Deploy method cannot find update_endpoint keyword argument 'update_endpoint'

[diesazul96]

Please fill out the form below.

System Information
Framework (e.g. TensorFlow) / Algorithm (e.g. KMeans):
I'm using from sklearn with IsolationForest

Framework Version:

Python Version: Python 3.8

CPU or GPU: CPU

Python SDK Version:

Are you using a custom image: Yes. I started from your example scikit_bring_your_own and modified it from there.

Describe the problem
I want to update an endpoint that already exists. According to your docs, this should be easy. https://sagemaker.readthedocs.io/en/stable/overview.html
Where the example given is

Deploys the model that was generated by fit() to an existing SageMaker endpoint
mxnet_predictor = mxnet_estimator.deploy(initial_instance_count=1,
instance_type='ml.p2.xlarge',
update_endpoint=True,
endpoint_name='existing-endpoint')

However, I when I try the deploy method it does not recognize update_endpoint (where sklearn_estimator is create with from sagemaker.sklearn.estimator import SKLearn ):
predictor = sklearn_estimator.deploy(instance_type="ml.t2.medium", initial_instance_count=1,endpoint_name=endpoint_name,update_endpoint=True)

Minimal repro / logs
TypeError Traceback (most recent call last)
in ()
----> 3 predictor = sklearn_estimator.deploy(instance_type="ml.t2.medium", initial_instance_count=1,endpoint_name=endpoint_name,update_endpoint=True)

~/anaconda3/envs/python3/lib/python3.6/site-packages/sagemaker/estimator.py in deploy(self, initial_instance_count, instance_type, accelerator_type, endpoint_name, use_compiled_model, **kwargs)
362 model = self._compiled_models[family]
363 else:
--> 364 model = self.create_model(**kwargs)
365 return model.deploy(
366 instance_type=instance_type,

~/anaconda3/envs/python3/lib/python3.6/site-packages/sagemaker/estimator.py in create_model(self, role, image, predictor_cls, serializer, deserializer, content_type, accept, vpc_config_override, **kwargs)
706 return Model(self.model_data, image or self.train_image(), role,
707 vpc_config=self.get_vpc_config(vpc_config_override),
--> 708 sagemaker_session=self.sagemaker_session, predictor_cls=predictor_cls, **kwargs)
709
710 @classmethod

TypeError: init() got an unexpected keyword argument 'update_endpoint'

Exact command to reproduce:
----> 3 sklearn_estimator.deploy(instance_type="ml.t2.medium", initial_instance_count=1,endpoint_name=endpoint_name,update_endpoint=True)

[danielcavalli]

I've got the sameproblem while using sagemaker 2.68.0

WARNING:sagemaker.deprecations:update_endpoint is a no-op in sagemaker>=2.
See: https://sagemaker.readthedocs.io/en/stable/v2.html for details.

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-120-5b86bd9b2ad7> in <module>
----> 1 predictor = estimator.deploy(initial_instance_count=1, instance_type=instance_type, update_endpoint=True, endpoint_name=endpoint_name)

/opt/conda/lib/python3.7/site-packages/sagemaker/estimator.py in deploy(self, initial_instance_count, instance_type, serializer, deserializer, accelerator_type, endpoint_name, use_compiled_model, wait, model_name, kms_key, data_capture_config, tags, **kwargs)
    945         else:
    946             kwargs["model_kms_key"] = self.output_kms_key
--> 947             model = self.create_model(**kwargs)
    948 
    949         model.name = model_name

/opt/conda/lib/python3.7/site-packages/sagemaker/sklearn/estimator.py in create_model(self, model_server_workers, role, vpc_config_override, entry_point, source_dir, dependencies, **kwargs)
    214             vpc_config=self.get_vpc_config(vpc_config_override),
    215             dependencies=(dependencies or self.dependencies),
--> 216             **kwargs
    217         )
    218 

/opt/conda/lib/python3.7/site-packages/sagemaker/sklearn/model.py in __init__(self, model_data, role, entry_point, framework_version, py_version, image_uri, predictor_cls, model_server_workers, **kwargs)
    133         self.py_version = py_version
    134 
--> 135         super(SKLearnModel, self).__init__(
    136             model_data, image_uri, role, entry_point, predictor_cls=predictor_cls, **kwargs
    137         )

TypeError: super(type, obj): obj must be an instance or subtype of type

As the warning says, this feature has been deprecated on SM >2.0 but the docs haven't been updated to let us know what is the solution to that problem.
Of course, I can use boto client to do that but it seems a little unintuitive.

[calunia]

Hello Everyone,
This is may be too late. update_endpoint argument is deprecated and not use in somes of sagemaker model class (tensorflow in my case).
But, you can use the pipelineModel where update_endpoint is not deprecated at all.

Here is an exemple :

import argparse
import boto3
import logging
import os
from botocore.exceptions import ClientError
import tarfile
import zipfile
from sagemaker import ModelPackage
from sagemaker.tensorflow.model import TensorFlowModel
from sagemaker.pipeline import PipelineModel
import sagemaker
from botocore.exceptions import ClientError
import sagemaker.session
import time
from sagemaker.serializers import CSVSerializer, IdentitySerializer
from sagemaker.deserializers import JSONDeserializer

sm_client = boto3.client("sagemaker")

region = boto3.Session().region_name
sagemaker_session = sagemaker.session.Session()
role = sagemaker.get_execution_role()

model = TensorFlowModel(
    role=role, 
    model_data="model.tar.gz",
    framework_version="2.3",
    sagemaker_session=sagemaker_session
)
pipelineModel = PipelineModel(models=[model],role=role,sagemaker_session=sagemaker_session)

endpoint_name = "your name"
 
pipelineModel.deploy(initial_instance_count=1, instance_type="ml.m5.xlarge", endpoint_name=endpoint_name, serializer=IdentitySerializer(content_type="application/json"),
        deserializer=JSONDeserializer(), update_endpoint=True)

I hope that'll help you.

Calunia

[pags]

Is there still no way to easily/rapidly update a previously-deployed model?