Snapshot Creation with Packer

Brandon Rozek

May 8, 2020

Packer is a tool to create automated machine images in both local virtual machine / container environments, as well as a variety of cloud platforms. My current cloud platform of choice is Digital Ocean, so this post will explain how to set up Packer with it. Though you can likely find your platform of choice on their docs page

In this post I am going to use the beta configuration language of HCL2. This requires a Packer version of at least 1.5.

Variables

First let us set up a variables file which we will later reference. This makes it easy to keep your main Packer configuration files in Git, while not committing your API key.

Create a file called variables.auto.pkrvars.hcl. Here is some example variables and values that I put in mine.

base_system_image = "ubuntu-20-04-x64"
region = "nyc3"
size = "512mb"
do_token = "DOTOKENHERE" # Secret

Then we need to create variables.pkr.hcl that define the types of each of these variables

variable "do_token" {
  type = string
}

variable "base_system_image" {
  type = string
}

variable "region" {
  type = string
}

variable "size" {
  type = string
}

Provisioning

Once the system is up and running we can use a variety of tools that setup the image.

I’ll use a simple bash script as an example

#!/bin/bash
apt update
apt upgrade -y

Piecing it together

Finally let us create a do.pkr.hcl file that contains the following information

source "digitalocean" "web" {
  api_token = var.do_token
  image = var.base_system_image
  region = var.region
  size = var.size
  ssh_username = "root"
  snapshot_name = "packer-example"
}


build {
  sources = [
    "source.digitalocean.web",
  ]

  provisioner "shell" {
    scripts = [ "setup.sh" ]
  }
}

Assuming all the files we’ve just created are in the same directory, Packer will automatically recognize where the value for var.do_token lives. We can then run packer to build the image

packer build .

This sets up a snapshot called packer-example which we can later spin up and use! Keep in mind that it does a small amount of money to store images on DigitalOcean.