Airflow Integration

note

If you're looking to schedule DataHub ingestion using Airflow, see the guide on scheduling ingestion with Airflow.

The DataHub Airflow plugin supports:

• Automatic column-level lineage extraction from various operators e.g. SQL operators (including MySqlOperator, PostgresOperator, SnowflakeOperator, BigQueryInsertJobOperator, and more), S3FileTransformOperator, and more.
• Airflow DAG and tasks, including properties, ownership, and tags.
• Task run information, including task successes and failures.
• Manual lineage annotations using inlets and outlets on Airflow operators.

There's two implementations of the plugin, with different Airflow version support.

Approach	Airflow Versions	Notes
Plugin v2	2.3.4+	Recommended. Requires Python 3.8+
Plugin v1	2.3 - 2.8	Deprecated. No automatic lineage extraction; may not extract lineage if the task fails.

If you're using Airflow older than 2.3, it's possible to use the v1 plugin with older versions of acryl-datahub-airflow-plugin. See the compatibility section for more details.

DataHub Plugin v2

Installation

The v2 plugin requires Airflow 2.3+ and Python 3.8+. If you don't meet these requirements, see the compatibility section for other options.

pip install 'acryl-datahub-airflow-plugin[plugin-v2]'

Configuration

Set up a DataHub connection in Airflow, either via command line or the Airflow UI.

Command Line

airflow connections add  --conn-type 'datahub-rest' 'datahub_rest_default' --conn-host 'http://datahub-gms:8080' --conn-password '<optional datahub auth token>'

If you are using DataHub Cloud then please use https://YOUR_PREFIX.acryl.io/gms as the --conn-host parameter.

Airflow UI

On the Airflow UI, go to Admin -> Connections and click the "+" symbol to create a new connection. Select "DataHub REST Server" from the dropdown for "Connection Type" and enter the appropriate values.

Optional Configurations

No additional configuration is required to use the plugin. However, there are some optional configuration parameters that can be set in the airflow.cfg file.

airflow.cfg

[datahub]
# Optional - additional config here.
enabled = True  # default

Name	Default value	Description
enabled	true	If the plugin should be enabled.
conn_id	datahub_rest_default	The name of the datahub rest connection.
cluster	prod	name of the airflow cluster, this is equivalent to the env of the instance
capture_ownership_info	true	Extract DAG ownership.
capture_ownership_as_group	false	When extracting DAG ownership, treat DAG owner as a group rather than a user
capture_tags_info	true	Extract DAG tags.
capture_executions	true	Extract task runs and success/failure statuses. This will show up in DataHub "Runs" tab.
materialize_iolets	true	Create or un-soft-delete all entities referenced in lineage.
enable_extractors	true	Enable automatic lineage extraction.
disable_openlineage_plugin	true	Disable the OpenLineage plugin to avoid duplicative processing.
log_level	no change	[debug] Set the log level for the plugin.
debug_emitter	false	[debug] If true, the plugin will log the emitted events.

DataHub Plugin v1

Installation

The v1 plugin requires Airflow 2.3 - 2.8 and Python 3.8+. If you're on older versions, it's still possible to use an older version of the plugin. See the compatibility section for more details.

Note that the v1 plugin is less featureful than the v2 plugin, and is overall not actively maintained. Since datahub v0.15.0, the v2 plugin has been the default. If you need to use the v1 plugin with acryl-datahub-airflow-plugin v0.15.0+, you must also set the environment variable DATAHUB_AIRFLOW_PLUGIN_USE_V1_PLUGIN=true.

pip install 'acryl-datahub-airflow-plugin[plugin-v1]'

# The DataHub rest connection type is included by default.
# To use the DataHub Kafka connection type, install the plugin with the kafka extras.
pip install 'acryl-datahub-airflow-plugin[plugin-v1,datahub-kafka]'

Configuration

Disable lazy plugin loading

airflow.cfg

[core]
lazy_load_plugins = False

On MWAA you should add this config to your Apache Airflow configuration options.

Setup a DataHub connection

You must configure an Airflow connection for Datahub. We support both a Datahub REST and a Kafka-based connections, but you only need one.

# For REST-based:
airflow connections add  --conn-type 'datahub_rest' 'datahub_rest_default' --conn-host 'http://datahub-gms:8080' --conn-password '<optional datahub auth token>'
# For Kafka-based (standard Kafka sink config can be passed via extras):
airflow connections add  --conn-type 'datahub_kafka' 'datahub_kafka_default' --conn-host 'broker:9092' --conn-extra '{}'

Configure the plugin

If your config doesn't align with the default values, you can configure the plugin in your airflow.cfg file.

airflow.cfg

[datahub]
enabled = true
conn_id = datahub_rest_default  # or datahub_kafka_default
# etc.

