Apache Airflow - OpenApi Client for Python

Clone this repo:
  1. 8bf55db Synchronize the build scripts/config/docs with apache-airflow repo (#105) by Jarek Potiuk · 3 months ago main
  2. 11726ed Modernize Python Client packaging (#104) by Jarek Potiuk · 3 months ago
  3. bc7f81e Housekeeping of Python client repository (#103) by Jarek Potiuk · 3 months ago
  4. 4775ec8 Add Client Version 2.8.0 (#102) by Ephraim Anierobi · 4 months ago 2.8.0 2.8.0rc1
  5. 9dea92f Add instructions on how to verify sources for PMC members (#101) by Jarek Potiuk · 5 months ago

Apache Airflow Python Client


To facilitate management, Apache Airflow supports a range of REST API endpoints across its objects. This section provides an overview of the API design, methods, and supported use cases.

Most of the endpoints accept JSON as input and return JSON responses. This means that you must usually add the following headers to your request:

Content-type: application/json
Accept: application/json


The term resource refers to a single type of object in the Airflow metadata. An API is broken up by its endpoint's corresponding resource. The name of a resource is typically plural and expressed in camelCase. Example: dagRuns.

Resource names are used as part of endpoint URLs, as well as in API parameters and responses.

CRUD Operations

The platform supports Create, Read, Update, and Delete operations on most resources. You can review the standards for these operations and their standard parameters below.

Some endpoints have special behavior as exceptions.


To create a resource, you typically submit an HTTP POST request with the resource‘s required metadata in the request body. The response returns a 201 Created response code upon success with the resource’s metadata, including its internal id, in the response body.


The HTTP GET request can be used to read a resource or to list a number of resources.

A resource‘s id can be submitted in the request parameters to read a specific resource. The response usually returns a 200 OK response code upon success, with the resource’s metadata in the response body.

If a GET request does not include a specific resource id, it is treated as a list request. The response usually returns a 200 OK response code upon success, with an object containing a list of resources' metadata in the response body.

When reading resources, some common query parameters are usually available. e.g.:

Query ParameterTypeDescription
limitintegerMaximum number of objects to fetch. Usually 25 by default
offsetintegerOffset after which to start returning objects. For use with limit query parameter.


Updating a resource requires the resource id, and is typically done using an HTTP PATCH request, with the fields to modify in the request body. The response usually returns a 200 OK response code upon success, with information about the modified resource in the response body.


Deleting a resource requires the resource id and is typically executing via an HTTP DELETE request. The response usually returns a 204 No Content response code upon success.


  • Resource names are plural and expressed in camelCase.

  • Names are consistent between URL parameter name and field name.

  • Field names are in snake_case.

    \"name\": \"string\",
    \"slots\": 0,
    \"occupied_slots\": 0,
    \"used_slots\": 0,
    \"queued_slots\": 0,
    \"open_slots\": 0

Update Mask

Update mask is available as a query parameter in patch endpoints. It is used to notify the API which fields you want to update. Using update_mask makes it easier to update objects by helping the server know which fields to update in an object instead of updating all fields. The update request ignores any fields that aren't specified in the field mask, leaving them with their current values.


import requests

resource = requests.get("/resource/my-id").json()
resource["my_field"] = "new-value"
requests.patch("/resource/my-id?update_mask=my_field", data=json.dumps(resource))

Versioning and Endpoint Lifecycle

  • API versioning is not synchronized to specific releases of the Apache Airflow.
  • APIs are designed to be backward compatible.
  • Any changes to the API will first go through a deprecation phase.

Trying the API

You can use a third party client, such as curl, HTTPie, Postman or the Insomnia rest client to test the Apache Airflow API.

Note that you will need to pass credentials data.

For e.g., here is how to pause a DAG with curl, when basic authorization is used:

curl -X PATCH 'https://example.com/api/v1/dags/{dag_id}?update_mask=is_paused' \\
-H 'Content-Type: application/json' \\
--user \"username:password\" \\
-d '{
    \"is_paused\": true

Using a graphical tool such as Postman or Insomnia, it is possible to import the API specifications directly:

  1. Download the API specification by clicking the Download button at top of this document.
  2. Import the JSON specification in the graphical tool of your choice.
  • In Postman, you can click the import button at the top
  • With Insomnia, you can just drag-and-drop the file on the UI

Note that with Postman, you can also generate code snippets by selecting a request and clicking on the Code button.

