scheduler

Apache Airflow version

2.2.4 (latest released)

What happened

When triggering this dag below it runs when it should fail. A set is being passed to default_args instead of what should be a dictionary yet the dag still succeeds.

What you expected to happen

I expected the dag to fail as the default_args parameter should only be a dictionary.

How to reproduce

from airflow.mo

Right now, we've got this chunk of code:

def _columns_to_markdown(columns: Mapping[str, Any]) -> str:
    return (
        textwrap.dedent(
            """
        | Name | Description |
        | - | - |
    """
        )
        + "\n".join([f"| {name} | {metadata['description']}" for name, metadata in columns.items()])
    )

Instead of representing them with markdown, we

Is there a way to make a job re-occur N milliseconds after it finished previous time?

The tests fail sometimes due to millisecond based timing on some of the tests. I over optimized for local testing and it's causing tests to fail in CI occasionally. The CI pipeline is running in docker with limited resources, so not able to handle quite as precise time as my local machine.