pecan/docs/source/deployment.rst

.. _deployment:

Deploying Pecan in Production
=============================

There are a variety of ways to deploy a Pecan project to a production
environment.  The following examples are meant to provide *direction*,
not explicit instruction; deployment is usually heavily dependent upon
the needs and goals of individual applications, so your mileage will
probably vary.

.. note::

    While Pecan comes packaged with a simple server *for development use*
    (:command:`pecan serve`), using a *production-ready* server similar to the ones
    described in this document is **very highly encouraged**.

Installing Pecan
----------------

A few popular options are available for installing Pecan in production
environments:

*  Using `setuptools <https://pypi.python.org/pypi/setuptools>`_.  Manage
   Pecan as a dependency in your project's ``setup.py`` file so that it's
   installed alongside your project (e.g., ``python
   /path/to/project/setup.py install``).  The default Pecan project
   described in :ref:`quick_start` facilitates this by including Pecan as
   a dependency for your project.

*  Using `pip <http://www.pip-installer.org/>`_.
   Use ``pip freeze`` and ``pip install`` to create and install from
   a ``requirements.txt`` file for your project.

*  Via the manual instructions found in :ref:`Installation`.

.. note::
    Regardless of the route you choose, it's highly recommended that all
    deployment installations be done in a Python `virtual environment
    <http://www.virtualenv.org/>`_.

Disabling Debug Mode
--------------------

.. warning::
    One of the most important steps to take before deploying a Pecan app
    into production is to ensure that you have disabled **Debug Mode**, which
    provides a development-oriented debugging environment for tracebacks
    encountered at runtime.  Failure to disable this development tool in your
    production environment *will* result in serious security issues.  In your
    production configuration file, ensure that ``debug`` is set to ``False``.

    ::

        # myapp/production_config.py
        app = {
            ...
            'debug': False
        }

Pecan and WSGI
--------------

WSGI is a Python standard that describes a standard interface between servers
and an application.  Any Pecan application is also known as a "WSGI
application" because it implements the WSGI interface, so any server that is
"WSGI compatible" may be used to serve your application.  A few popular
examples are:

* `mod_wsgi <http://code.google.com/p/modwsgi/>`__
* `uWSGI <http://projects.unbit.it/uwsgi/>`__
* `Gunicorn <http://gunicorn.org/>`__
* `waitress <http://docs.pylonsproject.org/projects/waitress/en/latest/>`__
* `CherryPy <http://cherrypy.org/>`__

Generally speaking, the WSGI entry point to any Pecan application can be
generated using :func:`~pecan.deploy.deploy`::

    from pecan.deploy import deploy
    application = deploy('/path/to/some/app/config.py')

Considerations for Static Files
-------------------------------

Pecan comes with static file serving (e.g., CSS, Javascript, images)
middleware which is **not** recommended for use in production.

In production, Pecan doesn't serve media files itself; it leaves that job to
whichever web server you choose.

For serving static files in production, it's best to separate your concerns by
serving static files separately from your WSGI application (primarily for
performance reasons).  There are several popular ways to accomplish this.  Here
are two:

1.  Set up a proxy server (such as `nginx <http://nginx.org/en>`__, `cherokee
    <http://www.cherokee-project.com>`__, :ref:`cherrypy`, or `lighttpd
    <http://www.lighttpd.net/>`__) to serve static files and proxy application
    requests through to your WSGI application:

    ::

      <HTTP Client> ─── <Production/Proxy Server>, e.g., Apache, nginx, cherokee (0.0.0.0:80) ─── <Static Files>
                         │
                         ├── <WSGI Server> Instance e.g., mod_wsgi, Gunicorn, uWSGI (127.0.0.1:5000 or /tmp/some.sock)
                         ├── <WSGI Server> Instance e.g., mod_wsgi, Gunicorn, uWSGI (127.0.0.1:5001 or /tmp/some.sock)
                         ├── <WSGI Server> Instance e.g., mod_wsgi, Gunicorn, uWSGI (127.0.0.1:5002 or /tmp/some.sock)
                         └── <WSGI Server> Instance e.g., mod_wsgi, Gunicorn, uWSGI (127.0.0.1:5003 or /tmp/some.sock)


2.  Serve static files via a separate service, virtual host, or CDN.

Common Recipes
--------------

Apache + mod_wsgi
+++++++++++++++++

`mod_wsgi <http://code.google.com/p/modwsgi/>`_ is a popular Apache
module which can be used to host any WSGI-compatible Python
application (including your Pecan application).

To get started, check out the `installation and configuration
documentation
<http://code.google.com/p/modwsgi/wiki/InstallationInstructions>`_ for
mod_wsgi.

