sagemaker.mlops.workflow.retry

Pipeline parameters and conditions for workflow.

Classes

RetryPolicy([backoff_rate, ...])	RetryPolicy base class
SageMakerJobExceptionTypeEnum(*args[, value])	SageMaker Job ExceptionType enum.
SageMakerJobStepRetryPolicy([...])	RetryPolicy for exception thrown by SageMaker Job.
StepExceptionTypeEnum(*args[, value])	Step ExceptionType enum.
StepRetryPolicy(exception_types[, ...])	RetryPolicy for a retryable step.

class sagemaker.mlops.workflow.retry.RetryPolicy(backoff_rate: float = 2.0, interval_seconds: int = 1, max_attempts: int | None = None, expire_after_mins: int | None = None)

Bases: Entity

RetryPolicy base class

backoff_rate

The multiplier by which the retry interval increases during each attempt (default: 2.0)

Type:

float

interval_seconds

An integer that represents the number of seconds before the first retry attempt (default: 1)

Type:

int

max_attempts

A positive integer that represents the maximum number of retry attempts. (default: None)

Type:

int

expire_after_mins

A positive integer that represents the maximum minute to expire any further retry attempt (default: None)

Type:

int

backoff_rate: float

expire_after_mins: int

interval_seconds: int

max_attempts: int

to_request() → Dict[str, Any] | List[Dict[str, Any]]

Get the request structure for workflow service calls.

validate_backoff_rate(_, value)

Validate the input back off rate type

validate_expire_after_mins(_, value)

Validate expire after mins

validate_interval_seconds(_, value)

Validate the input interval seconds

validate_max_attempts(_, value)

Validate the input max attempts

class sagemaker.mlops.workflow.retry.SageMakerJobExceptionTypeEnum(*args, value=<object object>, **kwargs)

Bases: Enum

SageMaker Job ExceptionType enum.

CAPACITY_ERROR = 'SageMaker.CAPACITY_ERROR'

INTERNAL_ERROR = 'SageMaker.JOB_INTERNAL_ERROR'

RESOURCE_LIMIT = 'SageMaker.RESOURCE_LIMIT'

class sagemaker.mlops.workflow.retry.SageMakerJobStepRetryPolicy(exception_types: List[SageMakerJobExceptionTypeEnum] | None = None, failure_reason_types: List[SageMakerJobExceptionTypeEnum] | None = None, backoff_rate: float = 2.0, interval_seconds: int = 1, max_attempts: int | None = None, expire_after_mins: int | None = None)

Bases: RetryPolicy

RetryPolicy for exception thrown by SageMaker Job.

exception_types

The SageMaker exception to match for this policy. The SageMaker exceptions captured here are the exceptions thrown by synchronously creating the job. For instance the resource limit exception.

Type:

List[SageMakerJobExceptionTypeEnum]

failure_reason_types

the SageMaker failure reason types to match for this policy. The failure reason type is presented in FailureReason field of the Describe response, it indicates the runtime failure reason for a job.

Type:

List[SageMakerJobExceptionTypeEnum]

backoff_rate

The multiplier by which the retry interval increases during each attempt (default: 2.0)

Type:

float

interval_seconds

An integer that represents the number of seconds before the first retry attempt (default: 1)

Type:

int

max_attempts

A positive integer that represents the maximum number of retry attempts. (default: None)

Type:

int

expire_after_mins

A positive integer that represents the maximum minute to expire any further retry attempt (default: None)

Type:

int

backoff_rate: float

exception_type_list: List[SageMakerJobExceptionTypeEnum]

expire_after_mins: int

interval_seconds: int

max_attempts: int

to_request() → Dict[str, Any] | List[Dict[str, Any]]

Gets the request structure for retry policy.

class sagemaker.mlops.workflow.retry.StepExceptionTypeEnum(*args, value=<object object>, **kwargs)

Bases: Enum

Step ExceptionType enum.

SERVICE_FAULT = 'Step.SERVICE_FAULT'

THROTTLING = 'Step.THROTTLING'

class sagemaker.mlops.workflow.retry.StepRetryPolicy(exception_types: List[StepExceptionTypeEnum], backoff_rate: float = 2.0, interval_seconds: int = 1, max_attempts: int | None = None, expire_after_mins: int | None = None)

Bases: RetryPolicy

RetryPolicy for a retryable step. The pipeline service will retry

sagemaker.workflow.retry.StepRetryExceptionTypeEnum.SERVICE_FAULT and sagemaker.workflow.retry.StepRetryExceptionTypeEnum.THROTTLING regardless of pipeline step type by default. However, for step defined as retryable, you can override them by specifying a StepRetryPolicy.

exception_types

the exception types to match for this policy

Type:

List[StepExceptionTypeEnum]

backoff_rate

The multiplier by which the retry interval increases during each attempt (default: 2.0)

Type:

float

interval_seconds

An integer that represents the number of seconds before the first retry attempt (default: 1)

Type:

int

max_attempts

A positive integer that represents the maximum number of retry attempts. (default: None)

Type:

int

expire_after_mins

A positive integer that represents the maximum minute to expire any further retry attempt (default: None)

Type:

int

backoff_rate: float

expire_after_mins: int

interval_seconds: int

max_attempts: int

to_request() → Dict[str, Any] | List[Dict[str, Any]]

Gets the request structure for retry policy.