Name	Default value	Description
enabled	true	If the plugin should be enabled.
conn_id	datahub_rest_default	The name of the datahub connection you set in step 1.
cluster	prod	name of the airflow cluster
capture_ownership_info	true	If true, the owners field of the DAG will be capture as a DataHub corpuser.
capture_ownership_as_group	false	When extracting DAG ownership, treat DAG owner as a group rather than a user.
capture_tags_info	true	If true, the tags field of the DAG will be captured as DataHub tags.
capture_executions	true	If true, we'll capture task runs in DataHub in addition to DAG definitions.
materialize_iolets	true	Create or un-soft-delete all entities referenced in lineage.
render_templates	true	If true, jinja-templated fields will be automatically rendered to improve the accuracy of SQL statement extraction.
datajob_url_link	taskinstance	If taskinstance, the datajob url will be taskinstance link on airflow. It can also be grid.
graceful_exceptions	true	If set to true, most runtime errors in the lineage backend will be suppressed and will not cause the overall task to fail. Note that configuration issues will still throw exceptions.
dag_filter_str	{ "allow": [".*"] }	AllowDenyPattern value in form of JSON string to filter the DAGs from running.

Validate that the plugin is working

1. Go and check in Airflow at Admin -> Plugins menu if you can see the DataHub plugin
2. Run an Airflow DAG. In the task logs, you should see Datahub related log messages like:

Emitting DataHub ...

Automatic lineage extraction

Only the v2 plugin supports automatic lineage extraction. If you're using the v1 plugin, you must use manual lineage annotation or emit lineage directly.

To automatically extract lineage information, the v2 plugin builds on top of Airflow's built-in OpenLineage extractors. As such, we support a superset of the default operators that Airflow/OpenLineage supports.

The SQL-related extractors have been updated to use DataHub's SQL lineage parser, which is more robust than the built-in one and uses DataHub's metadata information to generate column-level lineage.

Supported operators:

• SQLExecuteQueryOperator, including any subclasses. Note that in newer versions of Airflow (generally Airflow 2.5+), most SQL operators inherit from this class.
• AthenaOperator and AWSAthenaOperator
• BigQueryOperator and BigQueryExecuteQueryOperator
• BigQueryInsertJobOperator (incubating)
• MySqlOperator
• PostgresOperator
• RedshiftSQLOperator
• SnowflakeOperator and SnowflakeOperatorAsync
• SqliteOperator
• TrinoOperator

Known limitations:

• We do not fully support operators that run multiple SQL statements at once. In these cases, we'll only capture lineage from the first SQL statement.

Manual Lineage Annotation

Using inlets and outlets

You can manually annotate lineage by setting inlets and outlets on your Airflow operators. This is useful if you're using an operator that doesn't support automatic lineage extraction, or if you want to override the automatic lineage extraction.

We have a few code samples that demonstrate how to use inlets and outlets:

• lineage_backend_demo.py
• lineage_backend_taskflow_demo.py - uses the TaskFlow API

For more information, take a look at the Airflow lineage docs.

Custom Operators

If you have created a custom Airflow operator that inherits from the BaseOperator class, when overriding the execute function, set inlets and outlets via context['ti'].task.inlets and context['ti'].task.outlets. The DataHub Airflow plugin will then pick up those inlets and outlets after the task runs.

class DbtOperator(BaseOperator):
    ...

    def execute(self, context):
        # do something
        inlets, outlets = self._get_lineage()
        # inlets/outlets are lists of either datahub_airflow_plugin.entities.Dataset or datahub_airflow_plugin.entities.Urn
        context['ti'].task.inlets = self.inlets
        context['ti'].task.outlets = self.outlets

    def _get_lineage(self):
        # Do some processing to get inlets/outlets

        return inlets, outlets

If you override the pre_execute and post_execute function, ensure they include the @prepare_lineage and @apply_lineage decorators respectively. Reference the Airflow docs for more details.

Custom Extractors

Note: these are only supported in the v2 plugin.

You can also create a custom extractor to extract lineage from any operator. This is useful if you're using a built-in Airflow operator for which we don't support automatic lineage extraction.

See this example PR which adds a custom extractor for the BigQueryInsertJobOperator operator.

