Deploying Python WSGI applications can be a bit tedious. My preferred way of hosting such applications on Debian GNU/Linux involves:

1. Checking out a copy of the application code.
2. Installing Virtualenv and creating a virtualenv for the application.
3. Installing all dependencies of the application from a requirements file into the virtualenv.
4. Installing Gunicorn into the virtualenv for the application.
5. Create a init script which starts Gunicorn with the application.
6. Installing Monit and configure it to monitor the Gunicorn process.
7. Installing and configuring Nginx.
8. Setting up a virtual host in Nginx which proxies through a UNIX socket to the Gunicorn process.

Enter Puppet -- a declarative configuration management tool. With Puppet you describe how you want the setup of your stack to look like. Puppet can then recreate that stack from any known starting point. Contrast this with automation tools like Fabric which is imperative in nature. So in stead of saying "Install that package, delete that file, run that script" with imperative tools your conversation looks more like "I have a machine with this package and its started on boot with this user" with declarative tools like Puppet.

Using Puppet and some Puppet modules I recently created the process is distilled to the following steps for a WSGI blog application:

1. Place your applications source code including a requirements.txt under /usr/local/src/blog.

2. Install Puppet:

   apt-get install puppet
   

3. Paste the following into a puppet manifest file (blog.pp):

    Exec {
      path => "/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
    }
   
    include webapp::python
   
    webapp::python::instance { "blog":
      domain => "blog.uggedal.com",
      wsgi_module => "blog:app",
      requirements => true,
    }
   

4. Clone the following Puppet modules into a modules directory:

    mkdir modules
    git clone git://github.com/puppetmodules/puppet-module-webapp.git modules/webapp
    git clone git://github.com/puppetmodules/puppet-module-python.git modules/python
    git clone git://github.com/puppetmodules/puppet-module-monit.git modules/monit
    git clone git://github.com/puppetmodules/puppet-module-nginx.git modules/nginx
   

5. Run the file with Puppet (as root or with sudo):

    puppet apply --modulepath=modules blog.pp
   

Note that running this Puppet manifest as root will potentially overwrite files you already have configured on your machine like:

• /etc/nginx/nginx.conf
• /etc/monit/monitrc

Do you need a database? Get PostgreSQL installed, configured, and a user and database created by:

1. Add the following to blog.pp:

    include postgresql::server
    include postgresql::python
   
    postgresql::database { "blog":
      owner => "bloguser",
    }
   

2. Clone my PostgreSQL Puppet module:

    git clone git://github.com/puppetmodules/puppet-module-postgresql.git modules/postgresql
   

3. Run Puppet again:

    puppet apply --modulepath=modules blog.pp
   

Take a look at the README of my Puppet webapp module for more detailed instructions including how to use it with Django applications. You should also check out the modules used by the webapp module: Nginx module, Monit module, and Python module.