Going to be throwing more cycles at this today. Is there a specific MCP server I should be using? Also, would it make sense to throw an integration test together that uses all the vendoring features together that my PR is trying to solve for?

Ready again 🤞 @Erik Osterman (Cloud Posse)

also I think the windows integration testing failed.. not related to my PR

saw it failed in some others

https://github.com/cloudposse/atmos/actions/runs/24362882711/job/71147830095?pr=2322#step:16:62 some other PR

@Erik Osterman (Cloud Posse) @Andriy Knysh (Cloud Posse) it looks like with this fix, specifically with atmos now respecting hook events, there's a bug if someone is using

!terraform.state

lookups, or maybe other yaml templating to generate their backends specifically when using

after-

hooks

We ran into this because we just moved to using

!terraform.state

for backend generation 🫠

will be putting in an MR for fix

Some kind of race condition

basically on initial apply it generates the proper

backend.tf

file, succeeds. Then it goes to store SSM outputs:

hooks:
        store-outputs:
        events:
          - after-terraform-apply
        command: store
        outputs:
          rds_properties: .rds_properties

it actually regenerates all the files again, the backend.tf, and I need to double check the vars files, and it doesn't properly template the yaml functions, so it tries to init with like

!terraform.state

... as the bucket name for example