Skip to content

Estimator .fit() fails in Jakarta region with built-in algo, due to added "USE_SMDEBUG" environment variable when debugger is not supported #3454

@yudho

Description

@yudho

Describe the bug
When run from Jakarta region which does not support SageMaker Debugger yet (refer to https://docs.aws.amazon.com/sagemaker/latest/dg/debugger-supported-frameworks.html), the Estimator .fit() adds "Environment" variable with value "USE_SMDEBUG": "0". When used with built-in algo (e.g. XGBoost), this will produce error "Environment variables are not allowed when using SageMaker built-in algorithm" and blocks the training.

To reproduce

  1. Set sagemaker session (boto session) to use Jakarta region
  2. training_image = sagemaker.image_uris.retrieve("xgboost", region=region, version="1.2-1")
  3. Create an estimator and run .fit() method

Expected behavior
Training will fail to create with error "Environment variables are not allowed when using SageMaker built-in algorithm"

Screenshots or logs
Do not have

System information
A description of your system. Please provide:

  • SageMaker Python SDK version: AWS-SageMaker-Python-SDK/2.105.0 Boto3/1.24.58 Linux/4.14.291-218.527.amzn2.x86_64 Botocore/1.27.58
  • Framework name (eg. PyTorch) or algorithm (eg. KMeans): XGBoost
  • Framework version: 1.2-1
  • Python version: Python/3.8.13
  • CPU or GPU: CPU
  • Custom Docker image (Y/N): N

Additional context
Do not have

Metadata

Metadata

Assignees

Labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions