migrating-airflow-2-to-3

To migrate Airflow 2 DAGs to Airflow 3, you need to fix breaking changes across legacy imports, execution_date context usage, direct ORM session queries, and deprecated APIs. A systematic upgrade plan involves running Ruff with Airflow-specific rules and manually refactoring metadata access to use the REST API.

Airflow 3 migration requires addressing removed and renamed DAG arguments, deprecated context keys, XCom pickling changes, and Datasets to Assets migration. You must also replace direct ORM and metadata DB access with the Airflow Python client or REST API due to new runtime restrictions.

Yes, you can automatically fix common Airflow 2 to 3 migration breakages by running Ruff with specific Airflow rule sets including AIR30, AIR301, AIR302, AIR31, AIR311, and AIR312. These rules target legacy operators and outdated code patterns for automated upgrades.

When upgrading to Airflow 3, direct ORM and metadata DB access is restricted and must be refactored. You should replace direct ORM session queries and metadata database interactions by using replacement approaches like the Airflow Python client or the REST API.

Yes, upgrading to Airflow 2.11 first is recommended before migrating to Airflow 3. This intermediate upgrade provides safer rollback constraints and helps ensure a smoother transition when modernizing your codebase for Airflow 3 runtime restrictions.

The best way to scan for deprecated context keys, legacy imports, and old scheduling defaults in Airflow 2 is to use a targeted migration checklist. This helps identify XCom pickling changes and Datasets to Assets migration requirements before upgrading to Airflow 3.