gcp-managed-airflow-dag-authoring

Validate Airflow DAGs locally by running ruff for linting and using composer-dev to check import errors with the dags list-import-errors command. After deploying to the GCS bucket, run gcloud composer environments run with dags list-import-errors to confirm no parsing failures.

Run gcloud composer environments describe with the environment name and location, formatting the output on config.softwareConfig.imageVersion. This reveals the exact Airflow and image version so your DAG code matches the target constraints.

Airflow 3 introduces breaking changes in imports, operators, and configuration compared to Airflow 2. Check your target environment image version first, then adjust the DAG code to the matching version and lint with Airflow 3 rulesets when targeting that version.

Performance issues occur when database queries, API calls, or heavy computations run at the top level of the DAG file, since this code executes every few seconds during parsing. Move such logic inside tasks and access variables through Jinja templates.

Without GCP access you can only work offline by relying on user-provided environment details and validating locally with ruff or composer-dev. Deployment to the GCS DAGs bucket and remote import-error verification require authorized GCP access.