Version information
Start using this module
Add this module to your Puppetfile:
mod 'autostructure-image_build', '0.2.8'
Learn more about managing modules with a PuppetfileDocumentation
image_build
Table of Contents
- Module description - What is the image_build module, and what does it do?
- Setup - The basics of getting started with image_build
- Usage - How to build Docker containers with Puppet
- Reference - Sample help output from the tool
- Limitations - OS compatibility, etc.
- Maintainers - who maintains this project
Module description
The basic purpose of image_build
is to enable building various images,
including Docker images, from Puppet code. There are two main cases
where this can be useful:
- You have an existing Puppet codebase and you're moving some of your services to using containers. By sharing the same code between container and non-container based infrastructure you can cut down on duplication of effort, and take advantage of work you've already done.
- You're building a lot of images, but scaling Dockerfile means either
a complex hierachy of images or copy-and-pasting snippets between
many individual Dockerfiles.
image_build
allows for sharing common functionality as Puppet modules, and Puppet itself provides a rich domain-specific language for declarative composition of images.
Setup
puppetlabs/image_build
is a Puppet Module and is available on the Forge.
The following should work in most cases:
puppet module install puppetlabs/image_build
You don't need any additional gems installed unless you are looking to
work on developing the module. All you need is a working Docker environment or
acbuild
, for which I'd recommend Docker for Mac or Docker for Windows
or just installing Docker if you're on Linux. For acbuild you can use
the rkt module.
Usage
With the module installed you should have access to two new puppet
commands; puppet docker
and puppet aci
. These have two subcommands,
one will trigger a build of an image, the other can be used to output
the intermediary dockerfile or shell script.
The examples directory contains a set of examples for experimenting with.
Simply open up examples/nginx
and run:
puppet docker build
The above is the simplest example of a build. Some settings are provided
in the accompanying metadata.yaml
file, while others are defaults
specific to the tool. You can change values in the metadata file (useful
for version control) or you can override those values on the command
line.
puppet docker build --image-name puppet/sample --cmd nginx --expose 80
See the full help page for other arguments for specifying different base images, setting a maintainer, using Rocker instead of Docker for the build and much more.
puppet docker build --help
You can also output the intermediary dockerfile using another subcommand. This is useful for both debugging and if you want to do something not natively supported by the tool.
puppet docker dockerfile
A hello world example - Nginx
Lets see a simple hello world example. We'll create a Docker image running Nginx and serving a simple text file.
First lets use a few Puppet modules from the Forge. We'll use the existing nginx module and we'll specify it's dependencies. We're also using dummy_service to ignore service resources in the Nginx module.
$ cat Puppetfile
forge 'https://forgeapi.puppetlabs.com'
mod 'puppet/nginx'
mod 'puppetlabs/stdlib'
mod 'puppetlabs/concat'
mod 'puppetlabs/apt'
mod 'puppetlabs/dummy_service'
Then lets write a simple manifest. Disabling nginx daemon mode isn't
supported by the module yet so we drop a file in place. Have a look at
manifests/init.pp
:
include 'dummy_service'
class { 'nginx': }
nginx::resource::vhost { 'default':
www_root => '/var/www/html',
}
file { '/var/www/html/index.html':
ensure => present,
content => 'Hello Puppet and Docker',
}
exec { 'Disable Nginx daemon mode':
path => '/bin',
command => 'echo "daemon off;" >> /etc/nginx/nginx.conf',
unless => 'grep "daemon off" /etc/nginx/nginx.conf',
}
And finally lets store the metadata in a file rather than pass on the
command line. Take a look at metadata.yaml
:
cmd: nginx
expose: 80
image_name: puppet/nginx
Now lets build a Docker image. Note that you'll need docker available on
your host to do so, along with the image_build
module installed.
puppet docker build
And finally lets run our new image. We expose the webserver on port 8080 to the local host.
$ docker run -d -p 8080:80 puppet/nginx
83d5fbe370e84d424c71c1c038ad1f5892fec579d28b9905cd1e379f9b89e36d
$ curl http://0.0.0.0:8080
Hello Puppet and Docker%
Involving hiera - Elasticsearch
The Elasticsearch example is similar to the above, with a few additional
features demonstrated. In particular the use of Hiera to provide
additional context for the Puppet build. You can find this in the
examples/elasticsearch
directory.
puppet docker build manifests/init.pp --image-name puppet/es --expose 9200 --cmd /docker-entrypoint.sh
Building multiple images from one manifest
One advantage of using Puppet for building Docker images is you are removed from the need to have a single Dockerfile per image. Meaning a single repository of Puppet code can be used to describe multiple images. This makes ensuring all images use (for example) the same repositories or same hardening scripts much easier to enforce. Change code in one place and rebuild multiple images.
Describing multiple images in Puppet is done using the existing node
resource in your manifest. For instance:
node 'node1' {
webserver { 'hello node 1': }
}
node 'node2' {
webserver { 'hello node 2': }
}
You can then select which image to build when running the build command,
by explicitly passing the image-name
.
puppet docker build --image-name puppet/node1
The match for the node resource in the Puppet code is done without the
repository name, in this case the puppet/
before node1
.
