def describe(self, deployment_pb, repo=None):
deployment_spec = deployment_pb.spec
aws_config = deployment_spec.aws_lambda_operator_config
bento_path = repo.get(deployment_spec.bento_name,
deployment_spec.bento_version)
bento_config = load_bentoml_config(bento_path)
with TemporaryServerlessConfig(
archive_path=bento_path,
deployment_name=deployment_pb.name,
region=aws_config.region,
stage=deployment_pb.namespace,
provider_name='aws',
functions=generate_aws_handler_functions_config(
bento_config['apis']),
) as tempdir:
try:
response = call_serverless_command(["serverless", "info"],
tempdir)
info_json = parse_serverless_info_response_to_json_string(
response)
state = DeploymentState(state=DeploymentState.RUNNING,
info_json=info_json)
except BentoMLException as e:
state = DeploymentState(state=DeploymentState.ERROR,
error_message=str(e))
return DescribeDeploymentResponse(status=Status.OK(), state=state)
def describe(self, deployment_pb, yatai_service=None):
try:
deployment_spec = deployment_pb.spec
aws_config = deployment_spec.aws_lambda_operator_config
info_json = {'endpoints': []}
bento_pb = yatai_service.GetBento(
GetBentoRequest(
bento_name=deployment_spec.bento_name,
bento_version=deployment_spec.bento_version,
)
)
bento_service_metadata = bento_pb.bento.bento_service_metadata
api_names = (
[aws_config.api_name]
if aws_config.api_name
else [api.name for api in bento_service_metadata.apis]
)
try:
cloud_formation_stack_result = boto3.client(
'cloudformation'
).describe_stacks(
StackName='{name}-{ns}'.format(
ns=deployment_pb.namespace, name=deployment_pb.name
)
)
outputs = cloud_formation_stack_result.get('Stacks')[0]['Outputs']
except Exception as error:
state = DeploymentState(
state=DeploymentState.ERROR, error_message=str(error)
)
state.timestamp.GetCurrentTime()
return DescribeDeploymentResponse(
status=Status.INTERNAL(str(error)), state=state
)
base_url = ''
for output in outputs:
if output['OutputKey'] == 'ServiceEndpoint':
base_url = output['OutputValue']
break
if base_url:
info_json['endpoints'] = [
base_url + '/' + api_name for api_name in api_names
]
state = DeploymentState(
state=DeploymentState.RUNNING, info_json=json.dumps(info_json)
)
state.timestamp.GetCurrentTime()
return DescribeDeploymentResponse(status=Status.OK(), state=state)
except BentoMLException as error:
return DescribeDeploymentResponse(status=exception_to_return_status(error))
def apply(self, deployment_pb, repo, prev_deployment=None):
deployment_spec = deployment_pb.spec
gcp_config = deployment_spec.gcp_function_operator_config
bento_path = repo.get(deployment_spec.bento_name,
deployment_spec.bento_version)
bento_config = load_bentoml_config(bento_path)
with TemporaryServerlessContent(
archive_path=bento_path,
deployment_name=deployment_pb.name,
bento_name=deployment_spec.bento_name,
template_type='google-python',
) as output_path:
generate_main_py(bento_config['name'], bento_config['apis'],
output_path)
generate_serverless_configuration_for_gcp_function(
service_name=bento_config['name'],
apis=bento_config['apis'],
output_path=output_path,
region=gcp_config.region,
stage=deployment_pb.namespace,
)
call_serverless_command(["deploy"], output_path)
res_deployment_pb = Deployment(state=DeploymentState())
res_deployment_pb.CopyFrom(deployment_pb)
state = self.describe(res_deployment_pb, repo).state
res_deployment_pb.state.CopyFrom(state)
return ApplyDeploymentResponse(status=Status.OK(),
deployment=res_deployment_pb)
def describe(self, deployment_pb):
try:
deployment_spec = deployment_pb.spec
sagemaker_config = deployment_spec.sagemaker_operator_config
sagemaker_client = boto3.client('sagemaker', sagemaker_config.region)
_, _, sagemaker_endpoint_name = _get_sagemaker_resource_names(deployment_pb)
try:
endpoint_status_response = sagemaker_client.describe_endpoint(
EndpointName=sagemaker_endpoint_name
)
except ClientError as e:
raise _aws_client_error_to_bentoml_exception(
e,
f"Failed to fetch current status of sagemaker endpoint "
f"'{sagemaker_endpoint_name}'",
)
logger.debug("AWS describe endpoint response: %s", endpoint_status_response)
endpoint_status = endpoint_status_response["EndpointStatus"]
service_state = ENDPOINT_STATUS_TO_STATE[endpoint_status]
deployment_state = DeploymentState(
state=service_state,
info_json=json.dumps(endpoint_status_response, default=str),
)
deployment_state.timestamp.GetCurrentTime()
return DescribeDeploymentResponse(
state=deployment_state, status=Status.OK()
)
except BentoMLException as error:
return DescribeDeploymentResponse(status=error.status_proto)
def describe(self, deployment_pb, repo=None):
deployment_spec = deployment_pb.spec
sagemaker_config = deployment_spec.sagemaker_operator_config
if sagemaker_config is None:
raise BentoMLDeploymentException(
'Sagemaker configuration is missing.')
