Getting started with swarm mode

This tutorial introduces you to the features of Docker Engine Swarm mode. You may want to familiarize yourself with the key concepts before you begin.

The tutorial guides you through the following activities:

• initializing a cluster of Docker Engines in swarm mode
• adding nodes to the swarm
• deploying application services to the swarm
• managing the swarm once you have everything running

This tutorial uses Docker Engine CLI commands entered on the command line of a terminal window. You should be able to install Docker on networked machines and be comfortable with running commands in the shell of your choice.

If you are brand new to Docker, see About Docker Engine.

Set up

To run this tutorial, you need the following:

• three Linux hosts which can communicate over a network, with Docker installed
• Docker Engine 1.12 or later installed
• the IP address of the manager machine
• open ports between the hosts

Three networked host machines

This tutorial requires three Linux hosts which have Docker installed and can communicate over a network. These can be physical machines, virtual machines, Amazon EC2 instances, or hosted in some other way. You can even use Docker Machine from a Linux, Mac, or Windows host. Check out Getting started - Swarms for one possible set-up for the hosts.

One of these machines will be a manager (called manager1) and two of them will be workers (worker1 and worker2).

Note: You can follow many of the tutorial steps to test single-node swarm as well, in which case you need only one host. Multi-node commands will not work, but you can initialize a swarm, create services, and scale them.

Docker Engine 1.12 or newer

This tutorial requires Docker Engine 1.12 or newer on each of the host machines. Install Docker Engine and verify that the Docker Engine daemon is running on each of the machines. You can get the latest version of Docker Engine as follows:

• install Docker Engine on Linux machines

• use Docker for Mac or Docker for Windows

Install Docker Engine on Linux machines

If you are using Linux based physical computers or cloud-provided computers as hosts, simply follow the Linux install instructions for your platform. Spin up the three machines, and you are ready. You can test both single-node and multi-node swarm scenarios on Linux machines.

Use Docker for Mac or Docker for Windows

Alternatively, install the latest Docker for Mac or Docker for Windows application on one computer. You can test both single-node and multi-node swarm from this computer, but you will need to use Docker Machine to test the multi-node scenarios.

• You can use Docker for Mac or Windows to test single-node features of swarm mode, including initializing a swarm with a single node, creating services, and scaling services. Docker “Moby” on Hyperkit (Mac) or Hyper-V (Windows) will serve as the single swarm node.

• Currently, you cannot use Docker for Mac or Windows alone to test a multi-node swarm. However, you can use the included version of Docker Machine to create the swarm nodes (see Get started with Docker Machine and a local VM), then follow the tutorial for all multi-node features. For this scenario, you run commands from a Docker for Mac or Docker for Windows host, but that Docker host itself is not participating in the swarm (i.e., it will not be manager1, worker1, or worker2 in our example). After you create the nodes, you can run all swarm commands as shown from the Mac terminal or Windows PowerShell with Docker for Mac or Docker for Windows running.

The IP address of the manager machine

The IP address must be assigned to a network interface available to the host operating system. All nodes in the swarm must be able to access the manager at the IP address.

Because other nodes contact the manager node on its IP address, you should use a fixed IP address.

You can run ifconfig on Linux or macOS to see a list of the available network interfaces.

If you are using Docker Machine, you can get the manager IP with either docker-machine ls or docker-machine ip <MACHINE-NAME> — for example, docker-machine ip manager1.

The tutorial uses manager1 : 192.168.99.100.

Open protocols and ports between the hosts

The following ports must be available. On some systems, these ports are open by default.

• TCP port 2377 for cluster management communications
• TCP and UDP port 7946 for communication among nodes
• UDP port 4789 for overlay network traffic

If you are planning on creating an overlay network with encryption (--opt encrypted), you will also need to ensure ip protocol 50 (ESP) traffic is allowed.

What’s next?

After you have set up your environment, you are ready to create a swarm.