Ponder the idea of moving the dynamic_inventory.py script into an application that has entry points and a library that can be extended to support multiple storage backends; ie sql, reddis, json...
Presently the dynamic_inventory.py script is a bit one dementional. while it works well enough to build / scale out an infrastructure of both physical and container nodes it could be better developed, implement testing, and provide for a user importable API.
- The present dynamic_inventory.py would be changed to a wrapper script / function.
- The new lib would be installable with the ability to provide / scale inventory as well as manage it.
leave things the way they are.
No impact on the library, though providing multiple storage backends for the library would open the inventory storage to possible attack vectors assuming the user used a shared storage system as the backend.
End User Impact
The deployer would have to be aware of the inventory library in the sense that they would need to install it. Besides the added dependencies there shouldnt be a specific deployer impact.
Added software dependencies.
The new lib / entry point scripts would need to be documented.