airflow.providers.amazon.aws.sensors.sagemaker

Module Contents

Classes

SageMakerBaseSensor

Contains general sensor behavior for SageMaker.

SageMakerEndpointSensor

Poll the endpoint state until it reaches a terminal state; raise AirflowException with the failure reason.

SageMakerTransformSensor

Poll the transform job until it reaches a terminal state; raise AirflowException with the failure reason.

SageMakerTuningSensor

Poll the tuning state until it reaches a terminal state; raise AirflowException with the failure reason.

SageMakerTrainingSensor

Poll the training job until it reaches a terminal state; raise AirflowException with the failure reason.

SageMakerPipelineSensor

Poll the pipeline until it reaches a terminal state; raise AirflowException with the failure reason.

SageMakerAutoMLSensor

Poll the auto ML job until it reaches a terminal state; raise AirflowException with the failure reason.

class airflow.providers.amazon.aws.sensors.sagemaker.SageMakerBaseSensor(*, aws_conn_id='aws_default', resource_type='job', **kwargs)[source]

Bases: airflow.sensors.base.BaseSensorOperator

Contains general sensor behavior for SageMaker.

Subclasses should implement get_sagemaker_response() and state_from_response() methods. Subclasses should also implement NON_TERMINAL_STATES and FAILED_STATE methods.

ui_color = '#ededed'[source]
get_hook()[source]

Get SageMakerHook.

hook()[source]
poke(context)[source]

Override when deriving this class.

abstract non_terminal_states()[source]

Return states with should not terminate.

abstract failed_states()[source]

Return states with are considered failed.

abstract get_sagemaker_response()[source]

Check status of a SageMaker task.

get_failed_reason_from_response(response)[source]

Extract the reason for failure from an AWS response.

abstract state_from_response(response)[source]

Extract the state from an AWS response.

class airflow.providers.amazon.aws.sensors.sagemaker.SageMakerEndpointSensor(*, endpoint_name, **kwargs)[source]

Bases: SageMakerBaseSensor

Poll the endpoint state until it reaches a terminal state; raise AirflowException with the failure reason.

See also

For more information on how to use this sensor, take a look at the guide: Wait on an Amazon SageMaker endpoint state

Parameters

endpoint_name – Name of the endpoint instance to watch.

template_fields: Sequence[str] = ('endpoint_name',)[source]
template_ext: Sequence[str] = ()[source]
non_terminal_states()[source]

Return states with should not terminate.

failed_states()[source]

Return states with are considered failed.

get_sagemaker_response()[source]

Check status of a SageMaker task.

get_failed_reason_from_response(response)[source]

Extract the reason for failure from an AWS response.

state_from_response(response)[source]

Extract the state from an AWS response.

class airflow.providers.amazon.aws.sensors.sagemaker.SageMakerTransformSensor(*, job_name, **kwargs)[source]

Bases: SageMakerBaseSensor

Poll the transform job until it reaches a terminal state; raise AirflowException with the failure reason.

See also

For more information on how to use this sensor, take a look at the guide: Wait on an Amazon SageMaker transform job state

Parameters

job_name (str) – Name of the transform job to watch.

template_fields: Sequence[str] = ('job_name',)[source]
template_ext: Sequence[str] = ()[source]
non_terminal_states()[source]

Return states with should not terminate.

failed_states()[source]

Return states with are considered failed.

get_sagemaker_response()[source]

Check status of a SageMaker task.

get_failed_reason_from_response(response)[source]

Extract the reason for failure from an AWS response.

state_from_response(response)[source]

Extract the state from an AWS response.

class airflow.providers.amazon.aws.sensors.sagemaker.SageMakerTuningSensor(*, job_name, **kwargs)[source]

Bases: SageMakerBaseSensor

Poll the tuning state until it reaches a terminal state; raise AirflowException with the failure reason.

See also

For more information on how to use this sensor, take a look at the guide: Wait on an Amazon SageMaker tuning job state

Parameters

job_name (str) – Name of the tuning instance to watch.

template_fields: Sequence[str] = ('job_name',)[source]
template_ext: Sequence[str] = ()[source]
non_terminal_states()[source]

Return states with should not terminate.

failed_states()[source]

Return states with are considered failed.

get_sagemaker_response()[source]

Check status of a SageMaker task.

get_failed_reason_from_response(response)[source]

Extract the reason for failure from an AWS response.

state_from_response(response)[source]

Extract the state from an AWS response.

class airflow.providers.amazon.aws.sensors.sagemaker.SageMakerTrainingSensor(*, job_name, print_log=True, **kwargs)[source]

Bases: SageMakerBaseSensor

Poll the training job until it reaches a terminal state; raise AirflowException with the failure reason.

See also

For more information on how to use this sensor, take a look at the guide: Wait on an Amazon SageMaker training job state

Parameters
  • job_name – Name of the training job to watch.

  • print_log – Prints the cloudwatch log if True; Defaults to True.

template_fields: Sequence[str] = ('job_name',)[source]
template_ext: Sequence[str] = ()[source]
init_log_resource(hook)[source]

Set tailing LogState for associated training job.

non_terminal_states()[source]

Return states with should not terminate.

failed_states()[source]

Return states with are considered failed.

get_sagemaker_response()[source]

Check status of a SageMaker task.

get_failed_reason_from_response(response)[source]

Extract the reason for failure from an AWS response.

state_from_response(response)[source]

Extract the state from an AWS response.

class airflow.providers.amazon.aws.sensors.sagemaker.SageMakerPipelineSensor(*, pipeline_exec_arn, verbose=True, **kwargs)[source]

Bases: SageMakerBaseSensor

Poll the pipeline until it reaches a terminal state; raise AirflowException with the failure reason.

See also

For more information on how to use this sensor, take a look at the guide: Wait on an Amazon SageMaker pipeline execution state

Parameters
  • pipeline_exec_arn (str) – ARN of the pipeline to watch.

  • verbose (bool) – Whether to print steps details while waiting for completion. Defaults to true, consider turning off for pipelines that have thousands of steps.

template_fields: Sequence[str] = ('pipeline_exec_arn',)[source]
non_terminal_states()[source]

Return states with should not terminate.

failed_states()[source]

Return states with are considered failed.

get_sagemaker_response()[source]

Check status of a SageMaker task.

state_from_response(response)[source]

Extract the state from an AWS response.

class airflow.providers.amazon.aws.sensors.sagemaker.SageMakerAutoMLSensor(*, job_name, **kwargs)[source]

Bases: SageMakerBaseSensor

Poll the auto ML job until it reaches a terminal state; raise AirflowException with the failure reason.

See also

For more information on how to use this sensor, take a look at the guide: Wait on an Amazon SageMaker AutoML experiment state

Parameters

job_name (str) – unique name of the AutoML job to watch.

template_fields: Sequence[str] = ('job_name',)[source]
non_terminal_states()[source]

Return states with should not terminate.

failed_states()[source]

Return states with are considered failed.

get_sagemaker_response()[source]

Check status of a SageMaker task.

state_from_response(response)[source]

Extract the state from an AWS response.

Was this entry helpful?