hi all, it seems that the Metaflow-to-Airflow conversion process loses any k8s tolerations that were configured on the steps. I was poking at the codebase and it looks like this function does not actually configure tolerations anywhere? Someone please correct me if I am wrong or if there is another step where they should be configured.

def _kubernetes_pod_operator_args(operator_args):
    from kubernetes import client

    from airflow.kubernetes.secret import Secret

    # Set dynamic env variables like run-id, task-id etc from here.
    secrets = [
        Secret("env", secret, secret) for secret in operator_args.get("secrets", [])
    ]
    args = operator_args
    args.update(
        {
            "secrets": secrets,
            # Question for (savin):
            # Default timeout in airflow is 120. I can remove `startup_timeout_seconds` for now. how should we expose it to the user?
        }
    )

yes - that's expected. we haven't ported the entire feature set to airflow yet, but it is easy to add in. @thankful-ambulance-42457 can you look into this?

cc: @red-accountant-23764

Yes please - much appreciated thankyou