sagemaker_client = boto3.client('sagemaker', sagemaker_config.region)
endpoint_name = generate_aws_compatible_string(
deployment_pb.namespace + '-' + deployment_spec.bento_name)
try:
endpoint_status_response = sagemaker_client.describe_endpoint(
EndpointName=endpoint_name)
except ClientError as e:
status = _parse_aws_client_exception_or_raise(e)
status.error_message = (
'Failed to describe SageMaker deployment: %s',
status.error_message,
)
return DescribeDeploymentResponse(status=status)
logger.debug("AWS describe endpoint response: %s",
endpoint_status_response)
endpoint_status = endpoint_status_response["EndpointStatus"]
service_state = ENDPOINT_STATUS_TO_STATE[endpoint_status]
deployment_state = DeploymentState(
state=service_state,
info_json=json.dumps(endpoint_status_response, default=str),
)
return DescribeDeploymentResponse(state=deployment_state,
status=Status.OK())
def _apply(self, deployment_pb, bento_pb, yatai_service, bento_path):
if loader._is_remote_path(bento_path):
with loader._resolve_remote_bundle_path(bento_path) as local_path:
return self._apply(deployment_pb, bento_pb, yatai_service,
local_path)
deployment_spec = deployment_pb.spec
aws_config = deployment_spec.aws_lambda_operator_config
bento_service_metadata = bento_pb.bento.bento_service_metadata
template = 'aws-python3'
if version.parse(bento_service_metadata.env.python_version
) < version.parse('3.0.0'):
template = 'aws-python'
api_names = ([aws_config.api_name] if aws_config.api_name else
[api.name for api in bento_service_metadata.apis])
ensure_deploy_api_name_exists_in_bento(
[api.name for api in bento_service_metadata.apis], api_names)
with TempDirectory() as serverless_project_dir:
init_serverless_project_dir(
serverless_project_dir,
bento_path,
deployment_pb.name,
deployment_spec.bento_name,
template,
)
generate_aws_lambda_handler_py(deployment_spec.bento_name,
api_names, serverless_project_dir)
generate_aws_lambda_serverless_config(
bento_service_metadata.env.python_version,
deployment_pb.name,
api_names,
serverless_project_dir,
aws_config.region,
# BentoML deployment namespace is mapping to serverless `stage`
# concept
stage=deployment_pb.namespace,
)
logger.info(
'Installing additional packages: serverless-python-requirements'
)
install_serverless_plugin("serverless-python-requirements",
serverless_project_dir)
logger.info('Deploying to AWS Lambda')
call_serverless_command(["deploy"], serverless_project_dir)
res_deployment_pb = Deployment(state=DeploymentState())
res_deployment_pb.CopyFrom(deployment_pb)
state = self.describe(res_deployment_pb, yatai_service).state
res_deployment_pb.state.CopyFrom(state)
return ApplyDeploymentResponse(status=Status.OK(),
deployment=res_deployment_pb)
def ApplyDeployment(self, request, context=None):
try:
# apply default namespace if not set
request.deployment.namespace = (
request.deployment.namespace or self.default_namespace
)
validation_errors = validate_deployment_pb_schema(request.deployment)
if validation_errors:
return ApplyDeploymentResponse(
status=Status.ABORTED(
'Failed to validate deployment. {errors}'.format(
errors=validation_errors
)
)
)
previous_deployment = self.deployment_store.get(
request.deployment.name, request.deployment.namespace
)
