annotating-task-lineage

To add data lineage without automatic extraction, manually annotate Airflow tasks by defining OpenLineage Dataset objects, Airflow 3+ Assets, or Airflow 2.4+ Datasets and assigning them to the operator inlets and outlets properties.

You need to manually set inlets and outlets when an Airflow DAG operator reads from or writes to external systems and lacks a built-in OpenLineage extractor, serving as the lowest-priority fallback to capture lineage visibility.

You can use OpenLineage Dataset objects, Airflow 3+ Assets, or Airflow 2.4+ Datasets as inlet and outlet types to document task data lineage and make dependencies visible in Astro Lineage.

Yes, you can track multiple inputs and outputs by assigning multiple dataset objects to the inlets and outlets of a single Airflow operator, covering multi-input and multi-output transformations or data exports.

Yes, annotating task lineage applies to DAG operators that read from and write to external systems including databases and object storage, resolving missing lineage visibility by mapping those sources as inlets and outlets.

Data lineage is missing in Astro Lineage when operators lack automatic OpenLineage extraction, requiring you to manually define and assign OpenLineage Datasets or Airflow Assets to the task inlets and outlets.