When does the DAG run, and which dates does it cover?
Enter a cron schedule. See when each run starts, its logical date and its data interval, the way Airflow 2 and Airflow 3 compute them.
| Starts | Logical date | Data interval | In plain words |
|---|
Tested against Airflow 2.11.2 and Airflow 3.4.0 on 14 November 2026. Nothing you enter leaves the browser.
A run belongs to a period, not to a moment.
Airflow was built for batch jobs that process a closed period of data. A daily job for Monday can only run once Monday is over, so the scheduler starts it on Tuesday at 00:00 and labels it with Monday. That label is the logical date, and the period it opens is the data interval. In templates it surfaces as ds, data_interval_start and data_interval_end.
That is CronDataIntervalTimetable, the timetable a cron string gets in Airflow 2. Airflow 3 changed the default to CronTriggerTimetable, which treats the schedule as a clock: the run for Tuesday 00:00 starts on Tuesday 00:00 and its data interval is that single instant. A DAG that reads data_interval_start to decide which partition to load will read a different date after the upgrade, without any change to its file.
Daylight saving adds one more twist. Cron expressions are evaluated on the wall clock of the DAG timezone, so a run scheduled inside the hour that is skipped in spring or repeated in autumn moves, and the data interval around it is an hour shorter or longer. The table above shows the exact instants in UTC next to the local time so the shift is visible.
Airflow schedule FAQ.
Why does my Airflow DAG run a day late?
It usually does not. With CronDataIntervalTimetable, the default for a cron schedule in Airflow 2, a run starts when its data interval ends. A daily DAG whose logical date is Monday covers Monday 00:00 up to Tuesday 00:00 and starts on Tuesday at 00:00. The run is on time, and its logical date names the start of the data it covers.
What is the difference between the logical date and the time a DAG run starts?
The logical date is the start of the run's data interval. With CronDataIntervalTimetable the run starts one schedule period after its logical date, when the interval closes. With CronTriggerTimetable the data interval is a single instant, so the logical date and the start time are the same.
What changed for cron schedules in Airflow 3?
Two defaults. The create_cron_data_intervals setting is False in Airflow 3, so a cron string passed to schedule gets CronTriggerTimetable instead of CronDataIntervalTimetable. The catchup_by_default setting is also False. After an upgrade the same DAG file can produce different logical dates, and it no longer backfills missed runs unless catchup is set.
What does catchup do?
With catchup on, Airflow creates one run for every schedule period between the start date and now as soon as the DAG is unpaused. With catchup off it creates only the most recent due run and continues from there.
Does this tool send my schedule anywhere?
No. The calculation runs in your browser. The only analytics the page sends go to analyticslab.ro: the page path, the Airflow version, the timetable type, the catchup setting, whether a preset was used, whether the page was opened from a shared link, and whether the contact link was clicked. As with any web request, our server also sees your IP address, your browser's user agent and the site you came from, and stores them in the analytics system we operate. None of it carries the cron expression, the timezone or any date.
Which Airflow versions was this checked against?
The results are tested against fixtures generated by the real timetable classes in Airflow 2.11.2 and Airflow 3.4.0, on 14 November 2026. Timedelta schedules, asset triggers, custom timetables and the interval argument of CronTriggerTimetable are not covered.
Built by AnalyticsLab.
We build and repair data platforms: lakehouses on Iceberg, dbt transformation layers and Airflow orchestration. If the schedule is the least of the problems in your pipeline, tell us what else is going on.