def disable_data_capture(self):
"""Updates the DataCaptureConfig for the Predictor's associated Amazon SageMaker Endpoint
to disable data capture. For a more customized experience, refer to
update_data_capture_config, instead.
"""
self.update_data_capture_config(data_capture_config=DataCaptureConfig(
enable_capture=False, sagemaker_session=self.sagemaker_session))
def deploy_model(self):
if self.monitor:
from sagemaker.model_monitor import DataCaptureConfig
data_capture_config = DataCaptureConfig(
enable_capture=True,
sampling_percentage=100,
destination_s3_uri="s3://{}/ezsmdeploy/model-{}/datacapture".
format(self.bucket, self.name),
)
else:
data_capture_config = None
self.predictor = self.sagemakermodel.deploy(
initial_instance_count=self.instance_count,
instance_type=self.instance_type,
accelerator_type=self.ei,
endpoint_name="ezsmdeploy-endpoint-" + self.name,
update_endpoint=False,
wait=self.wait,
data_capture_config=data_capture_config,
)
self.endpoint_name = "ezsmdeploy-endpoint-" + self.name
def endpoint_name(sagemaker_session):
endpoint_name = unique_name_from_base("model-quality-monitor-integ")
xgb_model_data = sagemaker_session.upload_data(
path=os.path.join(XGBOOST_DATA_PATH, "xgb_model.tar.gz"),
key_prefix="integ-test-data/xgboost/model",
)
xgb_image = image_uris.retrieve("xgboost",
sagemaker_session.boto_region_name,
version="1",
image_scope="inference")
with tests.integ.timeout.timeout_and_delete_endpoint_by_name(
endpoint_name=endpoint_name,
sagemaker_session=sagemaker_session,
hours=2):
xgb_model = Model(
model_data=xgb_model_data,
image_uri=xgb_image,
name=endpoint_name, # model name
role=ROLE,
sagemaker_session=sagemaker_session,
)
xgb_model.deploy(
INSTANCE_COUNT,
INSTANCE_TYPE,
endpoint_name=endpoint_name,
data_capture_config=DataCaptureConfig(
True, sagemaker_session=sagemaker_session),
)
yield endpoint_name
def enable_data_capture(self):
"""Enables data capture by updating DataCaptureConfig.
This function updates the DataCaptureConfig for the Predictor's associated Amazon SageMaker
Endpoint to enable data capture. For a more customized experience, refer to
update_data_capture_config, instead.
"""
self.update_data_capture_config(data_capture_config=DataCaptureConfig(
enable_capture=True, sagemaker_session=self.sagemaker_session))
def test_to_request_dict_returns_correct_default_params_when_optionals_not_provided():
data_capture_config = DataCaptureConfig(
enable_capture=DEFAULT_ENABLE_CAPTURE, destination_s3_uri=DEFAULT_DESTINATION_S3_URI
)
assert data_capture_config.enable_capture == DEFAULT_ENABLE_CAPTURE
assert data_capture_config.sampling_percentage == DEFAULT_SAMPLING_PERCENTAGE
assert data_capture_config.destination_s3_uri == DEFAULT_DESTINATION_S3_URI
assert data_capture_config.kms_key_id == DEFAULT_KMS_KEY_ID
assert data_capture_config.csv_content_types == DEFAULT_CSV_CONTENT_TYPES
assert data_capture_config.json_content_types == DEFAULT_JSON_CONTENT_TYPES
def test_init_when_optionals_not_provided():
sagemaker_session = Mock()
sagemaker_session.default_bucket.return_value = DEFAULT_BUCKET_NAME
data_capture_config = DataCaptureConfig(
enable_capture=DEFAULT_ENABLE_CAPTURE,
sagemaker_session=sagemaker_session)
assert data_capture_config.enable_capture == DEFAULT_ENABLE_CAPTURE
assert data_capture_config.sampling_percentage == DEFAULT_SAMPLING_PERCENTAGE
assert data_capture_config.destination_s3_uri == DEFAULT_DESTINATION_S3_URI
assert data_capture_config.kms_key_id == DEFAULT_KMS_KEY_ID
assert data_capture_config.csv_content_types == DEFAULT_CSV_CONTENT_TYPES
assert data_capture_config.json_content_types == DEFAULT_JSON_CONTENT_TYPES
def test_endpoint_config_step_creation(pca_model):
data_capture_config = DataCaptureConfig(
enable_capture=True,
sampling_percentage=100,
destination_s3_uri='s3://sagemaker/datacapture')
step = EndpointConfigStep(
'Endpoint Config',
endpoint_config_name='MyEndpointConfig',