if previous_deployment:
# check deployment platform
if (
previous_deployment.spec.operator
!= request.deployment.spec.operator
):
return ApplyDeploymentResponse(
status=Status.ABORTED(
'Can not change the target deploy platform of existing '
'active deployment. Try delete existing deployment and '
'deploy to new target platform again'
)
)
request.deployment.state = DeploymentState(
state=DeploymentState.PENDING
)
self.deployment_store.insert_or_update(request.deployment)
# find deployment operator based on deployment spec
operator = get_deployment_operator(request.deployment)
# deploying to target platform
response = operator.apply(
request.deployment, self.repo, previous_deployment
)
# update deployment state
self.deployment_store.insert_or_update(response.deployment)
return response
except BentoMLException as e:
logger.error("INTERNAL ERROR: %s", e)
return ApplyDeploymentResponse(status=Status.INTERNAL(str(e)))
def describe(self, deployment_pb, yatai_service=None):
try:
deployment_spec = deployment_pb.spec
gcp_config = deployment_spec.gcp_function_operator_config
bento_pb = yatai_service.GetBento(
GetBentoRequest(
bento_name=deployment_spec.bento_name,
bento_version=deployment_spec.bento_version,
))
bento_service_metadata = bento_pb.bento.bento_service_metadata
api_names = ([gcp_config.api_name] if gcp_config.api_name else
[api.name for api in bento_service_metadata.apis])
with TempDirectory() as serverless_project_dir:
generate_gcp_function_serverless_config(
deployment_pb.name,
api_names,
serverless_project_dir,
gcp_config.region,
# BentoML namespace is mapping to serverless stage.
stage=deployment_pb.namespace,
)
try:
response = call_serverless_command(["info"],
serverless_project_dir)
info_json = parse_serverless_info_response_to_json_string(
response)
state = DeploymentState(state=DeploymentState.RUNNING,
info_json=info_json)
state.timestamp.GetCurrentTime()
except BentoMLException as e:
state = DeploymentState(state=DeploymentState.ERROR,
error_message=str(e))
state.timestamp.GetCurrentTime()
return DescribeDeploymentResponse(status=Status.OK(), state=state)
except BentoMLException as error:
return DescribeDeploymentResponse(
status=exception_to_return_status(error))
def apply(self, deployment_pb, yatai_service, prev_deployment=None):
try:
deployment_spec = deployment_pb.spec
gcp_config = deployment_spec.gcp_function_operator_config
bento_pb = yatai_service.GetBento(
GetBentoRequest(
bento_name=deployment_spec.bento_name,
bento_version=deployment_spec.bento_version,
))
if bento_pb.bento.uri.type != BentoUri.LOCAL:
raise BentoMLException(
'BentoML currently only support local repository')
else:
bento_path = bento_pb.bento.uri.uri
bento_service_metadata = bento_pb.bento.bento_service_metadata
api_names = ([gcp_config.api_name] if gcp_config.api_name else
[api.name for api in bento_service_metadata.apis])
ensure_deploy_api_name_exists_in_bento(
[api.name for api in bento_service_metadata.apis], api_names)
with TempDirectory() as serverless_project_dir:
init_serverless_project_dir(
serverless_project_dir,
bento_path,
deployment_pb.name,
deployment_spec.bento_name,
'google-python',
)
generate_gcp_function_main_py(deployment_spec.bento_name,
api_names,
serverless_project_dir)
