Upgrading to Airflow 3

As of version 3.0.0, IQVIA NLP Data Factory is intended for Airflow 3. Airflow 3 is a major release with breaking changes for deployments, configuration, authentication, plugins, custom operators, and DAG code. Use this page together with the Airflow upgrade guide when planning an IQVIA NLP Data Factory upgrade.

Upgrade process

Start with the official Airflow 3 upgrade guide. It covers the recommended upgrade path from Airflow 2.x, including prerequisites, metadata database cleanup and backup, provider installation, configuration migration, database migration, startup command changes, and breaking changes.

Before upgrading, confirm that your target Airflow, Python, and IQVIA NLP Airflow provider versions are compatible. The Installation section lists the currently supported versions and includes installation examples.

The following Airflow documentation is also useful when planning the upgrade:

• Airflow release notes

• Public Interface for Airflow 3.0+

• Installing Airflow from PyPI

• Airflow Docker image documentation

• Airflow FAB auth manager documentation

If your deployment uses Flask App Builder authentication in Airflow 2, review the Authentication section during the upgrade. Airflow 3 uses a pluggable auth manager, and FAB authentication requires the FAB provider and related configuration changes.

Updating DAGs

Complete Airflow’s DAG compatibility step before migrating the Airflow metadata database. That gives you a chance to fix and deploy DAG parsing issues while the existing Airflow 2 environment is still available.

Use the Ruff AIR checks and import migration guidance in that section to identify DAG changes that are required before running those DAGs on Airflow 3.

After updating DAG code, test the DAGs in a staging Airflow 3 environment with the same provider wheels, constraints, connections, variables, and external system access that production will use.