Rump helps you run Puppet locally against a Git checkout.
Rump supports a Puppet workflow where you quickly + iteratively develop your Puppet manifests on a single machine, then push your changes up to a repository to deploy to the rest of your infrastructure.
This workflow also complements a Capistrano or MCollective-style deployment, where you remotely instruct Rump to check out a copy of the latest manifests and run them.
Rump also has the ability to freeze Puppet in to the manifests repository, letting you quickly test different versions of Puppet without waiting for packages to appear, and reducing the dependencies on a system to run Puppet down to just Ruby and git.
On the server you're configuring with Puppet, run:
$ sudo aptitude install puppet git-core
Make sure your hostname is set:
$ sudo vi /etc/hostname foo.bar.example.org $ sudo hostname -F /etc/hostname
Make sure you check out the man pages.
Check out your repository of Puppet manifests:
$ rump clone firstname.lastname@example.org:me_at_example_dot_org/puppet.git
cd into the directory, and do a Puppet run:
$ sudo rump go
There's nothing stopping you from running Rump against different checkouts/branches of manifests. This is especially powerful when developing locally with the following workflow:
rump clone email@example.com:me_at_example_dot_org/puppet.git
cd puppet && git checkout -b new_feature
- Make your changes &&
- Iterate until everything's working
git checkout master && git merge new_feature
When you make changes, run Puppet through Rump:
$ sudo rump go
You can append options you'd normally pass to the
puppet command at the end
$ sudo rump go --verbose --debug --noop
Alternatively, if you want to live on the bleeding edge and eschew your distribution's packages, you can run Rump entirely from RubyGems or Git. This gives you a whole bunch of advantages:
- You only need Ruby installed on your system to run Puppet
- A checkout of your manifests also gives you Puppet
- Bind your manifests to a particular version of Puppet
- Test your manifests against new versions of Puppet in a separate branch
- Manage upgrades of Puppet outside your operating system's release cycle
You can freeze Puppet and it's dependencies very easily:
$ rump freeze
This will freeze Puppet + Facter under
vendor/, using Bundler.
When you run
rump go, Rump checks whether you have frozen Puppet + Facter, and
runs the frozen Puppet if available.
You can manage the versions of Puppet you want frozen using the
the root of your project. To use a specific version of Puppet, edit your
``` ruby source :rubygems gem "puppet", "2.6.4" ```
If you want to live on the bleeding edge, you can run Puppet out of git:
``` ruby source :rubygems gem "puppet", "2.6.7", :git => "git://github.com/puppetlabs/puppet.git", :tag => "2.7.0rc1" gem "facter", "1.5.8", :git => "git://github.com/puppetlabs/facter.git", :tag => "1.5.9rc5" ```
Any dependency you bundle will automatically be added to the load path when you
There's a suite of Cucumber tests to fully exercise Rump in
$ cucumber features/
The scenarios are tagged with
@online, depending on whether
the test requires internet connectivity. Run all but
@online tests with:
$ cucumber --tags ~@online features/
Puppet's fileserver (
source => "puppet:///..."on File resources) doesn't behave as expected on Puppet < 2.6. If you are using Puppet < 2.6, all files need to be templates.
Manifests need to be in modules so Puppet can pick them up correctly.