Trigger Rule

An Airflow trigger rule specifies which upstream task states satisfy a task’s dependency condition. The default all_success requires upstream success. all_done allows execution once upstream tasks are terminal, including failed or skipped states.

Passing a trigger rule is not the same as starting immediately: scheduling and capacity constraints still apply. A cleanup task may need all_done, but that does not validate the data produced upstream. Branching can skip tasks, so choose downstream rules for the intended branch behavior.

DAG Run status depends on leaf task states. A successful all_done leaf can cause the run to be marked successful despite an earlier failure. Test the terminal-state path and preserve a meaningful validation outcome when adding cleanup or branches.

See Apache Airflow Fundamentals: DAGs, Tasks, and Scheduling for worked examples.

Further reading

• Apache Airflow 3.1.0 — Dag Runs