generate_gcp_function_serverless_config(
deployment_pb.name,
api_names,
serverless_project_dir,
gcp_config.region,
# BentoML namespace is mapping to serverless stage.
stage=deployment_pb.namespace,
)
call_serverless_command(["deploy"], serverless_project_dir)
res_deployment_pb = Deployment(state=DeploymentState())
res_deployment_pb.CopyFrom(deployment_pb)
state = self.describe(res_deployment_pb, yatai_service).state
res_deployment_pb.state.CopyFrom(state)
return ApplyDeploymentResponse(status=Status.OK(),
deployment=res_deployment_pb)
except BentoMLException as error:
return ApplyDeploymentResponse(
status=exception_to_return_status(error))
def apply(self, deployment_pb, repo, prev_deployment=None):
ensure_docker_available_or_raise()
deployment_spec = deployment_pb.spec
aws_config = deployment_spec.aws_lambda_operator_config
bento_path = repo.get(deployment_spec.bento_name,
deployment_spec.bento_version)
bento_config = load_bentoml_config(bento_path)
template = 'aws-python3'
minimum_python_version = version.parse('3.0.0')
bento_python_version = version.parse(
bento_config['env']['python_version'])
if bento_python_version < minimum_python_version:
template = 'aws-python'
with TemporaryServerlessContent(
archive_path=bento_path,
deployment_name=deployment_pb.name,
bento_name=deployment_spec.bento_name,
template_type=template,
) as output_path:
generate_handler_py(deployment_spec.bento_name,
bento_config['apis'], output_path)
generate_serverless_configuration_for_aws_lambda(
service_name=deployment_pb.name,
apis=bento_config['apis'],
output_path=output_path,
region=aws_config.region,
stage=deployment_pb.namespace,
)
logger.info(
'Installing additional packages: serverless-python-requirements, '
'serverless-apigw-binary')
install_serverless_plugin("serverless-python-requirements",
output_path)
install_serverless_plugin("serverless-apigw-binary", output_path)
logger.info('Deploying to AWS Lambda')
call_serverless_command(["deploy"], output_path)
res_deployment_pb = Deployment(state=DeploymentState())
res_deployment_pb.CopyFrom(deployment_pb)
state = self.describe(res_deployment_pb, repo).state
res_deployment_pb.state.CopyFrom(state)
return ApplyDeploymentResponse(status=Status.OK(),
deployment=res_deployment_pb)
def apply(self, deployment_pb, repo, prev_deployment=None):
deployment_spec = deployment_pb.spec
sagemaker_config = deployment_spec.sagemaker_operator_config
if sagemaker_config is None:
raise BentoMLDeploymentException('Sagemaker configuration is missing.')
archive_path = repo.get(
deployment_spec.bento_name, deployment_spec.bento_version
)
# config = load_bentoml_config(bento_path)...
sagemaker_client = boto3.client('sagemaker', sagemaker_config.region)
with TemporarySageMakerContent(
archive_path, deployment_spec.bento_name, deployment_spec.bento_version
) as temp_path:
ecr_image_path = create_push_image_to_ecr(
deployment_spec.bento_name, deployment_spec.bento_version, temp_path
)
execution_role_arn = get_arn_role_from_current_user()
model_name = create_sagemaker_model_name(
deployment_spec.bento_name, deployment_spec.bento_version
)
sagemaker_model_info = {
"ModelName": model_name,
"PrimaryContainer": {
"ContainerHostname": model_name,
"Image": ecr_image_path,
"Environment": {
"API_NAME": sagemaker_config.api_name,
"BENTO_SERVER_TIMEOUT": config().get(
'apiserver', 'default_timeout'
),
"BENTO_SERVER_WORKERS": config().get(
'apiserver', 'default_gunicorn_workers_count'
),
},
},
"ExecutionRoleArn": execution_role_arn,
}
logger.info("Creating sagemaker model %s", model_name)
create_model_response = sagemaker_client.create_model(**sagemaker_model_info)
logger.debug("AWS create model response: %s", create_model_response)
production_variants = [
{
"VariantName": generate_aws_compatible_string(
deployment_spec.bento_name
),
"ModelName": model_name,
"InitialInstanceCount": sagemaker_config.instance_count,
"InstanceType": sagemaker_config.instance_type,
}
]
endpoint_config_name = create_sagemaker_endpoint_config_name(
deployment_spec.bento_name, deployment_spec.bento_version
)
logger.info(
"Creating Sagemaker endpoint %s configuration", endpoint_config_name
)
create_endpoint_config_response = sagemaker_client.create_endpoint_config(
EndpointConfigName=endpoint_config_name,
ProductionVariants=production_variants,
)
logger.debug(
"AWS create endpoint config response: %s", create_endpoint_config_response
)
endpoint_name = generate_aws_compatible_string(
deployment_pb.namespace + '-' + deployment_spec.bento_name
)
if prev_deployment:
logger.info("Updating sagemaker endpoint %s", endpoint_name)
update_endpoint_response = sagemaker_client.update_endpoint(
EndpointName=endpoint_name, EndpointConfigName=endpoint_config_name
)
logger.debug("AWS update endpoint response: %s", update_endpoint_response)
else:
logger.info("Creating sagemaker endpoint %s", endpoint_name)
create_endpoint_response = sagemaker_client.create_endpoint(
EndpointName=endpoint_name, EndpointConfigName=endpoint_config_name
)
logger.debug("AWS create endpoint response: %s", create_endpoint_response)
res_deployment_pb = Deployment(state=DeploymentState())
res_deployment_pb.CopyFrom(deployment_pb)
return ApplyDeploymentResponse(status=Status.OK(), deployment=res_deployment_pb)
def apply(self, deployment_pb, yatai_service, prev_deployment=None):
try:
ensure_docker_available_or_raise()
deployment_spec = deployment_pb.spec
sagemaker_config = deployment_spec.sagemaker_operator_config
if sagemaker_config is None:
raise BentoMLDeploymentException('Sagemaker configuration is missing.')