For the sake of example, let's say that our project, ``simpleapp``, lives at
``/var/www/simpleapp``, and that a `virtualenv <http://www.virtualenv.org>`_
has been created at ``/var/www/venv`` with any necessary dependencies installed
(including Pecan).  Additionally, for security purposes, we've created a user,
``user1``, and a group, ``group1`` to execute our application under.

The first step is to create a ``.wsgi`` file which mod_wsgi will use
as an entry point for your application::

    # /var/www/simpleapp/app.wsgi
    from pecan.deploy import deploy
    application = deploy('/var/www/simpleapp/config.py')

Next, add Apache configuration for your application.  Here's a simple
example::

    <VirtualHost *>
        ServerName example.com

        WSGIDaemonProcess simpleapp user=user1 group=group1 threads=5 python-path=/var/www/venv/lib/python2.7/site-packages
        WSGIScriptAlias / /var/www/simpleapp/app.wsgi

        <Directory /var/www/simpleapp/>
            WSGIProcessGroup simpleapp
            WSGIApplicationGroup %{GLOBAL}
            Order deny,allow
            Allow from all
        </Directory>
    </VirtualHost>

For more instructions and examples of mounting WSGI applications using
mod_wsgi, consult the `mod_wsgi Documentation`_.

.. _mod_wsgi Documentation: http://code.google.com/p/modwsgi/wiki/QuickConfigurationGuide#Mounting_The_WSGI_Application

Finally, restart Apache and give it a try.

uWSGI
+++++

`uWSGI <http://projects.unbit.it/uwsgi/>`_ is a fast, self-healing and
developer/sysadmin-friendly application container server coded in pure C.  It
uses the `uwsgi <http://projects.unbit.it/uwsgi/wiki/uwsgiProtocol>`__
protocol, but can speak other protocols as well (http, fastcgi...).

Running Pecan applications with uWSGI is a snap::

    $ pip install uwsgi
    $ pecan create simpleapp && cd simpleapp
    $ python setup.py develop
    $ uwsgi --http-socket :8080 --venv /path/to/virtualenv --pecan config.py

or using a Unix socket (that nginx, for example, could be configured to
`proxy to <http://projects.unbit.it/uwsgi/wiki/RunOnNginx>`_)::

    $ uwsgi -s /tmp/uwsgi.sock --venv /path/to/virtualenv --pecan config.py

Gunicorn
++++++++

`Gunicorn <http://gunicorn.org/>`__, or "Green Unicorn", is a WSGI HTTP Server for
UNIX. It’s a pre-fork worker model ported from Ruby’s Unicorn project. It
supports both eventlet and greenlet.

Running a Pecan application on Gunicorn is simple.  Let's walk through it with
Pecan's default project::

    $ pip install gunicorn
    $ pecan create simpleapp && cd simpleapp
    $ python setup.py develop
    $ gunicorn_pecan config.py


.. _cherrypy:

CherryPy
++++++++

`CherryPy <http://cherrypy.org/>`__ offers a pure Python HTTP/1.1-compliant WSGI
thread-pooled web server. It can support Pecan applications easily and even
serve static files like a production server would do.

The examples that follow are geared towards using CherryPy as the server in
charge of handling a Pecan app along with serving static files.

::

    $ pip install cherrypy
    $ pecan create simpleapp && cd simpleapp
    $ python setup.py develop

To run with CherryPy, the easiest approach is to create a script in the root of
the project (alongside ``setup.py``), so that we can describe how our example
application should be served. This is how the script (named ``run.py``) looks::

    import os
    import cherrypy
    from cherrypy import wsgiserver

    from pecan import deploy

    simpleapp_wsgi_app = deploy('/path/to/production_config.py')

    public_path = os.path.abspath(os.path.join(os.path.dirname(__file__), 'public'))

    # A dummy class for our Root object
    # necessary for some CherryPy machinery
    class Root(object):
        pass

    def make_static_config(static_dir_name):
        """
        All custom static configurations are set here, since most are common, it
        makes sense to generate them just once.
        """
        static_path = os.path.join('/', static_dir_name)
        path = os.path.join(public_path, static_dir_name)
        configuration = {
            static_path: {
                'tools.staticdir.on': True,
                'tools.staticdir.dir': path
            }
        }
        return cherrypy.tree.mount(Root(), '/', config=configuration)

    # Assuming your app has media on different paths, like 'css', and 'images'
    application = wsgiserver.WSGIPathInfoDispatcher({
        '/': simpleapp_wsgi_app,
        '/css': make_static_config('css'),
        '/images': make_static_config('images')
        }
    )

    server = wsgiserver.CherryPyWSGIServer(('0.0.0.0', 8080), application,
    server_name='simpleapp')

    try:
        server.start()
    except KeyboardInterrupt:
        print "Terminating server..."
        server.stop()

To start the server, simply call it with the Python executable::

    $ python run.py