Backfilling Historical dbt Runs with Airflow
A backfill reruns a DAG for a historical date range to repair or populate missing partitions.
Before starting a backfill, confirm the date filter, warehouse capacity, dependencies, and whether the dbt models are idempotent. Run a small date range first, monitor the tasks, and then expand the range.
Backfills should be isolated from the regular schedule when possible so historical work does not delay current production data.