bento_pb = yatai_service.GetBento(
GetBentoRequest(
bento_name=deployment_spec.bento_name,
bento_version=deployment_spec.bento_version,
)
)
if bento_pb.bento.uri.type != BentoUri.LOCAL:
raise BentoMLException(
'BentoML currently only support local repository'
)
else:
bento_path = bento_pb.bento.uri.uri
ensure_deploy_api_name_exists_in_bento(
[api.name for api in bento_pb.bento.bento_service_metadata.apis],
[sagemaker_config.api_name],
)
sagemaker_client = boto3.client('sagemaker', sagemaker_config.region)
with TempDirectory() as temp_dir:
sagemaker_project_dir = os.path.jon(
temp_dir, deployment_spec.bento_name
)
init_sagemaker_project(sagemaker_project_dir, bento_path)
ecr_image_path = create_push_docker_image_to_ecr(
deployment_spec.bento_name,
deployment_spec.bento_version,
sagemaker_project_dir,
)
execution_role_arn = get_arn_role_from_current_aws_user()
model_name = create_sagemaker_model_name(
deployment_spec.bento_name, deployment_spec.bento_version
)
sagemaker_model_info = {
"ModelName": model_name,
"PrimaryContainer": {
"ContainerHostname": model_name,
"Image": ecr_image_path,
"Environment": {
"API_NAME": sagemaker_config.api_name,
"BENTO_SERVER_TIMEOUT": config().get(
'apiserver', 'default_timeout'
),
"BENTO_SERVER_WORKERS": config().get(
'apiserver', 'default_gunicorn_workers_count'
),
},
},
"ExecutionRoleArn": execution_role_arn,
}
logger.info("Creating sagemaker model %s", model_name)
try:
create_model_response = sagemaker_client.create_model(
**sagemaker_model_info
)
logger.debug("AWS create model response: %s", create_model_response)
except ClientError as e:
status = _parse_aws_client_exception_or_raise(e)
status.error_message = (
'Failed to create model for SageMaker Deployment: %s',
status.error_message,
)
return ApplyDeploymentResponse(status=status, deployment=deployment_pb)
production_variants = [
{
"VariantName": generate_aws_compatible_string(
deployment_spec.bento_name
),
"ModelName": model_name,
"InitialInstanceCount": sagemaker_config.instance_count,
"InstanceType": sagemaker_config.instance_type,
}
]
endpoint_config_name = create_sagemaker_endpoint_config_name(
deployment_spec.bento_name, deployment_spec.bento_version
)
logger.info(
"Creating Sagemaker endpoint %s configuration", endpoint_config_name
)
try:
create_config_response = sagemaker_client.create_endpoint_config(
EndpointConfigName=endpoint_config_name,
ProductionVariants=production_variants,
)
logger.debug(
"AWS create endpoint config response: %s", create_config_response
)
except ClientError as e:
# create endpoint failed, will remove previously created model
cleanup_model_error = _cleanup_sagemaker_model(
sagemaker_client,
deployment_spec.bento_name,
deployment_spec.bento_version,
)
if cleanup_model_error:
cleanup_model_error.error_message = (
'Failed to clean up model after unsuccessfully '
'create endpoint config: %s',
cleanup_model_error.error_message,
)
return ApplyDeploymentResponse(
status=cleanup_model_error, deployment=deployment_pb
)
status = _parse_aws_client_exception_or_raise(e)
status.error_message = (
'Failed to create endpoint config for SageMaker deployment: %s',
status.error_message,
)
return ApplyDeploymentResponse(status=status, deployment=deployment_pb)
endpoint_name = generate_aws_compatible_string(
deployment_pb.namespace + '-' + deployment_spec.bento_name
)
try:
if prev_deployment:
logger.debug("Updating sagemaker endpoint %s", endpoint_name)
update_endpoint_response = sagemaker_client.update_endpoint(
EndpointName=endpoint_name,
EndpointConfigName=endpoint_config_name,
)
logger.debug(
"AWS update endpoint response: %s", update_endpoint_response
)
else:
logger.debug("Creating sagemaker endpoint %s", endpoint_name)
