there is a ssl validation fails to access s3 in python sdk without proxy

[jennylover]

this is a sample code,

    self.est = TensorFlow(
        entry_point = 'train.py',
        source_dir='code',
        role=sagemaker.role,
        framework_version='2.3.1',
        model_dir=False,
        py_version='py37',
        instance_type=instance_type,
        instance_count=1,
        output_path=s3_output_path_uri,
        hyperparameters={
            'batch-size':512,
            'epochs':10,
            'learning-rate': 1e-3,
            'beta_1':0.9,
            'beta_2':0.999
        },
    )

    self.est.fit(inputs=channels)

and when python sdk calls TensorFlow estimator, it returns botocore.exceptions.SSLError: SSL validation failed for https://s3-location [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: unable to get local issuer certificate (_ssl.c:1129)

boto3 can work with proxy configuration, but could not find how to set proxy with sagemaker python sdk.

    self.client = boto3.client('machinelearning',
                        AWS_REGION,
                        aws_access_key_id=AWS_ACCESS_KEY_ID,
                        aws_secret_access_key=AWS_SECRET_ACCESS_KEY,
                        config=Config(proxies={'https':'XX.XX.XX.XX:8080'},signature_version='s3v4'),
                        verify='cert.crt',
                        )

any chance to set proxy configuration in sagemaker python sdk?

Thanks,

[ntw-au]

I have tried to set a custom certificate bundle several ways but so far the SageMaker Python SDK does not respect the configuration. I've raised this as a feature request: #5108