MLflowLoggerCallback fails with botocore.exceptions.EndpointConnectionError: Could not connect to the endpoint URL

Ray Libraries (Data, Train, Tune, Serve) Ray Tune
1

I’ve added Using MLflow with Tune — Ray v2.0.0.dev0 to my code, but it fails with the following exception:
botocore.exceptions.EndpointConnectionError: Could not connect to the endpoint URL: “https://mlp-mlflow.s3.amazonaws.com/dev/10/8a790464051a4776b4338c859e5136fd/artifacts/bestmodel.h5”

How to correctly provide S3 credentials?

2

Hey @diman82 it’s a bit hard for me to help out here. Could you provide a longer stacktrace?

3

@rliaw Eventually, I’ve fixed the issue by providing the relevant env variables, as following:

    os.environ["MLFLOW_S3_ENDPOINT_URL"] = 'https://sampledomain.net'
    os.environ["MLFLOW_S3_IGNORE_TLS"] = 'true'
    os.environ["MLFLOW_S3_BUCKET"] = 's3://bucketname/dev'
    os.environ["AWS_ACCESS_KEY_ID"] = 'access_key'
    os.environ["AWS_SECRET_ACCESS_KEY"] = 'secret_key'