dumping all variables to a file for a reproducible run

Hi all
I’m using Ansible playbooks to run tests on EC2.
Most of my variables are define at group_vars/all, but I do override some of them using -e CLI option.
To make the test reproducible, I want to dump all the vars into a local file.
What is the idiomatic way to do so?

One solution I explored was using hostvars, which do hold all the relevant information, but also all kind of extra info I do not need like localhost info.
Is it the right path?
How can I filter only the top level, non nested vars from it?

followup question:
How to dump the vars in a format I can later easily include to reproduce the test?

Thanks!

I would try to go by including variables from a file with --extra-vars “@some_file.json”: http://docs.ansible.com/playbooks_variables.html#passing-variables-on-the-command-line
Then you wouldn’t need to worry about dumping playbook runtime variables from different nesting layers. You would always have one (or more) consistent files that you could use later to reproduce the tests.

Thanks Dan, its a good solution, but I’m kind of use (and like) passing parameters in the command line.
Since all variables are in hostvars, I probably can build a custom filter to extract them (or something).