PyNanny is a UNIX process management framework. It has a mini HTTP server for exposing statistics, logs and settings. It is useful for:

• Auto-daemonization.
• Periodic operations. Like cron, but per-application. Cron doesn't have access to your application's environment, nor is it aware of whether your process is running. Many real-world production periodic operations should only run when their application is running.
• Logging. STDOUT, STDERR are harvested by default - to rotated on-disk files and in-memory buffers accessible via the HTTP interface. Events such as health check failures, restarts and periodic operations are also logged.
• Statistics gathering.
• Automatically restarting processes when they crash or fail a health check. Extrmely useful when you have production processes which are known to need restarts (e.g. due to bugs, JVM/interpreter crashes, etc).
• Configuration of child environments.
• Management of groups of processes - a single nanny process can monitor as many children as you like. This is very usefull if you have e.g. multiple copies of the same server running on separate ports, and want a single process to control them.

The core of Nanny is written in C and is production tested. PyNanny is a Python binding combining the stable, performant C core with the extensibility and flexibility of Python.