@Andriy Knysh (Cloud Posse) https://github.com/cloudposse/atmos/pull/2093

@Esteban Angee Déjà vu? @Erik Osterman (Cloud Posse) silly question. Is it possible for this issue to happen in CI but not locally?

It would be odd, but this PR fixes it

Sounds good. The reason I ask is because our team is currently working to decouple docker from atmos for everyday development as well as distribution of our service catalog. But here's the real kicker…in our unit testing, where the issue happens in CI, none of the components run in parallel. The scenarios, stacks, do run parallel. We also tried running

git clean -fdx

against the test directory before running to ensure there's no files to cleanup (this is in addition to the unit testing running

atmos terraform clean

for each stack as part of our initialization)

Confirming you enabled the workdir setting?

Explicitly, no. Can you provide documentation?

Oh that's the problem; and the above issue is related to that feature.

So I need to add the following to every component in my catalog?

provision:
        workdir:
          enabled: true

No, you can definitely set it globally.

like this?

terraform:
  provision:
    workdir:
      enabled: true

However, to your point, when this pull request is merged, there was a bug where it didn't work when you set it globally.

let me check

Oh, great. I was coming here to ask if there were any reasons NOT to always enable workdirs. If none, why not make it the default?

We should probably make it the default in a future release, once we get enough validation