Note that you may want different metadata for different images.
image_build
will attempt to detect additional metadata in the
metadata
folder, and will merge items from metadata/metadata.yaml
with node specific metadata, for instance from metadata/node1.yaml
You can see an example of this settup in the examples/multi
directory.
Using a Puppet Master
The above examples all use local manifests copied to the image during
build, but image_build
also supports using a Puppet Master. You can
provide metadata via a local metadata file or directory, or by passing
command line arguments to the build command as shown in the examples
above. The only change is passing --master
like so.
puppet docker dockerfile --master puppet.example.com --image-name puppet/node1 --expose 80 --cmd nginx
The hostname passed to the Puppet Master will take the form node1.{datetime}.dockerbuilder. This means you can match on that pattern in your manifests, for instance like so:
node /^node1/ {
webserver { 'hello node 1': }
}
A worked example is provided in the examples/master
folder. You can
either upload this to an existing Puppet Master or Puppet Enterprise
install, or run a new local master using Docker.
First install the dependent modules into the local environment:
r10k puppetfile install --moduledir code/environments/production/modules
Then, from the examples/master
folder, use Docker to run an instance
of Puppet Server:
docker run --name puppet -P --hostname puppet -v $(pwd)/code:/etc/puppetlabs/code puppet/puppetserver-standalone
Determine the port on which the Puppet Server is exposed locally:
docker port puppet
You'll also need the IP address of your local machine. Replace the {ip} and {port} in the following with your own values.
puppet docker dockerfile --master {ip}:{port} --image-name puppet/node1 --expose 80 --cmd nginx
This should use the code on the Puppet Master to build the image.
Note that this currently assumes the requested certificate is autosigned in some way, expect more guidance on that in the future.
Minimizing image size with Rocker
image_build
supports using the
Rocker build tool in place of the
standard Docker build command. The Rocker output provides a little more
detail about the build process, but also allows for mounting of folders
at build time which minimizes the size of the resulting image.
puppet docker build --rocker
Note that when using Rocker the Puppet tools are not left in the final image, reducing it's file size.
Building ACI images
As well as Docker support, image_build
also experimentally supports building
ACI compatible
images for use with Rkt or other supported runtimes. This works in the
same manner as above. The following command should generate a shell
script which, when run, generates an ACI:
puppet aci script
And if you simply want to build the ACI directly you can just run:
puppet aci build
Reference
$ puppet docker --help
USAGE: puppet docker <action> [--from STRING]
[--maintainer STRING]
[--os STRING]
[--os-version STRING]
[--puppet-agent-version STRING]
[--r10k-version STRING]
[--expose STRING]
[--cmd STRING]
[--entrypoint STRING]
[--labels KEY=VALUE]
[--rocker]
[--[no-]inventory]
[--hiera-config STRING]
[--hiera-data STRING]
[--puppetfile STRING]
[--image-name STRING]
[--config-file STRING]
[--config-directory STRING]
[--master STRING]
Build Docker images and Dockerfiles using Puppet code
OPTIONS:
--render-as FORMAT - The rendering format to use.
--verbose - Whether to log verbosely.
--debug - Whether to log debug information.
--cmd STRING - The default command to be executed by the
resulting image
--config-directory STRING - A folder where metadata can be loaded from
--config-file STRING - A configuration file with all the metadata
--entrypoint STRING - The default entrypoint for the resulting
image
--expose STRING - A list of ports to be exposed by the
resulting image
--from STRING - The base docker image to use for the
resulting image
--hiera-config STRING - Hiera config file to use
--hiera-data STRING - Hieradata directory to use
--image-name STRING - The name of the resulting image
--[no-]inventory - Enable or disable the generation of an
inventory file at /inventory.json
--labels KEY=VALUE - A set of labels to be applied to the
resulting image
--maintainer STRING - Name and email address for the maintainer of
the resulting image
--master STRING - A Puppet Master to use for building images
--os STRING - The operating system used by the image if not
autodetected
--os-version STRING - The version of the operating system used by
the image if not autodetected
--puppet-agent-version STRING - Version of the Puppet Agent package to
install
--puppetfile STRING - Enable use of Puppetfile to install
dependencies during build
--r10k-version STRING - Version of R10k to use for installing modules
from Puppetfile
--rocker - Use Rocker as the build tool
ACTIONS:
build Build a Docker image from Puppet code
dockerfile Generate a Dockerfile which will run the specified Puppet code
See 'puppet man docker' or 'man puppet-docker' for full help.
Limitations
The module currently does not support building Windows containers, or building containers from a Windows machine. We'll be adding support for these in the future.
The inventory functionality does not work correctly on Centos 6 based
images, so if you're using Centos 6 then you need to pass the
--no-inventory
flag.
Maintainers
This repository is maintained by: Gareth Rushgrove gareth@puppet.com. The forked Autostructure repository is maintained by: Bryan Belanger bryan@autostructure.io
2017-01-24 Release 0.2.5
- Dockerfile switches to root user before work
2017-01-11 Release 0.2.4
- Keep ca-certificates package
2017-01-11 Release 0.2.3
- Fix template error
2017-01-11 Release 0.2.2
- Do not remove git-core
2016-12-02 Release 0.2.0
- Allow users to skip OS updates and Puppet install