Inter-VM Communication

Introduction

Sometimes you need to have one of your VMs reach another VM in your stack file. In this guide, we will create a stack file with two VMs and make one VM instance ping the other VM as an example.

Full Stack

provider:
  aws:
    access_key: '${var.aws_access_key}'
    secret_key: '${var.aws_secret_key}'
resource:
  aws_instance:
    web-server:
      instance_type: t2.nano
      tags:
        Name: '${var.koding_user_username}-${var.koding_group_slug}'
      user_data: |-
        echo "Pinging DB server at ${aws_instance.db-server.public_ip}"
        ping ${aws_instance.db-server.public_ip}

    db-server:
      instance_type: t2.nano
      tags:
        Name: '${var.koding_user_username}-${var.koding_group_slug}'

web-server instance

The first instance web server we run two commands under the user_data section

Commands under the user_data section run after VM boots and run as root user.

Explanation

1. Display the second instance db-server IP address

   echo "Pinging DB server at ${aws_instance.db-server.public_ip}"

   We make use of the public_ip attribute to display the second instance db-server IP address.

2. Ping the other instance

   ping ${aws_instance.db-server.public_ip}

   This is the actual ping command, we ping the second instance here using the same public_ip attribute.

Result

After building our stack, we can check web-server instance building logs to see the results of our user_data commands. As you can see the ping command ran successfully using the db-server IP as intended.

If you can not see the building logs of an instance, you can always click on VM settings (the three dots beside the VM name) and click on Show Logs

Quick note

VMs cannot ping each other in the same stack, as this will result in a Terraform cycle error. Terraform checks the stack template file to create a dependency list and therefore know which resources should be created first. If both instances are to ping each other this will create a cycle error and stack save will fail.