AIP-84 Grid failing to load on deleted TIs

[pierrejeambrun]

How to reproduce

Run this dag and let it settle (fail or success)

def try_something():
    if randint(0, 9) <= 2:
        return
    raise ValueError("Task Failed")


with DAG(
    dag_id="simple_with_tries",
    schedule="0 0 * * *",
    dagrun_timeout=datetime.timedelta(minutes=10),
    tags=["example2"],
    params={"example_key": "example_value"},
    catchup=False,
) as dag:
    run_this = PythonOperator(
        task_id="python_operator1", python_callable=try_something, retries=5, retry_delay=10
    )

    run_this_2 = PythonOperator(
        task_id="python_operator2", python_callable=try_something, retries=5, retry_delay=10
    )

Then edit the dag to create a new version, that does not contain the python_operator2 task, for instance by commenting it:

def try_something():
    if randint(0, 9) <= 2:
        return
    raise ValueError("Task Failed")


with DAG(
    dag_id="simple_with_tries",
    schedule="0 0 * * *",
    dagrun_timeout=datetime.timedelta(minutes=10),
    tags=["example2"],
    params={"example_key": "example_value"},
    catchup=False,
) as dag:
    run_this = PythonOperator(
        task_id="python_operator1", python_callable=try_something, retries=5, retry_delay=10
    )

    # run_this_2 = PythonOperator(
    #     task_id="python_operator2", python_callable=try_something, retries=5, retry_delay=10
    # )

Trigger this dag again and observe the console, or refresh the UI:

Committer

• I acknowledge that I am a maintainer/committer of the Apache Airflow project.

[bbovenzi]

It's a 500 error, so this is something in the API. not with how we're rendering it. right?

[pierrejeambrun]

Yes server side. Updating the title to AIP-84