当 MLFlow 尝试访问存储在 S3 上的人工制品时，获取“（InvalidToken）调用 ListObjectsV2 操作时”

Question

I am trying to start the MLFlow server on my local machine inside a python virtual environment using the following command:

mlflow server --backend-store-uri postgresql://mlflow_user:mlflow@localhost/mlflow  --artifacts-destination S3://<S3 bucket name>/mlflow/ --serve-artifacts  -h 0.0.0.0 -p 8000

I have exported the following environment variables inside the activated python venv :

export AWS_ACCESS_KEY_ID=<access key>
export AWS_SECRET_ACCESS_KEY=<secret key>
export DEFAULT_REGION_NAME=<region name>
export DEFAULT_OUTPUT_FORMAT=<output format>

MLFlow gives the following error while accessing the model artefacts for all the runs:

botocore.exceptions.ClientError: An error occurred (InvalidToken) when calling the ListObjectsV2 operation: The provided token is malformed or otherwise invalid.

Any workaround for this?

Answer 1

According to the Environment Variables section in the Boto3 official documentation, setting AWS_SESSION_TOKEN was also necessary.

With reference to this answer added by Stack Overflow user @Francis Lewis, I just reset the AWS access token inside the activated python environment as follows:

export AWS_SESSION_TOKEN=""

And the problem was solved. Thanks to @Francis Lewis!