Sagemaker Asynchronous Endpoint Configuration

Question

We deployed a LighGBM Regression model and endpoint using Sagemaker Jumpstart.
We have attempted to configure this endpoint as 'asynchronous' via the console.
Receiving Error: ValidationException-Network Isolation is not supported when specifying an AsyncInferenceConfig.

Looking at the model's network details the model has Enable Network Isolation set as 'True'.
This was default output setting set by JumpStart.

How can we diasble Network Isolation to in order to make this endpoint asynchronous?

Accepted Answer

Vanilla SageMaker "Models" (as opposed to versioned ModelPackages) are immutable in the API with no "UpdateModel" action... But I think you should be able to create a new Model copying the settings of the current one.

I'd suggest to:

1. Use [DescribeModel](https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_DescribeModel.html) (via [boto3.client("sagemaker").describe_model()](https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/sagemaker.html#SageMaker.Client.describe_model), assuming you're using Python) to fetch all the parameters of the existing JumpStart model such as the S3 artifact location and other settings
2. Use [CreateModel](https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_CreateModel.html) ([create_model()](https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/sagemaker.html#SageMaker.Client.create_model)) to create a new model with same configuration but network isolation disabled
3. Use your new model to try and deploy an async endpoint

Probably you'd find the low-level boto3 SDK more intuitive for this task than the high-level `sagemaker` SDK's [Model class](https://sagemaker.readthedocs.io/en/stable/api/inference/model.html) - because the latter does some magic that makes typical build/train/deploy workflows easier but can be less natural for hacking around with existing model definitions. For example, creating an SMSDK `Model` object doesn't actually create a Model in the SageMaker API, because deployment instance type affects choice of container image so that gets deferred until a `.deploy()` call or similar later.

Sagemaker Asynchronous Endpoint Configuration

Contenus pertinents