create_endpoint_response = sagemaker_client.create_endpoint(
EndpointName=endpoint_name,
EndpointConfigName=endpoint_config_name,
)
logger.debug(
"AWS create endpoint response: %s", create_endpoint_response
)
except ClientError as e:
# create/update endpoint failed, will remove previously created config
# and then remove the model
cleanup_endpoint_config_error = _cleanup_sagemaker_endpoint_config(
client=sagemaker_client,
name=deployment_spec.bento_name,
version=deployment_spec.bento_version,
)
if cleanup_endpoint_config_error:
cleanup_endpoint_config_error.error_message = (
'Failed to clean up endpoint config after unsuccessfully '
'apply SageMaker deployment: %s',
cleanup_endpoint_config_error.error_message,
)
return ApplyDeploymentResponse(
status=cleanup_endpoint_config_error, deployment=deployment_pb
)
cleanup_model_error = _cleanup_sagemaker_model(
client=sagemaker_client,
name=deployment_spec.bento_name,
version=deployment_spec.bento_version,
)
if cleanup_model_error:
cleanup_model_error.error_message = (
'Failed to clean up model after unsuccessfully '
'apply SageMaker deployment: %s',
cleanup_model_error.error_message,
)
return ApplyDeploymentResponse(
status=cleanup_model_error, deployment=deployment_pb
)
status = _parse_aws_client_exception_or_raise(e)
status.error_message = (
'Failed to apply SageMaker deployment: %s',
status.error_message,
)
return ApplyDeploymentResponse(status=status, deployment=deployment_pb)
res_deployment_pb = Deployment(state=DeploymentState())
res_deployment_pb.CopyFrom(deployment_pb)
return ApplyDeploymentResponse(
status=Status.OK(), deployment=res_deployment_pb
)
except BentoMLException as error:
return ApplyDeploymentResponse(status=exception_to_return_status(error))
def describe(self, deployment_pb):
try:
deployment_spec = deployment_pb.spec
lambda_deployment_config = deployment_spec.aws_lambda_operator_config
lambda_deployment_config.region = (lambda_deployment_config.region
or get_default_aws_region())
if not lambda_deployment_config.region:
raise InvalidArgument('AWS region is missing')
bento_pb = self.yatai_service.GetBento(
GetBentoRequest(
bento_name=deployment_spec.bento_name,
bento_version=deployment_spec.bento_version,
))
bento_service_metadata = bento_pb.bento.bento_service_metadata
api_names = ([lambda_deployment_config.api_name]
if lambda_deployment_config.api_name else
[api.name for api in bento_service_metadata.apis])
try:
cf_client = boto3.client('cloudformation',
lambda_deployment_config.region)
cloud_formation_stack_result = cf_client.describe_stacks(
StackName='{ns}-{name}'.format(ns=deployment_pb.namespace,
name=deployment_pb.name))
stack_result = cloud_formation_stack_result.get('Stacks')[0]
# https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/\
# using-cfn-describing-stacks.html
success_status = ['CREATE_COMPLETE', 'UPDATE_COMPLETE']
if stack_result['StackStatus'] in success_status:
if stack_result.get('Outputs'):
outputs = stack_result['Outputs']
else:
return DescribeDeploymentResponse(
status=Status.ABORTED(
'"Outputs" field is not present'),
state=DeploymentState(
state=DeploymentState.ERROR,
error_message='"Outputs" field is not present',
),
)
elif stack_result[
'StackStatus'] in FAILED_CLOUDFORMATION_STACK_STATUS:
state = DeploymentState(state=DeploymentState.FAILED)
state.timestamp.GetCurrentTime()
return DescribeDeploymentResponse(status=Status.OK(),
state=state)
else:
state = DeploymentState(state=DeploymentState.PENDING)
state.timestamp.GetCurrentTime()
return DescribeDeploymentResponse(status=Status.OK(),
state=state)
except Exception as error: # pylint: disable=broad-except
