Allow for a diff created by a dry run to be later re-used without re-calculating it

[Kircheneer]

Environment

• Nautobot version: latest
• nautobot-ssot version: latest

Proposed Functionality

Using the dry-run functionality, we can generate a diff. It would be nice if we could then later apply that diff without recalculating it.

Use Case

This would allow for more sophisticated and deterministic approval processes, i.e. having a process like:

• generate diff with dry run
• validate diff manually and approve it for deployment
• deploy diff (and only that diff, not any changes that happened in the meantime)

Caveats

I believe care needs to be taken at what happens if the target systems state changes in the meantime. If it does, the diff might be corrupt?

[chadell]

I think that you will always need to get the actual state... what you could use it for is if there is an approval state, using a previous diff, you skip it (similar to what Terraform does with inputting a previous plan in the apply stage)

[Kircheneer]

Yeah terraform with the plan feature is something that mirrors this very well

[lampwins]

This would also open the door to distributed processing of the diff.

[Kircheneer]

This may be particularly interesting for 'outgoing' jobs, i.e. ones where we mutate the state of an external system (think a network device controller) using data from Nautobot.