Add useful logging to a small application

Use this guide when a script or small web app runs somewhere you cannot watch it, such as a server or scheduled job, and you struggle to work out what went wrong. The examples use Python's built-in logging module, but the ideas apply to any language.

Step by step

1. Decide what questions logs must answer

   Write down three questions you wish you could answer after a failure, such as: Which customer request failed? What input caused it? How long did the job take? Logging should answer these, not record every line the program runs. This list stops logs becoming noise that nobody reads.