API Authentication

Authentication for the API is handled separately to the Web Authentication. The default is to check the user session:

[api]
auth_backends = airflow.api.auth.backend.session

Changed in version 1.10.11: In Airflow <1.10.11, the default setting was to allow all API requests without authentication, but this posed security risks for if the Webserver is publicly accessible.

Changed in version 2.3.0: In Airflow <2.3.0 this setting was auth_backend and allowed only one value. In 2.3.0 it was changed to support multiple backends that are tried in turn.

If you want to check which authentication backends are currently set, you can use airflow config get-value api auth_backends command as in the example below.

$ airflow config get-value api auth_backends
airflow.api.auth.backend.basic_auth

Disable authentication

If you wish to have the experimental API work, and aware of the risks of enabling this without authentication (or if you have your own authentication layer in front of Airflow) you can set the following in airflow.cfg:

[api]
auth_backends = airflow.api.auth.backend.default

Note

You can only disable authentication for experimental API, not the stable REST API.

See Modules Management for details on how Python and Airflow manage modules.

Kerberos authentication

Kerberos authentication is currently supported for the API.

To enable Kerberos authentication, set the following in the configuration:

[api]
auth_backends = airflow.api.auth.backend.kerberos_auth

[kerberos]
keytab = <KEYTAB>

The Kerberos service is configured as airflow/fully.qualified.domainname@REALM. Make sure this principal exists in the keytab file.

You have to make sure to name your users with the kerberos full username/realm in order to make it works. This means that your user name should be user_name@KERBEROS-REALM.

Basic authentication

Basic username password authentication is currently supported for the API. This works for users created through LDAP login or within Airflow Metadata DB using password.

To enable basic authentication, set the following in the configuration:

[api]
auth_backends = airflow.api.auth.backend.basic_auth

Username and password needs to be base64 encoded and send through the Authorization HTTP header in the following format:

Authorization: Basic Base64(username:password)

Here is a sample curl command you can use to validate the setup:

ENDPOINT_URL="http://localhost:8080/"
curl -X GET  \
    --user "username:password" \
    "${ENDPOINT_URL}/api/v1/pools"

Note, you can still enable this setting to allow API access through username password credential even though Airflow webserver might be using another authentication method. Under this setup, only users created through LDAP or airflow users create command will be able to pass the API authentication.

Roll your own API authentication

Each auth backend is defined as a new Python module. It must have 2 defined methods:

  • init_app(app: Flask) - function invoked when creating a flask application, which allows you to add a new view.

  • requires_authentication(fn: Callable) - a decorator that allows arbitrary code execution before and after or instead of a view function.

and may have one of the following to support API client authorizations used by remote mode for CLI:

  • function create_client_session() -> requests.Session

  • attribute CLIENT_AUTH: tuple[str, str] | requests.auth.AuthBase | None

After writing your backend module, provide the fully qualified module name in the auth_backends key in the [api] section of airflow.cfg.

Additional options to your auth backend can be configured in airflow.cfg, as a new option.

Was this entry helpful?