Working with Docker Swarm

Docker Swarm is a container orchestration tool that enables you to manage a cluster of Docker nodes as a single virtual system. It provides native clustering functionality for Docker containers, allowing you to deploy and scale applications across multiple machines while maintaining high availability and load balancing.

In a Docker Swarm cluster, there are two types of nodes: manager nodes that control the cluster and make orchestration decisions, and worker nodes that run the actual containers. The swarm manager handles scheduling, load balancing, and service discovery automatically.

Docker Swarm operates in two service modes: Replicated Service mode where you specify the number of replica tasks distributed across available nodes, and Global Service mode where one task runs on every available node in the cluster.

Creating Docker Machines

First, create multiple Docker machines with one serving as the swarm manager and others as worker nodes −

sudo docker-machine create --driver hyperv manager

sudo docker-machine create --driver hyperv worker1
sudo docker-machine create --driver hyperv worker2
sudo docker-machine create --driver hyperv worker3
sudo docker-machine create --driver hyperv worker4
sudo docker-machine create --driver hyperv worker5

Verify the machines have been created −

sudo docker-machine ls

Setting Up Swarm Cluster

Get the manager's IP address and SSH into the manager node −

sudo docker-machine ip manager
sudo docker-machine ssh manager

Initialize the swarm from within the manager node −

docker swarm init --advertise-addr <manager-ip>

Check the swarm status −

docker node ls

Adding Worker Nodes

From the manager node, get the join tokens for worker and manager nodes −

docker swarm join-token worker
docker swarm join-token manager

SSH into each worker node and execute the join command provided by the token output −

sudo docker-machine ssh worker1
# Inside worker1 SSH session:
docker swarm join --token <worker-token> <manager-ip>:2377

Repeat this process for all worker nodes. Verify the cluster from the manager node −

docker node ls

Creating and Managing Services

Deploy a service with multiple replicas from the manager node −

docker service create --replicas 5 -p 80:80 --name web nginx

Check service status and container distribution −

docker service ps web

Scaling Services

Scale the service up or down as needed −

docker service scale web=7

Verify the scaling operation −

docker service ps web

Key Features

• Load Balancing − Automatically distributes incoming requests across healthy containers

• Service Discovery − Built-in DNS-based service discovery for inter-service communication

• Rolling Updates − Update services without downtime by gradually replacing containers

• High Availability − Multiple manager nodes provide fault tolerance for the control plane

• Declarative Configuration − Define desired state and let Docker Swarm maintain it

Best Practices

• Use an odd number of manager nodes (3, 5, 7) to avoid split-brain scenarios

• Separate manager and worker roles for production environments

• Implement health checks for services to enable automatic recovery

• Use secrets management for sensitive data like passwords and certificates

Conclusion

Docker Swarm provides a straightforward approach to container orchestration, offering built-in clustering, load balancing, and service management. It's ideal for teams looking for a simple yet powerful solution to deploy and scale containerized applications across multiple machines while maintaining high availability.