packer-cn/website/source/intro/getting-started/remote-builds.html.markdown

layout	page_title	prev_url	next_url	next_title	description
intro	Remote Builds and Storage	/intro/getting-started/vagrant.html	/intro/getting-started/next.html	Next Steps	Up to this point in the guide, you have been running Packer on your local machine to build and provision images on AWS and DigitalOcean. However, you can use Atlas by HashiCorp to both run Packer builds remotely and store the output of builds.

Remote Builds and Storage

Up to this point in the guide, you have been running Packer on your local machine to build and provision images on AWS and DigitalOcean. However, you can use Atlas by HashiCorp to run Packer builds remotely and store the output of builds.

Why Build Remotely?

By building remotely, you can move access credentials off of developer machines, release local machines from long-running Packer processes, and automatically start Packer builds from trigger sources such as vagrant push, a version control system, or CI tool.

Run Packer Builds Remotely

To run Packer remotely, there are two changes that must be made to the Packer template. The first is the addition of the push configuration, which sends the Packer template to Atlas so it can run Packer remotely. The second modification is updating the variables section to read variables from the Atlas environment rather than the local environment. Remove the post-processors section for now if it is still in your template.

{
  "variables": {
    "aws_access_key": "{{env `aws_access_key`}}",
    "aws_secret_key": "{{env `aws_secret_key`}}"
  },
  "builders": [{
    "type": "amazon-ebs",
    "access_key": "{{user `aws_access_key`}}",
    "secret_key": "{{user `aws_secret_key`}}",
    "region": "us-east-1",
    "source_ami": "ami-9eaa1cf6",
    "instance_type": "t2.micro",
    "ssh_username": "ubuntu",
    "ami_name": "packer-example {{timestamp}}"
  }],
  "provisioners": [{
    "type": "shell",
    "inline": [
      "sleep 30",
      "sudo apt-get update",
      "sudo apt-get install -y redis-server"
    ]
  }],
  "push": {
    "name": "ATLAS_USERNAME/packer-tutorial"
  }
}

To get an Atlas username, create an account here. Replace "ATLAS_USERNAME" with your username, then run packer push -create example.json to send the configuration to Atlas, which automatically starts the build.

This build will fail since neither aws_access_key or aws_secret_key are set in the Atlas environment. To set environment variables in Atlas, navigate to the operations tab, click the "packer-tutorial" build configuration that was just created, and then click 'variables' in the left navigation. Set aws_access_key and aws_secret_key with their respective values. Now restart the Packer build by either clicking 'rebuild' in the Atlas UI or by running packer push example.json again. Now when you click on the active build, you can view the logs in real-time.

-> Note: Whenever a change is made to the Packer template, you must packer push to update the configuration in Atlas.

Store Packer Outputs

Now we have Atlas building an AMI with Redis pre-configured. This is great, but it's even better to store and version the AMI output so it can be easily deployed by a tool like Terraform. The atlas post-processor makes this process simple:

{
 "variables": ["..."],
 "builders": ["..."],
 "provisioners": ["..."],
 "push": ["..."],
 "post-processors": [
   {
     "type": "atlas",
     "artifact": "ATLAS_USERNAME/packer-tutorial",
     "artifact_type": "aws.ami"
   }
 ]
}

Update the post-processors block with your Atlas username, then packer push example.json and watch the build kick off in Atlas! When the build completes, the resulting artifact will be saved and stored in Atlas.