Docker Stack

In this comprehensive guide, you'll learn how to move from single-host Docker Compose deployments to robust, multi-node orchestration using Docker Stack on a Swarm cluster. We’ll compare docker run vs. Compose, introduce Docker Swarm concepts, and walk through a real-world voting application example—complete with replicas, placement constraints, resource limits, and health checks.

1. Docker Run vs. Docker Compose

When you start with containers, you often use docker run for each service:

docker run simple-webapp
docker run mongodb
docker run redis:alpine

However, for multi-service applications, Docker Compose simplifies management by defining services in a single YAML file:

# docker-compose.yml
version: "3"
services:
  web:
    image: simple-webapp
  database:
    image: mongodb
  messaging:
    image: redis:alpine

Launch all services together:

docker-compose up

This approach centralizes configuration but is limited to a single host.

2. Introducing Docker Swarm and Stacks

Docker Swarm enables clustering multiple Docker engines into a single, fault-tolerant Swarm cluster. Instead of docker run, you create services:

docker service create --name web simple-webapp
docker service create --name database mongodb
docker service create --name messaging redis:alpine

With Docker Stack, you can use your Compose file to deploy across the Swarm:

docker stack deploy --compose-file docker-compose.yml mystack

3. Containers, Services, and Stacks

• Container: A running instance of an image, isolated with its dependencies.
• Service: A scalable set of containers of the same image, distributed across Swarm nodes.
• Stack: A collection of related services that define an application.

4. Example: Voting Application

We’ll deploy a simple voting app with Redis, PostgreSQL, vote service, result service, and a worker.

4.1 Single-Host Deployment with Docker Compose

version: "3"
services:
  redis:
    image: redis
  db:
    image: postgres:9.4
  vote:
    image: voting-app
  result:
    image: result
  worker:
    image: worker

docker-compose up

All services run on your local Docker host.

4.2 Multi-Node Deployment with Docker Stack

Assume a Swarm with one manager and two workers. Enhance the Compose file with a deploy section for Swarm-specific settings.

4.2.1 Replicas

Scale services by defining replica counts:

version: "3"
services:
  redis:
    image: redis
    deploy:
      replicas: 1
  db:
    image: postgres:9.4
    deploy:
      replicas: 1
  vote:
    image: voting-app
    deploy:
      replicas: 2
  result:
    image: result
    deploy:
      replicas: 1
  worker:
    image: worker
    deploy:
      replicas: 1

Deploy the stack:

docker stack deploy --compose-file docker-compose.yml voting

4.2.2 Placement Constraints

Ensure critical services run only on manager nodes:

  db:
    image: postgres:9.4
    deploy:
      replicas: 1
      placement:
        constraints:
          - node.role == manager

4.2.3 Resource Limits

Protect node resources by setting CPU and memory limits:

  vote:
    image: voting-app
    deploy:
      replicas: 2
      resources:
        limits:
          cpus: "0.01"
          memory: 50M

4.2.4 Health Checks

Automatically monitor container health:

  vote:
    image: voting-app
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost"]
      interval: 1m30s
      timeout: 10s
      retries: 3
      start_period: 40s
    deploy:
      replicas: 2
      resources:
        limits:
          cpus: "0.01"
          memory: 50M

5. Common Stack Commands

References

• Docker Stack Deploy
• Compose File Reference
• Docker Swarm Overview
• Docker Healthcheck Documentation