model_name='pca-model',
initial_instance_count=1,
instance_type='ml.p2.xlarge',
data_capture_config=data_capture_config,
tags=DEFAULT_TAGS,
)
assert step.to_dict() == {
'Type': 'Task',
'Parameters': {
'EndpointConfigName':
'MyEndpointConfig',
'ProductionVariants': [{
'InitialInstanceCount': 1,
'InstanceType': 'ml.p2.xlarge',
'ModelName': 'pca-model',
'VariantName': 'AllTraffic'
}],
'DataCaptureConfig': {
'EnableCapture':
True,
'InitialSamplingPercentage':
100,
'DestinationS3Uri':
's3://sagemaker/datacapture',
'CaptureOptions': [{
'CaptureMode': 'Input'
}, {
'CaptureMode': 'Output'
}],
'CaptureContentTypeHeader': {
'CsvContentTypes': ['text/csv'],
'JsonContentTypes': ['application/json']
}
},
'Tags':
DEFAULT_TAGS_LIST
},
'Resource': 'arn:aws:states:::sagemaker:createEndpointConfig',
'End': True
}
def test_init_when_non_defaults_provided():
data_capture_config = DataCaptureConfig(
enable_capture=NON_DEFAULT_ENABLE_CAPTURE,
sampling_percentage=NON_DEFAULT_SAMPLING_PERCENTAGE,
destination_s3_uri=NON_DEFAULT_DESTINATION_S3_URI,
kms_key_id=NON_DEFAULT_KMS_KEY_ID,
csv_content_types=NON_DEFAULT_CSV_CONTENT_TYPES,
json_content_types=NON_DEFAULT_JSON_CONTENT_TYPES,
)
assert data_capture_config.enable_capture == NON_DEFAULT_ENABLE_CAPTURE
assert data_capture_config.sampling_percentage == NON_DEFAULT_SAMPLING_PERCENTAGE
assert data_capture_config.destination_s3_uri == NON_DEFAULT_DESTINATION_S3_URI
assert data_capture_config.kms_key_id == NON_DEFAULT_KMS_KEY_ID
assert data_capture_config.csv_content_types == NON_DEFAULT_CSV_CONTENT_TYPES
assert data_capture_config.json_content_types == NON_DEFAULT_JSON_CONTENT_TYPES
def capture():
# Change parameters as you would like - adjust sampling percentage,
# chose to capture request or response or both.
# Learn more from our documentation
data_capture_config = DataCaptureConfig(
enable_capture = True,
sampling_percentage=50,
destination_s3_uri=s3_capture_upload_path,
kms_key_id=None,
capture_options=["REQUEST", "RESPONSE"],
csv_content_types=["text/csv"],
json_content_types=["application/json"])
# Now it is time to apply the new configuration and wait for it to be applied
predictor = RealTimePredictor(endpoint=endpoint_name)
predictor.update_data_capture_config(data_capture_config=data_capture_config)
sm_session.wait_for_endpoint(endpoint=endpoint_name)
def test_disabling_data_capture_on_endpoint_shows_correct_data_capture_status(
sagemaker_session, tensorflow_inference_latest_version):
endpoint_name = unique_name_from_base("sagemaker-tensorflow-serving")
model_data = sagemaker_session.upload_data(
path=os.path.join(tests.integ.DATA_DIR,
"tensorflow-serving-test-model.tar.gz"),
key_prefix="tensorflow-serving/models",
)
with tests.integ.timeout.timeout_and_delete_endpoint_by_name(
endpoint_name, sagemaker_session):
model = TensorFlowModel(
model_data=model_data,
role=ROLE,
framework_version=tensorflow_inference_latest_version,
sagemaker_session=sagemaker_session,
)
destination_s3_uri = os.path.join("s3://",
sagemaker_session.default_bucket(),
endpoint_name, "custom")
predictor = model.deploy(
initial_instance_count=INSTANCE_COUNT,
instance_type=INSTANCE_TYPE,
endpoint_name=endpoint_name,
data_capture_config=DataCaptureConfig(
enable_capture=True,
sampling_percentage=CUSTOM_SAMPLING_PERCENTAGE,
destination_s3_uri=destination_s3_uri,
capture_options=CUSTOM_CAPTURE_OPTIONS,
csv_content_types=CUSTOM_CSV_CONTENT_TYPES,
json_content_types=CUSTOM_JSON_CONTENT_TYPES,
sagemaker_session=sagemaker_session,
),
)
endpoint_desc = sagemaker_session.sagemaker_client.describe_endpoint(
EndpointName=predictor.endpoint_name)
endpoint_config_desc = sagemaker_session.sagemaker_client.describe_endpoint_config(
EndpointConfigName=endpoint_desc["EndpointConfigName"])
assert endpoint_config_desc["DataCaptureConfig"]["EnableCapture"]
assert (endpoint_config_desc["DataCaptureConfig"]
["InitialSamplingPercentage"] == CUSTOM_SAMPLING_PERCENTAGE)
