Multi-VM Environments

Vagrant supports what are known as “Multi-VM Environments.” These are Vagrant environments which contain multiple virtual machines which typically work together or are somehow associated with each other. An example of some uses of Multi-VM environments:

• Accurately modeling a separate web and database server within the same development environment.
• Modeling a cluster of machines, and how they behave together.
• Testing an interface, such as API calls or a chat interface.
• Testing a load balancer configuration, or the effects of “unplugging” a machine.

Historically, running complex environments such as these was done by flattening them onto a developer’s machine. For example, in the case where there may typically be a separate worker server which may process images, the web, queue, and workers would all run side by side on the same development machine.

The problem with this is that it is an inaccurate model of the production setup. Vagrant and Multi-VM environments allow developers to model complex multi-server setups on a single development machine without getting in the way of the development process (continue to use the same editors, tools, etc.).

Defining Multiple VMs

The multiple VMs are all defined within the same single Vagrantfile which you’re probably already used to. A small example is shown below:

Vagrant::Config.run do |config|
  config.vm.define :web do |web_config|
    web_config.vm.box = "web"
    web_config.vm.forward_port 80, 8080
  end

  config.vm.define :db do |db_config|
    db_config.vm.box = "db"
    db_config.vm.forward_port 3306, 3306
  end
end

Using config.vm.define, you can create multiple VMs (as many as you’d like) within a single Vagrantfile. The web_config or db_config variables used within the definitions are the exact same, functionally, as a Vagrantfile’s typical config. This allows the VMs to have their own custom configuration on any configurable value.

In the case above, we’re defining two VMs: “web” and “db.” These two VMs are based on two different boxes. The web VM forwards port 80 while the database VM forwards port 3306.

Controlling Multiple VMs

The moment that multiple VMs are introduced in a Vagrantfile, the usage of the various vagrant command line commands change slightly. This change is hopefully mostly intuitive, and will be shown using a very simple example based on vagrant up.

In a single VM environment, vagrant up starts that VM. In a multi-VM environment vagrant up starts every VM. If a name is specified to the command such as vagrant up web then it will start only that specific VM.

This pattern follows for every other command as well, although some don’t implement the “every VM” functionality when it doesn’t make sense, such as vagrant ssh, which requires that a VM name be specified if it’s in a multi-VM environment.

Communication Between VMs

With multiple VMs up and running, the next step is to support inter-VM communication so that, say, a web server can talk to its associated database server.

This communication is typically done through host-only networking. There is an entire page dedicated to the topic, but a relatively simple example is given below, based on the example given earlier:

Vagrant::Config.run do |config|
  config.vm.define :web do |web_config|
    # ...
    web_config.vm.network :hostonly, "192.168.1.10"
  end

  config.vm.define :db do |db_config|
    # ...
    db_config.vm.network :hostonly, "192.168.1.11"
  end
end

The above assigns a static IP to both the web and database VMs. This static IP can then be used to communicate directly to the other VMs. For more details on how to do things such as creating separate networks, joining the same network from separate Vagrantfiles, etc. please read the host-only networking page.