Lesson 9 - Docker Swarm

15/02/2024 - 3 phút

Docker Swarm is an integrated orchestration tool within Docker, allowing you to manage a cluster of Docker nodes as a single unit. With Docker Swarm, you can easily deploy, scale, and manage containerized applications across multiple hosts. Here’s a more detailed guide on setting up and using Docker Swarm.

Below is a more detailed description of how a Docker Swarm operates, represented through a diagram and explained in English:

graph TD
    A[Manager Node] -->|Assigns Tasks| B[Worker Node 1]
    A -->|Assigns Tasks| C[Worker Node 2]
    A -->|Assigns Tasks| D[Worker Node 3]

    B -->|Runs Task| E[Service Container 1]
    C -->|Runs Task| F[Service Container 2]
    D -->|Runs Task| G[Service Container 3]

    E -.->|Registers| H[Service Discovery]
    F -.->|Registers| H
    G -.->|Registers| H

    H -->|Routes Requests| I[Routing Mesh]
    I -->|Published Port| J[Port on Host]

    classDef manager fill:#f9f,stroke:#333,stroke-width:4px;
    classDef worker fill:#bbf,stroke:#333,stroke-width:2px;
    classDef service fill:#fe9,stroke:#333,stroke-width:2px;
    classDef discovery fill:#bfe,stroke:#333,stroke-width:2px;
    classDef routing fill:#ffa,stroke:#333,stroke-width:2px;
    classDef published fill:#f96,stroke:#333,stroke-width:2px;

    class A manager;
    class B,C,D worker;
    class E,F,G service;
    class H discovery;
    class I routing;
    class J published;

Manager Node: The central control point of Docker Swarm, managing task distribution and maintaining the desired state of the cluster.
Worker Nodes (1, 2, 3): Nodes that execute tasks assigned by the Manager Node. Each Worker Node can run one or more Service Containers.
Service Containers (1, 2, 3): Containers that execute specific tasks of a service. Each container runs a replica of the service configured in the Swarm.
Service Discovery: A mechanism that allows containers within the Swarm to find and communicate with each other through services, rather than using direct IP addresses.
Routing Mesh: A routing mechanism that allows requests to a service to be automatically directed to any container running that service, regardless of which Worker Node the container is running on within the cluster.
Port on Host: Ports on Docker hosts used to access services from outside the cluster. The Routing Mesh forwards requests to the appropriate service through this port.

Setting Up Docker Swarm

Step 1: Initialize Docker Swarm

To turn a Docker host into a Swarm manager, use the following command on that machine:

docker swarm init --advertise-addr <manager_ip>

--advertise-addr specifies the IP address or interface that other nodes can use to join the Swarm manager.

After initialization, the command will return a join token that you will use to add worker nodes to the Swarm.

Step 2: Add Worker Nodes to the Swarm

On each worker node, use the join token received from the manager node to join the Swarm:

docker swarm join --token <SWMTKN-1-...> <manager_ip>:2377

This command will connect the current worker node to the Swarm cluster managed by the manager node at <manager_ip>.

Step 3: Deploy Services in the Swarm

A service in Docker Swarm is a high-level description of the tasks you want to run on the cluster. To deploy a service, use the following command:

docker service create --name my_service -p 80:80 --replicas 3 nginx

--name sets the name of the service.
-p configures port mapping from the host to the container.
--replicas sets the number of container replicas you want to deploy.
nginx is the name of the Docker image you want to use for this service.

Step 4: Managing and Scaling Services

List services:
```
docker service ls
```
Scale the number of replicas for a service:
```
docker service scale my_service=5
```
View tasks of a service:
```
docker service ps my_service
```

Step 5: Removing Services and Disbanding the Swarm

Remove a service:
```
docker service rm my_service
```
Disband the Swarm:
On the manager node:
```
docker swarm leave --force
```
On each worker node:
```
docker swarm leave
```

Notes on Using Docker Swarm

Ensure the security of the cluster by using --autolock when initializing the Swarm to encrypt the Raft log and other critical data.
Use Docker Secrets and Docker Configs to securely manage configurations and sensitive data within the Swarm.
Monitor the status and performance of the Swarm using tools like Docker Swarm Visualizer or Portainer.

Docker Swarm offers a simple and effective solution for managing containers and services across multiple hosts. With familiar Docker commands and built-in integration, you can easily scale and manage your applications in a distributed environment.