state = DeploymentState(state=DeploymentState.ERROR,
error_message=str(error))
state.timestamp.GetCurrentTime()
return DescribeDeploymentResponse(status=Status.INTERNAL(
str(error)),
state=state)
outputs = {o['OutputKey']: o['OutputValue'] for o in outputs}
info_json = {}
if 'EndpointUrl' in outputs:
info_json['endpoints'] = [
outputs['EndpointUrl'] + '/' + api_name
for api_name in api_names
]
if 'S3Bucket' in outputs:
info_json['s3_bucket'] = outputs['S3Bucket']
state = DeploymentState(state=DeploymentState.RUNNING,
info_json=json.dumps(info_json))
state.timestamp.GetCurrentTime()
return DescribeDeploymentResponse(status=Status.OK(), state=state)
except BentoMLException as error:
return DescribeDeploymentResponse(status=error.status_proto)
def describe(self, deployment_pb, repo):
# fetch deployment state
deployment_state = DeploymentState()
# deployment_state.state = ...
return deployment_state
def _apply(
self, deployment_pb, bento_pb, yatai_service, bento_path, prev_deployment=None
):
if loader._is_remote_path(bento_path):
with loader._resolve_remote_bundle_path(bento_path) as local_path:
return self._apply(
deployment_pb, bento_pb, yatai_service, local_path, prev_deployment
)
deployment_spec = deployment_pb.spec
sagemaker_config = deployment_spec.sagemaker_operator_config
ensure_deploy_api_name_exists_in_bento(
[api.name for api in bento_pb.bento.bento_service_metadata.apis],
[sagemaker_config.api_name],
)
sagemaker_client = boto3.client('sagemaker', sagemaker_config.region)
with TempDirectory() as temp_dir:
sagemaker_project_dir = os.path.join(temp_dir, deployment_spec.bento_name)
init_sagemaker_project(sagemaker_project_dir, bento_path)
ecr_image_path = create_push_docker_image_to_ecr(
sagemaker_config.region,
deployment_spec.bento_name,
deployment_spec.bento_version,
sagemaker_project_dir,
)
try:
model_name = _create_sagemaker_model(
sagemaker_client,
deployment_spec.bento_name,
deployment_spec.bento_version,
ecr_image_path,
sagemaker_config.api_name,
)
except ClientError as e:
status = _parse_aws_client_exception(e)
status.error_message = (
'Failed to create model for SageMaker'
' Deployment: {}'.format(status.error_message)
)
return ApplyDeploymentResponse(status=status, deployment=deployment_pb)
try:
endpoint_config_name = _create_sagemaker_endpoint_config(
sagemaker_client,
model_name,
deployment_spec.bento_name,
deployment_spec.bento_version,
sagemaker_config,
)
except ClientError as e:
# create endpoint failed, will remove previously created model
cleanup_model_error = _cleanup_sagemaker_model(
sagemaker_client,
deployment_spec.bento_name,
deployment_spec.bento_version,
)
if cleanup_model_error:
cleanup_model_error.error_message = (
'Failed to clean up model after unsuccessfully '
'create endpoint config: %s',
cleanup_model_error.error_message,
)
return ApplyDeploymentResponse(
status=cleanup_model_error, deployment=deployment_pb
)
status = _parse_aws_client_exception(e)
status.error_message = (
'Failed to create endpoint config for SageMaker deployment: %s',
status.error_message,
)
return ApplyDeploymentResponse(status=status, deployment=deployment_pb)
endpoint_name = generate_aws_compatible_string(
deployment_pb.namespace + '-' + deployment_spec.bento_name
)
try:
if prev_deployment:
logger.debug("Updating sagemaker endpoint %s", endpoint_name)
update_endpoint_response = sagemaker_client.update_endpoint(
EndpointName=endpoint_name, EndpointConfigName=endpoint_config_name
)
logger.debug(
"AWS update endpoint response: %s", update_endpoint_response
)
else:
logger.debug("Creating sagemaker endpoint %s", endpoint_name)
create_endpoint_response = sagemaker_client.create_endpoint(