Enabling CORS

Cross-origin resource sharing (CORS) is a browser security feature that restricts HTTP requests that are initiated from scripts running in the browser.

For details on enabling/configuring CORS, see Enabling CORS.


To be able to meet the requirements of many organizations, Airflow supports many authentication methods, and it is even possible to add your own method.

If you want to check which auth backend is 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

The default is to deny all requests.

For details on configuring the authentication, see API Authorization.


We follow the error response format proposed in RFC 7807 also known as Problem Details for HTTP APIs. As with our normal API responses, your client must be prepared to gracefully handle additional members of the response.


This indicates that the request has not been applied because it lacks valid authentication credentials for the target resource. Please check that you have valid credentials.


This response means that the server understood the request but refuses to authorize it because it lacks sufficient rights to the resource. It happens when you do not have the necessary permission to execute the action you performed. You need to get the appropriate permissions in other to resolve this error.


This response means that the server cannot or will not process the request due to something that is perceived to be a client error (e.g., malformed request syntax, invalid request message framing, or deceptive request routing). To resolve this, please ensure that your syntax is correct.


This client error response indicates that the server cannot find the requested resource.


Indicates that the request method is known by the server but is not supported by the target resource.


The target resource does not have a current representation that would be acceptable to the user agent, according to the proactive negotiation header fields received in the request, and the server is unwilling to supply a default representation.


The request could not be completed due to a conflict with the current state of the target resource, e.g. the resource it tries to create already exists.


This means that the server encountered an unexpected condition that prevented it from fulfilling the request.

This Python package is automatically generated by the OpenAPI Generator project:

  • API version: 2.8.0
  • Package version: 2.8.0
  • Build package: org.openapitools.codegen.languages.PythonClientCodegen For more information, please visit https://airflow.apache.org


Python >=3.8

Installation & Usage

pip install

You can install the client using standard Python installation tools. It is hosted in PyPI with apache-airflow-client package id so the easiest way to get the latest version is to run:

pip install apache-airflow-client

If the python package is hosted on a repository, you can install directly using:

pip install git+https://github.com/apache/airflow-client-python.git

Import check

Then import the package:

import airflow_client.client

Getting Started

Please follow the installation procedure and then run the following:

import time
import airflow_client.client
from pprint import pprint
from airflow_client.client.api import config_api
from airflow_client.client.model.config import Config
from airflow_client.client.model.error import Error

# Defining the host is optional and defaults to /api/v1
# See configuration.py for a list of all supported configuration parameters.
configuration = client.Configuration(host="/api/v1")

# The client must configure the authentication and authorization parameters
# in accordance with the API server security policy.
# Examples for each auth method are provided below, use the example that
# satisfies your auth use case.

# Configure HTTP basic authorization: Basic
configuration = client.Configuration(username="YOUR_USERNAME", password="YOUR_PASSWORD")

# Enter a context with an instance of the API client
with client.ApiClient(configuration) as api_client:
    # Create an instance of the API class
    api_instance = config_api.ConfigApi(api_client)

        # Get current configuration
        api_response = api_instance.get_config()
    except client.ApiException as e:
        print("Exception when calling ConfigApi->get_config: %s\n" % e)

