version 3:
Introducing Docker 1.13.0 blog post from Docker Core Engineering
Versioning, Version 3, and Upgrading in Compose file reference
In the next step, we’ll set up two Dockerized hosts.