EndpointName=endpoint_name, EndpointConfigName=endpoint_config_name
)
logger.debug(
"AWS create endpoint response: %s", create_endpoint_response
)
except ClientError as e:
# create/update endpoint failed, will remove previously created config
# and then remove the model
cleanup_endpoint_config_error = _cleanup_sagemaker_endpoint_config(
client=sagemaker_client,
name=deployment_spec.bento_name,
version=deployment_spec.bento_version,
)
if cleanup_endpoint_config_error:
cleanup_endpoint_config_error.error_message = (
'Failed to clean up endpoint config after unsuccessfully '
'apply SageMaker deployment: %s',
cleanup_endpoint_config_error.error_message,
)
return ApplyDeploymentResponse(
status=cleanup_endpoint_config_error, deployment=deployment_pb
)
cleanup_model_error = _cleanup_sagemaker_model(
client=sagemaker_client,
name=deployment_spec.bento_name,
version=deployment_spec.bento_version,
)
if cleanup_model_error:
cleanup_model_error.error_message = (
'Failed to clean up model after unsuccessfully apply '
'SageMaker deployment: {}'.format(cleanup_model_error.error_message)
)
return ApplyDeploymentResponse(
status=cleanup_model_error, deployment=deployment_pb
)
status = _parse_aws_client_exception(e)
status.error_message = 'Failed to apply SageMaker ' 'deployment: {}'.format(
status.error_message
)
return ApplyDeploymentResponse(status=status, deployment=deployment_pb)
res_deployment_pb = Deployment(state=DeploymentState())
res_deployment_pb.CopyFrom(deployment_pb)
return ApplyDeploymentResponse(status=Status.OK(), deployment=res_deployment_pb)
def apply(self, deployment_pb, yatai_service, prev_deployment=None):
try:
ensure_docker_available_or_raise()
deployment_spec = deployment_pb.spec
aws_config = deployment_spec.aws_lambda_operator_config
bento_pb = yatai_service.GetBento(
GetBentoRequest(
bento_name=deployment_spec.bento_name,
bento_version=deployment_spec.bento_version,
)
)
if bento_pb.bento.uri.type != BentoUri.LOCAL:
raise BentoMLException(
'BentoML currently only support local repository'
)
else:
bento_path = bento_pb.bento.uri.uri
bento_service_metadata = bento_pb.bento.bento_service_metadata
template = 'aws-python3'
if version.parse(bento_service_metadata.env.python_version) < version.parse(
'3.0.0'
):
template = 'aws-python'
api_names = (
[aws_config.api_name]
if aws_config.api_name
else [api.name for api in bento_service_metadata.apis]
)
ensure_deploy_api_name_exists_in_bento(
[api.name for api in bento_service_metadata.apis], api_names
)
with TempDirectory() as serverless_project_dir:
init_serverless_project_dir(
serverless_project_dir,
bento_path,
deployment_pb.name,
deployment_spec.bento_name,
template,
)
generate_aws_lambda_handler_py(
deployment_spec.bento_name, api_names, serverless_project_dir
)
generate_aws_lambda_serverless_config(
bento_service_metadata.env.python_version,
deployment_pb.name,
api_names,
serverless_project_dir,
aws_config.region,
# BentoML deployment namespace is mapping to serverless `stage`
# concept
stage=deployment_pb.namespace,
)
logger.info(
'Installing additional packages: serverless-python-requirements'
)
install_serverless_plugin(
"serverless-python-requirements", serverless_project_dir
)
logger.info('Deploying to AWS Lambda')
call_serverless_command(["deploy"], serverless_project_dir)
res_deployment_pb = Deployment(state=DeploymentState())
res_deployment_pb.CopyFrom(deployment_pb)
state = self.describe(res_deployment_pb, yatai_service).state
res_deployment_pb.state.CopyFrom(state)
return ApplyDeploymentResponse(
status=Status.OK(), deployment=res_deployment_pb
)
except BentoMLException as error:
return ApplyDeploymentResponse(status=exception_to_return_status(error))