Documentation for API Endpoints

All URIs are relative to /api/v1

ClassMethodHTTP requestDescription
ConfigApiget_configGET /configGet current configuration
ConnectionApidelete_connectionDELETE /connections/{connection_id}Delete a connection
ConnectionApiget_connectionGET /connections/{connection_id}Get a connection
ConnectionApiget_connectionsGET /connectionsList connections
ConnectionApipatch_connectionPATCH /connections/{connection_id}Update a connection
ConnectionApipost_connectionPOST /connectionsCreate a connection
ConnectionApitest_connectionPOST /connections/testTest a connection
DAGApidelete_dagDELETE /dags/{dag_id}Delete a DAG
DAGApiget_dagGET /dags/{dag_id}Get basic information about a DAG
DAGApiget_dag_detailsGET /dags/{dag_id}/detailsGet a simplified representation of DAG
DAGApiget_dag_sourceGET /dagSources/{file_token}Get a source code
DAGApiget_dagsGET /dagsList DAGs
DAGApiget_taskGET /dags/{dag_id}/tasks/{task_id}Get simplified representation of a task
DAGApiget_tasksGET /dags/{dag_id}/tasksGet tasks for DAG
DAGApipatch_dagPATCH /dags/{dag_id}Update a DAG
DAGApipatch_dagsPATCH /dagsUpdate DAGs
DAGApipost_clear_task_instancesPOST /dags/{dag_id}/clearTaskInstancesClear a set of task instances
DAGApipost_set_task_instances_statePOST /dags/{dag_id}/updateTaskInstancesStateSet a state of task instances
DAGRunApiclear_dag_runPOST /dags/{dag_id}/dagRuns/{dag_run_id}/clearClear a DAG run
DAGRunApidelete_dag_runDELETE /dags/{dag_id}/dagRuns/{dag_run_id}Delete a DAG run
DAGRunApiget_dag_runGET /dags/{dag_id}/dagRuns/{dag_run_id}Get a DAG run
DAGRunApiget_dag_runsGET /dags/{dag_id}/dagRunsList DAG runs
DAGRunApiget_dag_runs_batchPOST /dags/~/dagRuns/listList DAG runs (batch)
DAGRunApiget_upstream_dataset_eventsGET /dags/{dag_id}/dagRuns/{dag_run_id}/upstreamDatasetEventsGet dataset events for a DAG run
DAGRunApipost_dag_runPOST /dags/{dag_id}/dagRunsTrigger a new DAG run
DAGRunApiset_dag_run_notePATCH /dags/{dag_id}/dagRuns/{dag_run_id}/setNoteUpdate the DagRun note.
DAGRunApiupdate_dag_run_statePATCH /dags/{dag_id}/dagRuns/{dag_run_id}Modify a DAG run
DagWarningApiget_dag_warningsGET /dagWarningsList dag warnings
DatasetApiget_datasetGET /datasets/{uri}Get a dataset
DatasetApiget_dataset_eventsGET /datasets/eventsGet dataset events
DatasetApiget_datasetsGET /datasetsList datasets
DatasetApiget_upstream_dataset_eventsGET /dags/{dag_id}/dagRuns/{dag_run_id}/upstreamDatasetEventsGet dataset events for a DAG run
EventLogApiget_event_logGET /eventLogs/{event_log_id}Get a log entry
EventLogApiget_event_logsGET /eventLogsList log entries
ImportErrorApiget_import_errorGET /importErrors/{import_error_id}Get an import error
ImportErrorApiget_import_errorsGET /importErrorsList import errors
MonitoringApiget_healthGET /healthGet instance status
MonitoringApiget_versionGET /versionGet version information
PermissionApiget_permissionsGET /permissionsList permissions
PluginApiget_pluginsGET /pluginsGet a list of loaded plugins
PoolApidelete_poolDELETE /pools/{pool_name}Delete a pool
PoolApiget_poolGET /pools/{pool_name}Get a pool
PoolApiget_poolsGET /poolsList pools
PoolApipatch_poolPATCH /pools/{pool_name}Update a pool
PoolApipost_poolPOST /poolsCreate a pool
ProviderApiget_providersGET /providersList providers
RoleApidelete_roleDELETE /roles/{role_name}Delete a role
RoleApiget_roleGET /roles/{role_name}Get a role
RoleApiget_rolesGET /rolesList roles
RoleApipatch_rolePATCH /roles/{role_name}Update a role
RoleApipost_rolePOST /rolesCreate a role
TaskInstanceApiget_extra_linksGET /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances/{task_id}/linksList extra links
TaskInstanceApiget_logGET /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances/{task_id}/logs/{task_try_number}Get logs
TaskInstanceApiget_mapped_task_instanceGET /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances/{task_id}/{map_index}Get a mapped task instance
TaskInstanceApiget_mapped_task_instancesGET /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances/{task_id}/listMappedList mapped task instances
TaskInstanceApiget_task_instanceGET /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances/{task_id}Get a task instance
TaskInstanceApiget_task_instancesGET /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstancesList task instances
TaskInstanceApiget_task_instances_batchPOST /dags//dagRuns//taskInstances/listList task instances (batch)
TaskInstanceApipatch_mapped_task_instancePATCH /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances/{task_id}/{map_index}Updates the state of a mapped task instance
TaskInstanceApipatch_task_instancePATCH /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances/{task_id}Updates the state of a task instance
TaskInstanceApiset_mapped_task_instance_notePATCH /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances/{task_id}/{map_index}/setNoteUpdate the TaskInstance note.
TaskInstanceApiset_task_instance_notePATCH /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances/{task_id}/setNoteUpdate the TaskInstance note.
UserApidelete_userDELETE /users/{username}Delete a user
UserApiget_userGET /users/{username}Get a user
UserApiget_usersGET /usersList users
UserApipatch_userPATCH /users/{username}Update a user
UserApipost_userPOST /usersCreate a user
VariableApidelete_variableDELETE /variables/{variable_key}Delete a variable
VariableApiget_variableGET /variables/{variable_key}Get a variable
VariableApiget_variablesGET /variablesList variables
VariableApipatch_variablePATCH /variables/{variable_key}Update a variable
VariableApipost_variablesPOST /variablesCreate a variable
XComApiget_xcom_entriesGET /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances/{task_id}/xcomEntriesList XCom entries
XComApiget_xcom_entryGET /dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances/{task_id}/xcomEntries/{xcom_key}Get an XCom entry

