Go to file
Eran Chetz 569121519d minor change that was majorly bugging me 2015-09-21 16:32:18 +03:00
builder Fixed GCE builder after dependency change. 2015-09-20 10:14:07 +02:00
command Fixes #2667: Added vmware-rename to fix command help text 2015-08-24 20:19:49 +02:00
common Added [DEBUG] prefix to log messages 2015-08-19 13:15:23 -07:00
communicator Added a warning log so we can diagnose failure cases 2015-07-15 12:29:42 -07:00
contrib/zsh-completion
fix Enable headless mode by default on Parallels Desktop 11 2015-08-24 15:09:29 +02:00
helper Add s.SSHPort variable as the port WinRM uses to connect. This is needed on any builder where the port used to connect is not the guest winrm port but a nated port on the host. 2015-07-27 09:57:38 +12:00
packer Fix govet issues 2015-08-05 19:41:29 -07:00
plugin Added an artifice post-processor which allows you to override artifacts in a post-processor chain 2015-08-07 18:21:23 -07:00
post-processor Merge branch 'master' into artifice 2015-08-18 11:57:06 +00:00
provisioner Download chef from chef.io 2015-08-25 21:53:50 -07:00
scripts Overhaul the Makefile 2015-08-21 18:25:58 -07:00
template helper/config: copy buildname/buildtype properly 2015-06-30 10:42:55 -07:00
test
website minor change that was majorly bugging me 2015-09-21 16:32:18 +03:00
.gitignore
.travis.yml Allow failures in 1.5 since the build scripts don't support switching for 2015-08-22 14:14:58 -07:00
CHANGELOG.md Add 0.8.6 release date to the changelog 2015-08-22 13:43:33 -07:00
CONTRIBUTING.md Fix go version in docs 2015-07-15 13:54:41 +03:00
LICENSE
Makefile Add deps pre-target and remove testrace sadness from ci 2015-08-21 18:37:51 -07:00
README.md
Vagrantfile
appveyor.yml Added go vet and git rev-parse head to appveyor so we can see what we're actually building / testing 2015-08-06 12:24:13 -07:00
checkpoint.go
commands.go
config.go
config_unix.go
config_windows.go
log.go
main.go
main_test.go
panic.go
signal.go
stdin.go
version.go up version for dev 2015-08-26 21:24:47 -07:00

README.md

Packer

Build Status Windows Build Status

Packer is a tool for building identical machine images for multiple platforms from a single source configuration.

Packer is lightweight, runs on every major operating system, and is highly performant, creating machine images for multiple platforms in parallel. Packer comes out of the box with support for the following platforms:

  • Amazon EC2 (AMI). Both EBS-backed and instance-store AMIs
  • DigitalOcean
  • Docker
  • Google Compute Engine
  • OpenStack
  • Parallels
  • QEMU. Both KVM and Xen images.
  • VirtualBox
  • VMware

Support for other platforms can be added via plugins.

The images that Packer creates can easily be turned into Vagrant boxes.

Quick Start

Note: There is a great introduction and getting started guide for those with a bit more patience. Otherwise, the quick start below will get you up and running quickly, at the sacrifice of not explaining some key points.

First, download a pre-built Packer binary for your operating system or compile Packer yourself.

After Packer is installed, create your first template, which tells Packer what platforms to build images for and how you want to build them. In our case, we'll create a simple AMI that has Redis pre-installed. Save this file as quick-start.json. Be sure to replace any credentials with your own.

{
  "builders": [{
    "type": "amazon-ebs",
    "access_key": "YOUR KEY HERE",
    "secret_key": "YOUR SECRET KEY HERE",
    "region": "us-east-1",
    "source_ami": "ami-de0d9eb7",
    "instance_type": "t1.micro",
    "ssh_username": "ubuntu",
    "ami_name": "packer-example {{timestamp}}"
  }]
}

Next, tell Packer to build the image:

$ packer build quick-start.json
...

Packer will build an AMI according to the "quick-start" template. The AMI will be available in your AWS account. To delete the AMI, you must manually delete it using the AWS console. Packer builds your images, it does not manage their lifecycle. Where they go, how they're run, etc. is up to you.

Documentation

Full, comprehensive documentation is viewable on the Packer website:

http://www.packer.io/docs

Developing Packer

If you wish to work on Packer itself or any of its built-in providers, you'll first need Go installed (version 1.4+ is required). Make sure Go is properly installed, including setting up a GOPATH.

Next, install the following software packages, which are needed for some dependencies:

Then, install Gox, which is used as a compilation tool on top of Go:

$ go get -u github.com/mitchellh/gox

Next, clone this repository into $GOPATH/src/github.com/mitchellh/packer. Install the necessary dependencies by running make updatedeps and then just type make. This will compile some more dependencies and then run the tests. If this exits with exit status 0, then everything is working!

$ make updatedeps
...
$ make
...

To compile a development version of Packer and the built-in plugins, run make dev. This will put Packer binaries in the bin folder:

$ make dev
...
$ bin/packer
...

If you're developing a specific package, you can run tests for just that package by specifying the TEST variable. For example below, only packer package tests will be run.

$ make test TEST=./packer
...

Acceptance Tests

Packer has comprehensive acceptance tests covering the builders of Packer.

If you're working on a feature of a builder or a new builder and want verify it is functioning (and also hasn't broken anything else), we recommend running the acceptance tests.

Warning: The acceptance tests create/destroy/modify real resources, which may incur real costs in some cases. In the presence of a bug, it is technically possible that broken backends could leave dangling data behind. Therefore, please run the acceptance tests at your own risk. At the very least, we recommend running them in their own private account for whatever builder you're testing.

To run the acceptance tests, invoke make testacc:

$ make testacc TEST=./builder/amazon/ebs
...

The TEST variable is required, and you should specify the folder where the backend is. The TESTARGS variable is recommended to filter down to a specific resource to test, since testing all of them at once can sometimes take a very long time.

Acceptance tests typically require other environment variables to be set for things such as access keys. The test itself should error early and tell you what to set, so it is not documented here.