airflow.executors.celery_executor¶
CeleryExecutor
See also
For more information on how the CeleryExecutor works, take a look at the guide: Celery Executor
Module Contents¶
Classes¶
Wrapper class used to propagate exceptions to parent processes from subprocesses. |
|
CeleryExecutor is recommended for production use of Airflow. It allows |
|
Gets status for many Celery tasks using the best method available |
Functions¶
|
Executes command. |
|
Sends task to executor. |
|
Preload some "expensive" airflow modules so that every task process doesn't have to import it again and |
|
Fetch and return the state of the given celery task. The scope of this function is |
Attributes¶
To start the celery worker, run the command: |
|
- airflow.executors.celery_executor.CELERY_FETCH_ERR_MSG_HEADER = Error fetching Celery task state[source]¶
- airflow.executors.celery_executor.OPERATION_TIMEOUT[source]¶
To start the celery worker, run the command: airflow celery worker
- class airflow.executors.celery_executor.ExceptionWithTraceback(exception, exception_traceback)[source]¶
Wrapper class used to propagate exceptions to parent processes from subprocesses.
- airflow.executors.celery_executor.send_task_to_executor(task_tuple)[source]¶
Sends task to executor.
- airflow.executors.celery_executor.on_celery_import_modules(*args, **kwargs)[source]¶
Preload some “expensive” airflow modules so that every task process doesn’t have to import it again and again.
Loading these for each task adds 0.3-0.5s per task before the task can run. For long running tasks this doesn’t matter, but for short tasks this starts to be a noticeable impact.
- class airflow.executors.celery_executor.CeleryExecutor[source]¶
Bases:
airflow.executors.base_executor.BaseExecutorCeleryExecutor is recommended for production use of Airflow. It allows distributing the execution of task instances to multiple worker nodes.
Celery is a simple, flexible and reliable distributed system to process vast amounts of messages, while providing operations with the tools required to maintain such a system.
- sync()[source]¶
Sync will get called periodically by the heartbeat method. Executors should override this to perform gather statuses.
- change_state(key, state, info=None)[source]¶
Changes state of the task.
- Parameters
info – Executor information for the task instance
key (airflow.models.taskinstance.TaskInstanceKey) – Unique key for the task instance
state (str) – State to set for the task.
- end(synchronous=False)[source]¶
This method is called when the caller is done submitting job and wants to wait synchronously for the job submitted previously to be all done.
- try_adopt_task_instances(tis)[source]¶
Try to adopt running task instances that have been abandoned by a SchedulerJob dying.
Anything that is not adopted will be cleared by the scheduler (and then become eligible for re-scheduling)
- Returns
any TaskInstances that were unable to be adopted
- Return type
Sequence[airflow.models.taskinstance.TaskInstance]
- airflow.executors.celery_executor.fetch_celery_task_state(async_result)[source]¶
Fetch and return the state of the given celery task. The scope of this function is global so that it can be called by subprocesses in the pool.
- Parameters
async_result (celery.result.AsyncResult) – a tuple of the Celery task key and the async Celery object used to fetch the task’s state
- Returns
a tuple of the Celery task key and the Celery state and the celery info of the task
- Return type
tuple[str, str | ExceptionWithTraceback, Any]
- class airflow.executors.celery_executor.BulkStateFetcher(sync_parallelism=None)[source]¶
Bases:
airflow.utils.log.logging_mixin.LoggingMixinGets status for many Celery tasks using the best method available
If BaseKeyValueStoreBackend is used as result backend, the mget method is used. If DatabaseBackend is used as result backend, the SELECT …WHERE task_id IN (…) query is used Otherwise, multiprocessing.Pool will be used. Each task status will be downloaded individually.