Apache Airflow Core

3.2.0 · active · verified Thu Apr 09

Apache Airflow Core (part of the main `apache-airflow` distribution) provides the foundational components for the Apache Airflow platform, a powerful tool for programmatically authoring, scheduling, and monitoring data workflows. It includes the scheduler, API server, DAG file processor, and triggerer. Workflows, known as Directed Acyclic Graphs (DAGs), are defined as Python code, making them maintainable, versionable, and collaborative. Airflow follows semantic versioning, with minor releases typically every 2-3 months and patch releases as needed. Major versions, such as 3.x, introduce backwards-incompatible changes and are released on an irregular schedule. The current version of `apache-airflow` is 3.2.0.

Warnings

breaking Direct database access from worker nodes and tasks has been removed in Airflow 3.x. Tasks and custom operators must now communicate through the REST API instead of direct database connections. Existing DAGs and custom operators that accessed the metadata database directly will require refactoring.
Fix: Refactor DAGs and custom operators to use the Airflow REST API or the Task Execution API for runtime interactions (state transitions, heartbeats, XComs, resource fetching). Utilize the Airflow Python Client where appropriate.
breaking SubDAGs have been removed in Airflow 3.x. They are replaced by `TaskGroups`, Assets, and Data Aware Scheduling.
Fix: Migrate existing SubDAGs to use `TaskGroups` for grouping tasks, or explore Assets and Data Aware Scheduling for alternative architectural patterns.
breaking The `/api/v1` REST API has been replaced by the modern FastAPI-based `/api/v2`.
Fix: Update any integrations or custom tools that interact with the Airflow REST API to use the `/api/v2` endpoints. Refer to the Airflow API v2 documentation.
breaking The `catchup_by_default` DAG parameter is now `False` by default in Airflow 3.x, changing the default behavior for DAGs scheduled to run for past intervals.
Fix: Explicitly set `catchup=True` in your DAG definition if you intend for it to run for past missed schedules. Review existing DAGs to ensure the desired catchup behavior is maintained.
deprecated Airflow 2.x reaches its end-of-life in April 2026. Staying on Airflow 2.x after this date will introduce security risks as CVEs cannot be patched anymore due to old dependencies.
Fix: Plan and execute an upgrade to Airflow 3.x. Utilize tools like Ruff with AIR rules (version 0.13.1 or later) to check DAG compatibility and automate parts of the migration.
gotcha Plain `pip install apache-airflow` might occasionally lead to an unusable installation due to dependency conflicts. Airflow keeps its dependencies open (in `pyproject.toml`) for user flexibility, but this requires careful management.
Fix: Always use the official constraint files for reproducible Airflow installations. Example: `pip install "apache-airflow[EXTRAS]==AIRFLOW_VERSION" --constraint "https://raw.githubusercontent.com/apache/airflow/constraints-AIRFLOW_VERSION/constraints-PYTHON_VERSION.txt"`.

Install

pip install "apache-airflow[standard]==3.2.0" --constraint "https://raw.githubusercontent.com/apache/airflow/constraints-3.2.0/constraints-3.10.txt" Recommended reproducible installation for Airflow 3.2.0 with Python 3.10
pip install "apache-airflow[standard]==AIRFLOW_VERSION" --constraint "https://raw.githubusercontent.com/apache/airflow/constraints-AIRFLOW_VERSION/constraints-PYTHON_VERSION.txt" General reproducible installation (replace placeholders)

Imports

DAG
```
from airflow import DAG
```
datetime
```
from datetime import datetime
```

BashOperator

from airflow.operators.bash import BashOperator

PythonOperator

from airflow.operators.python import PythonOperator

Quickstart

This quickstart defines a simple 'Hello World' DAG. To run it locally, first install Apache Airflow (see 'install' section). Then, initialize the database and start Airflow components using `airflow standalone`. Place this Python file in your configured `AIRFLOW_HOME/dags` directory, then access the Airflow UI at `localhost:8080`, log in (admin/admin by default for `standalone`), and unpause the `hello_world_dag`.

from airflow import DAG
from airflow.operators.bash import BashOperator
from datetime import datetime, timedelta

with DAG(
    dag_id='hello_world_dag',
    start_date=datetime(2026, 1, 1),
    schedule_interval=timedelta(days=1),
    catchup=False, # Default in Airflow 3.x is False
    tags=['example'],
) as dag:
    start_task = BashOperator(
        task_id='start',
        bash_command='echo "Starting hello_world_dag"',
    )

    hello_task = BashOperator(
        task_id='hello',
        bash_command='echo "Hello, Airflow!"',
    )

    world_task = BashOperator(
        task_id='world',
        bash_command='echo "This is a simple DAG."',
    )

    end_task = BashOperator(
        task_id='end',
        bash_command='echo "Finished hello_world_dag"',
    )

    start_task >> hello_task >> world_task >> end_task

view raw JSON →