Documentation For Models

Documentation For Authorization

By default the generated client supports the three authentication schemes:

  • Basic
  • GoogleOpenID
  • Kerberos

However, you can generate client and documentation with your own schemes by adding your own schemes in the security section of the OpenAPI specification. You can do it with Breeze CLI by adding the --security-schemes option to the breeze release-management prepare-python-client command.

Basic “smoke” tests

You can run basic smoke tests to check if the client is working properly - we have a simple test script that uses the API to run the tests. To do that, you need to:

  • install the apache-airflow-client package as described above
  • install rich Python package
  • download the test_python_client.py file
  • make sure you have test airflow installation running. Do not experiment with your production deployment
  • configure your airflow webserver to enable basic authentication In the [api] section of your airflow.cfg set:
auth_backend = airflow.api.auth.backend.session,airflow.api.auth.backend.basic_auth

You can also set it by env variable: export AIRFLOW__API__AUTH_BACKENDS=airflow.api.auth.backend.session,airflow.api.auth.backend.basic_auth

  • configure your airflow webserver to load example dags In the [core] section of your airflow.cfg set:
load_examples = True

You can also set it by env variable: export AIRFLOW__CORE__LOAD_EXAMPLES=True

  • optionally expose configuration (NOTE! that this is dangerous setting). The script will happily run with the default setting, but if you want to see the configuration, you need to expose it. In the [webserver] section of your airflow.cfg set:
expose_config = True

You can also set it by env variable: export AIRFLOW__WEBSERVER__EXPOSE_CONFIG=True

  • Configure your host/ip/user/password in the test_python_client.py file
import airflow_client

# Configure HTTP basic authorization: Basic
configuration = airflow_client.client.Configuration(
    host="http://localhost:8080/api/v1", username="admin", password="admin"
  • Run scheduler (or dag file processor you have setup with standalone dag file processor) for few parsing loops (you can pass --num-runs parameter to it or keep it running in the background). The script relies on example DAGs being serialized to the DB and this only happens when scheduler runs with core/load_examples set to True.

  • Run webserver - reachable at the host/port for the test script you want to run. Make sure it had enough time to initialize.

Run python test_python_client.py and you should see colored output showing attempts to connect and status.

Notes for Large OpenAPI documents

If the OpenAPI document is large, imports in client.apis and client.models may fail with a RecursionError indicating the maximum recursion limit has been exceeded. In that case, there are a couple of solutions:

Solution 1: Use specific imports for apis and models like:

  • from airflow_client.client.api.default_api import DefaultApi
  • from airflow_client.client.model.pet import Pet

Solution 2: Before importing the package, adjust the maximum recursion limit as shown below:

import sys

import airflow_client.client
from airflow_client.client.apis import *
from airflow_client.client.models import *