Sessions at DjangoCon US 2011 about Python on Tuesday 6th September

Over the course of 6 years of Python development at NASA, Revsys, and Eldarion; Daniel Greenfeld has picked up a lot of easy tricks stolen from the best and brightest in the Python community that make him look good in front of his peers and clients. And now, at great risk to his career and reputation, he is going to show you how he does it.

Highlights:

• Get the most out of the least amount of work.
• Being forgetful is responsible for open source development.
• Using commonly available tools to make you look like you know what you are doing.
• All the sample code is accessible outside the slides

This session will cover the challenges of creating a production application performance monitoring system for Python. It includes an overview of the architecture of the system, as well as how it hooks into Django and captures details about web transactions, database transactions, memcache requests, exceptions and much more.

This session provides a technical review of the challenges associated with instrumenting Python for production monitoring. How do you place monitoring hooks into code that cannot be modified directly in order to collect meaningful performance data? Concepts covered in this session will include:

• Using context manager objects to time execution of code.
• Using context manager objects to catch details of exceptions.
• Applying a context manager to functions using wrappers and decorators.
• Applying a context manager to WSGI applications and components.
• How to go about applying the above to existing code using monkey patching.
• Deriving the name of a wrapped object for identification purposes.
• Using post import hooks as a means to trigger monkey patching.
• Driving the process of monkey patching from a configuration file.

The talk includes real-world examples of how these mechanisms have been used to add instrumentation to the Django web framework in order to collect actionable performance data.