Cleanup obsolete pipelines and tasks from Datahub

There might be a case where the DAGs are removed from the Airflow but the corresponding pipelines and tasks are still there in the Datahub, let's call such pipelines ans tasks, obsolete pipelines and tasks

Following are the steps to cleanup them from the datahub:

• create a DAG named Datahub_Cleanup, i.e.

from datetime import datetime

from airflow import DAG
from airflow.operators.bash import BashOperator

from datahub_airflow_plugin.entities import Dataset, Urn

with DAG(
    "Datahub_Cleanup",
    start_date=datetime(2024, 1, 1),
    schedule_interval=None,
    catchup=False,
) as dag:
    task = BashOperator(
        task_id="cleanup_obsolete_data",
        dag=dag,
        bash_command="echo 'cleaning up the obsolete data from datahub'",
    )

• ingest this DAG, and it will remove all the obsolete pipelines and tasks from the Datahub based on the cluster value set in the airflow.cfg

Get all dataJobs associated with a dataFlow

If you are looking to find all tasks (aka DataJobs) that belong to a specific pipeline (aka DataFlow), you can use the following GraphQL query:

query {
  dataFlow(urn: "urn:li:dataFlow:(airflow,db_etl,prod)") {
    childJobs: relationships(
      input: { types: ["IsPartOf"], direction: INCOMING, start: 0, count: 100 }
    ) {
      total
      relationships {
        entity {
          ... on DataJob {
            urn
          }
        }
      }
    }
  }
}

Emit Lineage Directly

If you can't use the plugin or annotate inlets/outlets, you can also emit lineage using the DatahubEmitterOperator.

Reference lineage_emission_dag.py for a full example.

In order to use this example, you must first configure the Datahub hook. Like in ingestion, we support a Datahub REST hook and a Kafka-based hook. See the plugin configuration for examples.

Debugging

Missing lineage

If you're not seeing lineage in DataHub, check the following:

• Validate that the plugin is loaded in Airflow. Go to Admin -> Plugins and check that the DataHub plugin is listed.
• With the v2 plugin, it should also print a log line like INFO [datahub_airflow_plugin.datahub_listener] DataHub plugin v2 using DataHubRestEmitter: configured to talk to <datahub_url> during Airflow startup, and the airflow plugins command should list datahub_plugin with a listener enabled.
• If using the v2 plugin's automatic lineage, ensure that the enable_extractors config is set to true and that automatic lineage is supported for your operator.
• If using manual lineage annotation, ensure that you're using the datahub_airflow_plugin.entities.Dataset or datahub_airflow_plugin.entities.Urn classes for your inlets and outlets.

Incorrect URLs

If your URLs aren't being generated correctly (usually they'll start with http://localhost:8080 instead of the correct hostname), you may need to set the webserver base_url config.

airflow.cfg

[webserver]
base_url = http://airflow.mycorp.example.com

TypeError ... missing 3 required positional arguments

If you see errors like the following with the v2 plugin:

ERROR - on_task_instance_success() missing 3 required positional arguments: 'previous_state', 'task_instance', and 'session'
Traceback (most recent call last):
  File "/home/airflow/.local/lib/python3.8/site-packages/datahub_airflow_plugin/datahub_listener.py", line 124, in wrapper
    f(*args, **kwargs)
TypeError: on_task_instance_success() missing 3 required positional arguments: 'previous_state', 'task_instance', and 'session'

The solution is to upgrade acryl-datahub-airflow-plugin>=0.12.0.4 or upgrade pluggy>=1.2.0. See this PR for details.

Compatibility

We no longer officially support Airflow <2.3. However, you can use older versions of acryl-datahub-airflow-plugin with older versions of Airflow. The first two options support Python 3.7+, and the last option supports Python 3.8+.

• Airflow 1.10.x, use DataHub plugin v1 with acryl-datahub-airflow-plugin <= 0.9.1.0.
• Airflow 2.0.x, use DataHub plugin v1 with acryl-datahub-airflow-plugin <= 0.11.0.1.
• Airflow 2.2.x, use DataHub plugin v2 with acryl-datahub-airflow-plugin <= 0.14.1.5.

DataHub also previously supported an Airflow lineage backend implementation. While the implementation is still in our codebase, it is deprecated and will be removed in a future release. Note that the lineage backend did not support automatic lineage extraction, did not capture task failures, and did not work in AWS MWAA. The documentation for the lineage backend has already been archived.

Additional references

Related Datahub videos:

• Airflow Lineage
• Airflow Run History in DataHub