Watch the recording
By proceeding you agree to our Privacy Policy, our Website Terms and to receive emails from Astronomer.
Upgrade from Airflow 2 to Airflow 3: On-Demand Webinar
Recorded on
Tamara FingerlinSenior Developer AdvocateAstronomer
Kenten DanasDirector, Developer RelationsAstronomer.png)
Alida LaneyData EngineerCity of Pittsburgh
Airflow 3 launched in April 2025, and leading data teams have upgraded to unlock new features like Dag versioning, remote execution, and more. Now that Airflow 2 has reached end of life and no longer gets updated with security patches and bug fixes, it has never been more urgent to upgrade.
In this webinar, we’ll cover everything you need to plan a smooth upgrade from Airflow 2 to 3, without breaking your workflows or stretching your team thin.
You’ll learn:
- What Airflow 2 end-of-life means for your deployments and the risks of staying on unsupported versions
- Breaking changes between Airflow 2 and 3
- Upgrade utilities like ruff AIR rules and airflow config lint to help you identify changes
- Testing and rollout recommendations for upgrading
Plus: hear directly from a Data Engineer who upgraded to Airflow 3 in production, with real lessons learned and tips you can use right away.
Frequently asked questions
When did Airflow 2 reach end of life?
Airflow 2 reached end of life on April 22, 2026.
The Apache Airflow project no longer ships updates such as security patches, bug fixes, or compatibility guarantees as Python versions, libraries, and provider packages move on. Astro customers will have extended Airflow 2 support until April 2027.
Which Airflow 2 version should I upgrade from?
Airflow 2.11 is the recommended starting version.
Once there, fix whatever its deprecation warnings surface, then upgrade to the latest patch of a supported Airflow 3 release. On Astro, in-place upgrades require Runtime 13.7.0 or above. See the Airflow 2 to Airflow 3 upgrade docs.
Can I roll back from Airflow 3 to Airflow 2?
Yes, on Astro.
You can only roll back to Airflow 2.11, and it requires that your Deployment was on Astro Runtime 13.7.0 or newer before the upgrade and is on 3.0-11 or newer now. See Airflow 3 to Airflow 2 rollback support.
What are the biggest breaking changes in Airflow 3?
Airflow 3 includes a new UI and architectural changes.
Workers no longer access the metadata database directly, so Dag code that queries it must use the Airflow Python client or REST API instead. Some imports and parameters were removed. See the full list of Airflow 3 breaking changes.
How do I find the changes my Dags need before upgrading?
Run the Airflow ruff rules against your Dag code.
AIR30 rules flag required changes such as removed parameters and imports, and AIR31 rules flag deprecated code you should fix. With the Astro CLI, astro dev upgrade-test includes ruff check options. See the Airflow 3 ruff rules and upgrade checklist.
