Providing Context In TriggerDagRunOperator

I have a dag that has been triggered by another dag. I have passed through to this dag some configuration variables via the DagRunOrder().payload dictionary in the same way the off

Solution 1:

Solved:

The dag_run object is stored in the context and so the configuration variables can be accessed in the python_callable of the TriggerDagRunOperator with this pattern:

def trigger(context, dag_run_obj):
    dag_run_obj.payload = {
        "message": context["dag_run"].conf["message"],
        "day": context["dag_run"].conf["day"]
    }
    return dag_run_obj

trigger_step = TriggerDagRunOperator(
    task_id="trigger_modelling",
    trigger_dag_id="Dummy_Modelling",
    python_callable=trigger,
    dag=dag
)

Solution 2:

In Airflow2.0.x, the equivalent of @efbbrown's answer is:

from airflow.operators.trigger_dagrun import TriggerDagRunOperator

trigger_step = TriggerDagRunOperator(
    task_id="trigger_modelling",
    trigger_dag_id="Dummy_Modelling",
    conf={"message": "{{ dag_run.conf['message'] }}", "day":"{{ 
    dag_run.conf['day'] }}"},
    dag=dag
)

The pull request is described here on GitHub.

See the documentation for external-triggers and for trigger_dagrun.

Here is a YouTube video on the topic that shows the correct imports.

Solution 3:

@efbbrown Yes either you can do that or whenever you access first dags params push it to xcom and pull it while triggering second dag