ploybot is a Fabric_ based tool for setting up Python WSGI apps on what I like to
call the SNUG stack:
* Supervisord_ for starting processes and keeping them alive.
* Nginx_ for proxying between your WSGI app and the big bad web.
* Ubuntu_ as the OS of choice, enabling resolution of system dependencies with
apt. Debian might work as well but hasn't been tested.
* Gunicorn_ for serving your WSGI app.
(I suppose it could also be the GUNS stack but that sounds far less friendly.)
* Deploy your site to one or more servers with a single command ('ploybot push').
* Automatic configuration of Nginx, Supervisord, and Gunicorn to get your site running.
* Isolation of each site into a separate Virtualenv_
* Support for differing app config depending on which role you deploy to (a
different DB in staging than production, for example).
pip install ploybot
You can also install the current development version straight from bitbucket::
pip install hg+http://bitbucket.org/dori/ploybot#egg=ploybot
(Almost) all of the commands below require that you specify a role name, like
'ploybot dosomething -R dev'.
Commands can generally be run from the site root directory or any subdirectory
ploybot push -R rolename
This command is the main reason for ploybot's existence. It does the work required
to get your app running on a host (or set of hosts) given the configuration
specified in site.yaml and the selected role .yaml file. 'push' does the
1. SSHes to the remote server(s) specified in the role config.
2. Creates a zipped up rollback archive of the old site if there's one
3. Creates a virtualenv for the site.
4. Installs apt and python dependencies.
5. Copies the site from your local machine to a temporary directory on the
6. Writes config file includes for nginx and supervisord.
7. Moves your code from the temp dir into its production location
(/srv/<sitename> by default).
8. Tells nginx and supervisord to reload their configs.
ploybot rollback -R rolename
This command is for when you have those 'OMG I BROKE THE SITE' moments. It will
SSH to the push_hosts specified in your role file and restore the most recent
archive of the site. ploybot keeps 3 rollback copies of your site, so you could
potentially run 'ploybot rollback' 3 times to go back to the state from 3
ploybot run -R rolename
This command runs the site from the local machine, on port 8000. (Nothing is
pushed of copied.) Static directories listed in the *static_dirs* section of
site.yaml will also be served. (CherryPy is used for this magic.)
ploybot install_server_deps -R rolename
When you get a shiny new server with that fresh Ubuntu smell, it needs just a
tiny bit of setup before it will know how to serve ploybot-deployed sites. This
command does that. It installs nginx and supervisord, and gives each of them a
wildcard include in their configs for loading from /srv/<sitename>/conf.
This command wraps 'pip install' to install all of the python packages listed
in requirements.txt into your local python environment. It's handy for grabbing all
the dependencies when you're working with a new virtualenv on an existing
Creates a virtual environment for the app to be deployed.
Reloads nginx config
Kills other running proccesses of this app
ploybot skel sitename
Creates a directory with a basic ploybot file and directory structure.
A ploybot-enabled project should be layed out something like this::
│ ├── dev.yaml
│ ├── staging.yaml
│ └── production.yaml
Some of those files/folders are required, other are optional:
1. site.yaml - This is the main config file (comparable to app.yaml in Google
2. deps.yaml - Lists Python packages, Ubuntu apt packages, and apt build
dependencies that need to be installed on the server running your site.
3. fabfile.py - A Fabric_-compatible fabfile that imports ploybot's Fabric
4. roles/\*.yaml - One or more 'role' files that contain the config to be
passed into your app depending on the deployment context.
All of the required files will be created for you with the 'ploybot skel' command.
1. membrane.py - For Django projects it's nice to have a little shim to expose
the project as a WSGI app. I like to call mine membrane.py. You can use
whatever you like, or nothing at all, depending on your setup.
2. my-django-project - ploybot isn't restricted to Django; any valid WSGI app on
your Python path should be servable. But for Django projects I think it
makes sense to stick them right there.
.. _Supervisord: http://supervisord.org/
.. _Nginx: http://nginx.org/
.. _Ubuntu: http://www.ubuntu.com/
.. _Gunicorn: http://gunicorn.org/
.. _Fabric: http://docs.fabfile.org/
.. _Virtualenv: http://virtualenv.openplans.org/