astronomer.providers.core.sensors.external_task

Classes

ExternalTaskSensorAsync	Waits for a different DAG, task group, or task to complete for a specific logical date.
ExternalDeploymentTaskSensorAsync	External deployment task sensor Make HTTP call and poll for the response state of externally

Module Contents

class astronomer.providers.core.sensors.external_task.ExternalTaskSensorAsync(poke_interval=5.0, **kwargs)

Bases: airflow.sensors.external_task.ExternalTaskSensor

Waits for a different DAG, task group, or task to complete for a specific logical date.

If both external_task_group_id and external_task_id are None (default), the sensor waits for the DAG. Values for external_task_group_id and external_task_id can’t be set at the same time.

By default, the ExternalTaskSensor will wait for the external task to succeed, at which point it will also succeed. However, by default it will not fail if the external task fails, but will continue to check the status until the sensor times out (thus giving you time to retry the external task without also having to clear the sensor).

By default, the ExternalTaskSensor will not skip if the external task skips. To change this, simply set skipped_states=[TaskInstanceState.SKIPPED]. Note that if you are monitoring multiple tasks, and one enters error state and the other enters a skipped state, then the external task will react to whichever one it sees first. If both happen together, then the failed state takes priority.

It is possible to alter the default behavior by setting states which cause the sensor to fail, e.g. by setting allowed_states=[DagRunState.FAILED] and failed_states=[DagRunState.SUCCESS] you will flip the behaviour to get a sensor which goes green when the external task fails and immediately goes red if the external task succeeds!

Note that soft_fail is respected when examining the failed_states. Thus if the external task enters a failed state and soft_fail == True the sensor will _skip_ rather than fail. As a result, setting soft_fail=True and failed_states=[DagRunState.SKIPPED] will result in the sensor skipping if the external task skips. However, this is a contrived example—consider using skipped_states if you would like this behaviour. Using skipped_states allows the sensor to skip if the target fails, but still enter failed state on timeout. Using soft_fail == True as above will cause the sensor to skip if the target fails, but also if it times out.

Parameters:

• external_dag_id – The dag_id that contains the task you want to wait for. (templated)

• external_task_id – The task_id that contains the task you want to wait for. (templated)

• external_task_ids – The list of task_ids that you want to wait for. (templated) If None (default value) the sensor waits for the DAG. Either external_task_id or external_task_ids can be passed to ExternalTaskSensor, but not both.

• external_task_group_id – The task_group_id that contains the task you want to wait for. (templated)

• allowed_states – Iterable of allowed states, default is ['success']

• skipped_states – Iterable of states to make this task mark as skipped, default is None

• failed_states – Iterable of failed or dis-allowed states, default is None

• execution_delta – time difference with the previous execution to look at, the default is the same logical date as the current task or DAG. For yesterday, use [positive!] datetime.timedelta(days=1). Either execution_delta or execution_date_fn can be passed to ExternalTaskSensor, but not both.

• execution_date_fn – function that receives the current execution’s logical date as the first positional argument and optionally any number of keyword arguments available in the context dictionary, and returns the desired logical dates to query. Either execution_delta or execution_date_fn can be passed to ExternalTaskSensor, but not both.

• check_existence – Set to True to check if the external task exists (when external_task_id is not None) or check if the DAG to wait for exists (when external_task_id is None), and immediately cease waiting if the external task or DAG does not exist (default value: False).

• poll_interval – polling period in seconds to check for the status

• deferrable – Run sensor in deferrable mode

is_deprecated = True

post_deprecation_replacement = 'from airflow.sensors.external_task import ExternalTaskSensor'

poke_interval

execute(context)

Correctly identify which trigger to execute, and defer execution as expected.

execute_complete(context, session, event=None)

Verifies that there is a success status for each task via execution date.

get_execution_dates(context)

Helper function to set execution dates depending on which context and/or internal fields are populated.

class astronomer.providers.core.sensors.external_task.ExternalDeploymentTaskSensorAsync(*args, **kwargs)

Bases: airflow.providers.http.sensors.http.HttpSensor

External deployment task sensor Make HTTP call and poll for the response state of externally deployed DAG task to complete. Inherits from HttpSensor, the host should be external deployment url, header with access token

See also

• Retrieve an access token and Deployment URL

Parameters:

• http_conn_id – The Connection ID to run the sensor against

• method – The HTTP request method to use

• endpoint – The relative part of the full url

• request_params – The parameters to be added to the GET url

• headers – The HTTP headers to be added to the GET request

• extra_options – Extra options for the ‘requests’ library, see the ‘requests’ documentation (options to modify timeout, ssl, etc.)

• tcp_keep_alive – Enable TCP Keep Alive for the connection.

• tcp_keep_alive_idle – The TCP Keep Alive Idle parameter (corresponds to socket.TCP_KEEPIDLE).

• tcp_keep_alive_count – The TCP Keep Alive count parameter (corresponds to socket.TCP_KEEPCNT)

• tcp_keep_alive_interval – The TCP Keep Alive interval parameter (corresponds to socket.TCP_KEEPINTVL)

• poke_interval – Time in seconds that the job should wait in between each tries

execute(context)

Defers trigger class to poll for state of the job run until it reaches a failure state or success state

execute_complete(context, event=None)

Callback for when the trigger fires - returns immediately. Return true and log the response if state is not success state raise ValueError