Apache Airflow

Sifflet connects to Apache Airflow to show the status of your DAGs next to the data they produce. You see failed or late DAG runs directly in the Data Catalog, on asset pages, and in the lineage, and when a DAG fails, you see which tables and dashboards downstream are affected. For the list of what Sifflet imports, see Airflow Collected Data.

Airflow DAG status in the Sifflet catalog.

Choose Your Deployment

Sifflet supports three ways of running Airflow. Each one uses a different identity and offers different network options, so each has its own setup page.

Deployment	Identity Sifflet uses	Network options	Setup page
Self-hosted Airflow	Airflow user with the `Viewer` role (basic authentication), or an OAuth2 client	Public endpoint, IP allowlisting	Self-Hosted Airflow
Amazon Managed Workflows for Apache Airflow (MWAA)	AWS IAM role assumed by Sifflet	Public web server; private web server through the MWAA `InvokeRestApi` API (private preview)	Amazon MWAA
Google Cloud Composer	GCP service account with a JSON key	Public web server, IP allowlisting	Cloud Composer

All three deployments import the same data: see Airflow Collected Data.

Related Guides

These guides aren't needed to connect Airflow, but build on the integration:

Linking Airflow DAGs to Data Assets: tag the queries of your DAGs so that Sifflet links each DAG to the tables it updates.
Airflow Custom Operators: trigger Sifflet monitors or send dbt artifacts from your DAGs.