assert endpoint_config_desc["DataCaptureConfig"]["CaptureOptions"] == [
{
"CaptureMode": "Input"
}
]
assert (endpoint_config_desc["DataCaptureConfig"]
["CaptureContentTypeHeader"]["CsvContentTypes"] ==
CUSTOM_CSV_CONTENT_TYPES)
assert (endpoint_config_desc["DataCaptureConfig"]
["CaptureContentTypeHeader"]["JsonContentTypes"] ==
CUSTOM_JSON_CONTENT_TYPES)
predictor.disable_data_capture()
# Wait for endpoint to finish updating
# Endpoint update takes ~7min. 25 retries * 60s sleeps = 25min timeout
for _ in retries(
max_retry_count=25,
exception_message_prefix=
"Waiting for 'InService' endpoint status",
seconds_to_sleep=60,
):
new_endpoint = sagemaker_session.sagemaker_client.describe_endpoint(
EndpointName=predictor.endpoint_name)
if new_endpoint["EndpointStatus"] == "InService":
break
endpoint_desc = sagemaker_session.sagemaker_client.describe_endpoint(
EndpointName=predictor.endpoint_name)
endpoint_config_desc = sagemaker_session.sagemaker_client.describe_endpoint_config(
EndpointConfigName=endpoint_desc["EndpointConfigName"])
assert not endpoint_config_desc["DataCaptureConfig"]["EnableCapture"]
def main(datacapture=False):
# Load config from environment and set required defaults
# AWS especific
AWS_DEFAULT_REGION = os.getenv('AWS_DEFAULT_REGION', 'eu-west-1')
AWS_PROFILE = os.getenv('AWS_PROFILE', 'default')
AWS_ACCESS_KEY_ID = os.getenv('AWS_ACCESS_KEY_ID', None)
AWS_SECRET_ACCESS_KEY = os.getenv('AWS_SECRET_ACCESS_KEY', None)
b3_session, sm_client, sm_runtime, sm_session = get_sm_session(
region=AWS_DEFAULT_REGION,
profile_name=AWS_PROFILE,
aws_access_key_id=AWS_ACCESS_KEY_ID,
aws_secret_access_key=AWS_SECRET_ACCESS_KEY
)
ROLE_ARN = os.getenv('AWS_ROLE', sagemaker.get_execution_role())
MODEL_PACKAGE_GROUP_NAME = os.getenv(
'MODEL_PACKAGE_GROUP_NAME', 'sts-sklearn-grp')
BASE_JOB_PREFIX = os.getenv('BASE_JOB_PREFIX', 'sts')
# define useful const's
bucket = sm_session.default_bucket()
endpoint_name = "{}-sklearn-{}".format(
BASE_JOB_PREFIX,
datetime.datetime.now().strftime("%Y%m%d%H%M")
)
prefix = "{}/{}".format(BASE_JOB_PREFIX, endpoint_name)
data_capture_prefix = "{}/datacapture".format(prefix)
s3_capture_upload_path = "s3://{}/{}".format(bucket, data_capture_prefix)
# outputs is a dict to save to json
outputs = dict()
if datacapture is True:
# if data capture was enabled output the S3 Uri for data capture
outputs['monitor'] = {
's3_capture_upload_path': s3_capture_upload_path
}
# get the last version aproved in the model package group
model_package_arn = get_approved_package(
MODEL_PACKAGE_GROUP_NAME, sm_client)
_l.info(f"Latest approved model package: {model_package_arn}")
model_info = sm_client.describe_model_package(
ModelPackageName=model_package_arn)
outputs['model_info'] = model_info
model_uri = model_info.get(
'InferenceSpecification')['Containers'][0]['ModelDataUrl']
_l.info(f"Model data uri: {model_uri}")
sk_model = SKLearnModel(
model_uri, # s3 uri for the model.tar.gz
ROLE_ARN, # sagemaker role to be used
'model_loader.py', # script to load the model
framework_version='0.23-1'
)
data_capture_config=None
if datacapture is True:
# if data capture was enabled generated the required config
_l.info("Enabling data capture as requested")
_l.info(f"s3_capture_upload_path: {s3_capture_upload_path}")
data_capture_config = DataCaptureConfig(
enable_capture=True, sampling_percentage=100,
destination_s3_uri=s3_capture_upload_path,
capture_options=["REQUEST", "RESPONSE"],
sagemaker_session=sm_session
)
# Deploy the endpoint
predictor = sk_model.deploy(
instance_type="ml.m5.xlarge",
initial_instance_count=1,
serializer=CSVSerializer(),
deserializer=CSVDeserializer(),
data_capture_config=data_capture_config,
endpoint_name=endpoint_name
)
_l.info(f"Endpoint name: {predictor.endpoint_name}")
outputs['endpoint'] = {
'name': endpoint_name,
'config_name': predictor.endpoint_name # is the same as the endpoint ?
