Docker Swarm

Is a single Docker instance struggling to handle the traffic? Swarm lets you scale from 1 to 3 instances and deploy new versions with zero downtime.

1. What You'll Learn

• Swarm Cluster Architecture and Initialization
• The Difference Between a Service and a Container
• Rolling Updates and Rollback Strategies
• Secrets and Configuration Management
• Multi-node network routing

2. A Story of E-commerce Operations

(1) Pain Point: A single server can't handle the traffic

Traffic to Charlie’s e-commerce business skyrocketed, and a single Docker instance couldn’t handle it—during peak hours, CPU usage hit 100%, API responses timed out, and user complaints poured in. He needed to scale to multiple servers, but manually starting containers on three separate servers was too complicated.

(2) Solutions for Docker Swarm Clusters

Charlie initialized a 3-node Swarm cluster, scaled the API to 5 replicas, and deployed the new version using rolling updates with zero downtime.

# Initialize Swarm cluster
docker swarm init --advertise-addr 10.0.0.1

# Deploy service with 5 replicas
docker service create --replicas 5 --name api -p 8080:80 myapp:v2

# Rolling update to v3
docker service update --image myapp:v3 api

(3) Benefits: From a single machine to a cluster—done in 5 minutes

Three servers were combined into a cluster, with the API automatically distributed across five replicas and rolling updates performed with zero downtime. The transition from a single server to a cluster took only five minutes.

3. Swarm Architecture

(1) Node Roles

graph TB
    MGR1["Manager Node 1<br/>Leader"] --- MGR2["Manager Node 2<br/>Standby"]
    MGR1 --- MGR3["Manager Node 3<br/>Standby"]
    MGR1 --> WRK1["Worker Node 1"]
    MGR2 --> WRK2["Worker Node 2"]
    MGR3 --> WRK3["Worker Node 3"]

graph TB
    MGR1["Manager Node 1<br/>Leader"] --- MGR2["Manager Node 2<br/>Standby"]
    MGR1 --- MGR3["Manager Node 3<br/>Standby"]
    MGR1 --> WRK1["Worker Node 1"]
    MGR2 --> WRK2["Worker Node 2"]
    MGR3 --> WRK3["Worker Node 3"]

(1) Service vs Container

4. Swarm Initialization

▶ Example: Initializing with swarm init (Difficulty: ⭐⭐)

# On Manager node: initialize the swarm
docker swarm init --advertise-addr 10.0.0.1

# Output includes the join token for workers
# docker swarm join --token SWMTKN-xxx 10.0.0.1:2377

# On Worker nodes: join the swarm
docker swarm join --token SWMTKN-xxx 10.0.0.1:2377

# List all nodes
docker node ls

ID            HOSTNAME   STATUS   AVAILABILITY   MANAGER STATUS
abc123 *      manager1   Ready    Active         Leader
def456        worker1    Ready    Active
ghi789        worker2    Ready    Active

5. Service Management

▶ Example: service create deployment (Difficulty: ⭐⭐)

# Create a service with 3 replicas
docker service create \
  --name web \
  --replicas 3 \
  --publish 80:80 \
  nginx:1.25-alpine

# List services
docker service ls

# Check service tasks (containers)
docker service ps web

▶ Example: Service Scale Scaling (Difficulty: ⭐⭐)

# Scale to 5 replicas
docker service scale web=5

# Verify
docker service ps web

▶ Example: service update—rolling update (Difficulty: ⭐⭐⭐)

# Rolling update: update 2 containers at a time with 30s delay
docker service update \
  --image nginx:1.26-alpine \
  --update-parallelism 2 \
  --update-delay 30s \
  --update-failure-action rollback \
  web

# Check update status
docker service inspect web --format='{{.UpdateStatus.State}}'

(1) Rolling Update Strategy

▶ Example: Service Rollback (Difficulty: ⭐⭐)

# Rollback to previous version
docker service rollback web

6. Secrets and Configuration Management

▶ Example: Creating and Using Docker Secrets (Difficulty: ⭐⭐⭐)

