Puppet Python Module with Virtualenv, Pip and Gunicorn support

Module for configuring Python with virtualenvs and installation of packages inside them with pip in addition to serving Python WSGI applications including Django through Gunicorn.

This module support installing packages specified in a requirements.txt file and update said packages when the file changes. This way you only have to define your requirements in one place: in the VCS for your application code.

Tested on Debian GNU/Linux 6.0 Squeeze and Ubuntu 10.4 LTS with Puppet 2.6. Patches for other operating systems welcome.

TODO

• Use /etc/gunicorn.d/ for instance configs and simplify /etc/init.d/gunicorn. Possibly use a single gunicorn init script like in Debian.
• Uninstallation of packages no longer provided in the requirements file.

Installation

Clone this repo to a python directory under your Puppet modules directory:

git clone git://github.com/uggedal/puppet-module-python.git python

If you don't have a Puppet Master you can create a manifest file based on the notes below and run Puppet in stand-alone mode providing the module directory you cloned this repo to:

puppet apply --modulepath=modules test_python.pp

Usage

To install Python with development dependencies simply include the module:

include python::dev

You can install a specific version of Python by including the module with this special syntax:

class { "python::dev": version => "2.5" }

Note that classes in Puppet are singletons and not more than one can be created even if you provide different paramters to them. This means that the python::dev class can only be used to install one version. If you need more coexising versions you could create a new class based on the current one prefixed with the actual version.

Virtualenv

To install and configure virtualenv, include the module:

include python::venv

You can also provide an owner and group which will be the owner of the virtualenv files by including the class with this special syntax:

class { "python::venv": owner => "www-mgr", group => "www-mgr" }

Setting up a virtualenv is done with the python::venv::isolate resource:

python::venv::isolate { "/usr/local/venv/mediaqueri.es": }

Note that you'll need to define a global search path for the exec resource to make the python::venv::isolate resource function properly. This should ideally be placed in manifests/site.pp:

Exec {
  path => "/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
}

If you have several version of Python installed you can specifiy which interpreter you'd like the virtualenv to contain:

python::venv::isolate { "/usr/local/venv/mediaqueri.es":
  version => "2.5",
}

If you point to a pip requirements file Puppet will install the specified packages and upgrade them when the file changes:

python::venv::isolate { "/usr/local/venv/mediaqueri.es":
  requirements => "/var/www/mediaqueri.es/requirements.txt",
}

Gunicorn

To use Gunicorn for serving WSGI applications you'll first have to include its class which manages a directory for storing pid files and unix sockets for all your Gunicorn instances:

include python::gunicorn

If you don't want all your Gunicorn instances running as root you should specify an unprivileged user when including the Gunicorn class:

class { "python::gunicorn": owner => "www-mgr", group => "www-mgr" }

Serving a WSGI application is done by using the following definition type and specifying a virtualenv, the source of your application code and the WSGI application module:

python::gunicorn::instance { "blog":
  venv => "/usr/local/venv/blog",
  src => "/usr/local/src/blog",
  wsgi_module => "blog:app",
}

A Django application does not need a WSGI application module argument:

python::gunicorn::instance { "cms":
  venv => "/usr/local/venv/cms",
  src => "/usr/local/src/cms",
  django => true,
}

You can optionally provide a specific settings file to use with Django:

python::gunicorn::instance { "cms":
  venv => "/usr/local/venv/cms",
  src => "/usr/local/src/cms",
  django => true,
  django_settings => "settings_production.py",
}

The gunicorn instance resource installs the latest gunicorn into the virtualenv the first time it's created. If you need a specific version simply provide a version argument:

python::gunicorn::instance { "cms":
  venv => "/usr/local/venv/cms",
  src => "/usr/local/src/cms",
  django => true,
  version => "0.12.1",
}