}
outputs['model_info'].update({"name": sk_model.name})
# ENDPOINT deploy done
# save useful outputs to a file
with open('deploymodel_out.json', 'w') as f:
json.dump(outputs, f, default=json_default)
sagemaker_session = sagemaker.session.Session(
boto_session=boto_session,
sagemaker_client=sagemaker_boto_client
)
# Enable real-time inference data capture
s3_capture_upload_path = f's3://{bucket}/{prefix}/endpoint-data-capture/' #example: s3://bucket-name/path/to/endpoint-data-capture/
# Change parameters as you would like - adjust sampling percentage,
# chose to capture request or response or both
data_capture_config = DataCaptureConfig(
enable_capture = True,
sampling_percentage=25,
destination_s3_uri=s3_capture_upload_path,
kms_key_id=None,
capture_options=["REQUEST", "RESPONSE"],
csv_content_types=["text/csv"],
json_content_types=["application/json"]
)
# Now it is time to apply the new configuration
predictor = Predictor(endpoint_name=endpoint_name, sagemaker_session=sagemaker_session)
predictor.update_data_capture_config(data_capture_config=data_capture_config)
print('Created Predictor at endpoint {}'.format(endpoint_name))
baseline_data_uri = args.baseline_data_uri ##'s3://bucketname/path/to/baseline/data' - Where your validation data is
baseline_results_uri = f's3://{bucket}/{prefix}/baseline/results' ##'s3://bucketname/path/to/baseline/data' - Where the results are to be stored in
print('Baseline data is at {}'.format(baseline_data_uri))
from sagemaker.model_monitor import DataCaptureConfig
s3_capture_path = "s3://monitoring/xgb-churn-data"
data_capture_config = DataCaptureConfig(
enable_capture=True, sampling_percentage=100, destination_s3_uri=s3_capture_path
)
from sagemaker.deserializers import CSVDeserializer
predictor = model.deploy(
initial_instance_count=1,
instance_type="ml.m4.large",
endpoint_name="xgb-churn-monitor",
data_capture_config=data_capture_config,
deserializer=CSVDeserializer(),
)
'Containers':
check_autopilot_job_status.output()['Payload']['InferenceContainers'],
'ModelName':
execution_input['ModelName'],
'ExecutionRoleArn':
sagemaker_exec_role
})
endpoint_config_step = EndpointConfigStep(
'CreateModelEndpointConfig',
endpoint_config_name=execution_input['EndpointConfigName'],
model_name=execution_input['ModelName'],
initial_instance_count=1,
instance_type='ml.m4.xlarge',
data_capture_config=DataCaptureConfig(
enable_capture=True,
sampling_percentage=100,
))
endpoint_step = EndpointStep(
'UpdateModelEndpoint',
endpoint_name=execution_input['EndpointName'],
endpoint_config_name=execution_input['EndpointConfigName'],
update=False)
# define Amazon CodeBuild Step Functions Task
deploy_rest_api_task = Task(
'DeployRestAPI',
resource='arn:aws:states:::codebuild:startBuild.sync',
parameters={
'ProjectName':
utils.get_api_codebuild_project(),