A Dag run is one execution of a Dag for one logical date. The scheduler creates scheduled runs from the schedule; you create manual runs with Trigger or airflow dags trigger. A run moves from queued to running to success or failed.
Two switches decide whether anything runs. A paused Dag gets no scheduled runs, and even a run you trigger by hand waits in Queued until you switch the Dag on. Switching a Dag back on does not replay every midnight it missed: with catchup=False, the Airflow 3 default, the scheduler starts only the latest one.
The logical date says which slice of time a run is about. A manual run gets the moment you triggered it unless you pick a date, which is how you rerun last Friday: airflow dags trigger marigold_daily --logical-date 2026-11-06. Tasks that read ds then load Friday's data.
On Dags, the scheduler clock in this workspace lets you move time forward and watch the scheduler react.
Switch the Dag on, rerun Friday, then list its runs with their types and states.