# Create a secret from a file
echo "my_super_secret_password" | docker secret create db_password -

# Use the secret in a service
docker service create \
  --name api \
  --secret db_password \
  -e DB_PASSWORD_FILE=/run/secrets/db_password \
  myapp:1.0

7. Routing Mesh

(1) Network Routing Mechanisms

Swarm's Routing Mesh allows any port on any node to route to any replica of a service—even if that replica is not on the current node.

graph LR
    A["Node 1<br/>:80"] -->|"request"| B["Routing Mesh"]
    C["Node 2<br/>:80"] --> B
    B -->|"round-robin"| D["Replica 1<br/>Node 1"]
    B --> E["Replica 2<br/>Node 2"]
    B --> F["Replica 3<br/>Node 3"]

graph LR
    A["Node 1<br/>:80"] -->|"request"| B["Routing Mesh"]
    C["Node 2<br/>:80"] --> B
    B -->|"round-robin"| D["Replica 1<br/>Node 1"]
    B --> E["Replica 2<br/>Node 2"]
    B --> F["Replica 3<br/>Node 3"]

8. Swarm vs Compose Comparison

▶ Example: Deploying a Compose file to Swarm using stack deploy (Difficulty: ⭐⭐⭐)

# Deploy a compose file as a stack
docker stack deploy -c docker-compose.yml mystack

# List stacks
docker stack ls

# List services in a stack
docker stack services mystack

# Remove a stack
docker stack rm mystack

9. Complete Example: Deploying a Three-Node Cluster

# ============================================
# Complete walkthrough: 3-node Swarm cluster
# Covers: init, join, service, scale, update
# ============================================

# --- On Manager (10.0.0.1) ---
# 1. Initialize Swarm
docker swarm init --advertise-addr 10.0.0.1

# 2. Get worker join token
docker swarm join-token worker

# --- On Workers (10.0.0.2, 10.0.0.3) ---
# 3. Join the swarm (paste the token from step 2)
docker swarm join --token SWMTKN-xxx 10.0.0.1:2377

# --- On Manager ---
# 4. Verify cluster
docker node ls

# 5. Deploy web service with 5 replicas
docker service create \
  --name web \
  --replicas 5 \
  --publish 80:80 \
  --restart-condition on-failure \
  nginx:1.25-alpine

# 6. Check distribution across nodes
docker service ps web

# 7. Rolling update to v1.26
docker service update \
  --image nginx:1.26-alpine \
  --update-parallelism 2 \
  --update-delay 30s \
  web

# 8. Simulate node failure
docker node update --availability drain worker1
docker service ps web  # Replicas rescheduled

# 9. Restore node
docker node update --availability active worker1

# 10. Scale down
docker service scale web=2

# 11. Clean up
docker service rm web
docker swarm leave --force  # On manager

❓ FAQ

📖 Summary

• Swarm = Docker's built-in container orchestration: Managers handle scheduling, while Workers run containers
• Service instead of Container: Specify the number of replicas, automatic self-healing, built-in load balancing
• Live Updates: --update-parallelism + --update-delay Set the Pace
• Routing Mesh: Ports on any node are automatically routed to all replicas of a service
• Secret encrypted storage: Not exposed in environment variables; read from a file within the container
• docker stack deploy -c compose.yml Compose files can be deployed directly to Swarm

📝 Exercises

1. Basic Exercise (Difficulty ⭐): Initialize Swarm on a single machine, create an Nginx service with 3 replicas, and verify that docker service ps displays the replica distribution.
2. Advanced Exercise (Difficulty ⭐⭐): Perform a rolling update (nginx: 1.25 → 1.26) and observe the update process in docker service ps.
3. Challenge (Difficulty: ⭐⭐⭐): Set up a Swarm cluster on three virtual machines, deploy an API service with five replicas, simulate the failure of a worker node, and verify that the replicas automatically migrate to other nodes.

Web-Tutorial Tech Team

A team of developers maintaining programming tutorials. Each tutorial is written and reviewed by developers with expertise in that field. We work to keep our content accurate and reliable — if you spot an issue, please let us know.

About Us