ApMan is a python wrapper which can be used to call other python applications (e.g. analytics packages like a customer churn model) in a standardised, robust, and logged manner - and email notifications in the event of an issue.

When ApMan is used to run a package:

• The initiation of the package is recorded to a central log
• The package is run in a seperate process, started by ApMan
• Typically, inside this process, a package dedicated python virtual environment is activated, and then the package work is done.
• If the package goes over a pre-configured package timeout, the process is killed, and the timeout result is recorded to the central log.
• If the package causes an unhandled exception, the process is killed, and the exception result recorded to the central log.
• On successful completion, the success result is recorded to the central log.
• In all cases, all logging (stdout/stderr) from the package is also recorded to the central log.
• Finally, in the event of timeout or exception (or on success, if so configured) ApMan emails the configured recipients.

• Packages are constructed as Python applications, with an additional 'apman' configuration file.
• It is also expected and supported that each package should have a requirements.txt file which can be used to build a dedicated virtual environment. The apman config is then updated to direct ApMan to use this environment on execution.

The config file should be a json dictionary such as the following:

{ "id":"test", "description":"this is a test package", "timeout": 4, "command": "venv/Scripts/python script.py", "notify-success": True "parameters": {"name":"Will", "age":28} }

• Once installed, ApMan can be called using the 'apman_run.py' command line tool.
• To run a package call 'apman_run.py config_file'.