Airflow does not support DAGs with loops. After all, the abbreviation DAG stands for Directed Acyclic Graph, so we can’t have cycles. It is also not the standard usage of Airflow, which was built to support daily batch processing.
Table of Contents
All of that does not stop us from using a simple trick that lets us run a DAG in a loop. To do that, we have to add a TriggerDagRunOperator as the last task in the DAG. In the task configuration, we specify the DAG id of the DAG that contains the task:
from airflow.operators.dagrun_operator import TriggerDagRunOperator
trigger_self = TriggerDagRunOperator(
task_id='repeat'
trigger_dag_id=dag.dag_id,
dag=dag
)
the_rest_of_the_dag >> trigger_self # add it as the last task
Get Weekly AI Implementation Insights
Join engineering leaders who receive my analysis of common AI production failures and how to prevent them. No fluff, just actionable techniques.