Middleware for Django to support plop profiling
Django-plop is a middleware for profiling your views with PLOP, the Python Low Overhead Profiler.
Install it with pip (or easy_install, if that’s how you roll):
pip install django-plop
In your project settings:
MIDDLEWARE_CLASSES = MIDDLEWARE_CLASSES + ('django-plop.middleware.PlopMiddleware',) PLOP_DIR = os.path.join(PROJECT_ROOT, 'plop') # will be created, defaults to /tmp/plop
Then start your server with python manage.py --noreload (the --noreload is very important.) Hit a few pages, and then start the plop viewer like so:
python -m plop.viewer --datadir=plop
And navigate to localhost:8888 to view the profile results.
--noreload is used in development to allow the signalling features PLOP uses. This means it sadly won’t work with some alternate runserver implementations such as django-devserver. It will, however, work in prefork environments like Gunicorn.
PLOP itself is used in production by Dropbox with 2% CPU overhead (or so). However, this middleware will write logs to the disk, which may not be acceptable for your use case (especially on Heroku and other PaaSes)
Django-plop won’t install tornado for you (you need it to use the plop visualizer.) This is because you can use it in production, and you may not want to have tornado kicking around in your production environment. If you you want to have it installed for you, install with the “viewer” extras:
pip install django-plop[viewer]