dbt test runs asynchronously as a repeated process.

What is implied by the "asynchronous" qualifier? I had an idea of adding a configuration option to enable running the test after the scheduled dbt_run, with the same selector.

Something like

- job_name: moderator_metrics
  scheduled: monthly
  models:
  - +active_moderators_monthly
  tags: []
  exclude: []
  run_test: True

which would translate into

dbt run --select +active_moderators_monthly

followed by

dbt test --select +active_moderators_monthly

Oh good point! I used "asynchronous" mostly to mean "not in CI/Gitlab", but you're right that this still leaves two options:

1. Run after the dbt run in Airflow, executing only the relevant tests.
2. Run in a completely independent DAG, executing tests for all models.

I don't have a super strong opinion there, but I feel like 2 is what gives us the biggest coverage for a reasonably low complexity, as it catches test failures caused by factors outside of the dbt run. But happy to be convinced otherwise!

Having discussed this again, I think what we have now implemented (configurable dbt test after dbt run for a set of models) is good enough to satisfy this milestone.

I think the remaining action here is to document how to configure this, in our main dbt documentation on Wikitech. Then we can close this :)

This is now documented at https://wikitech.wikimedia.org/wiki/Data_Platform/Systems/dbt/Schedule-dbt-on-Airflow#Run_tests_after_a_model, closing the milestone.