Proxy server for artifacts just doesn't work

Based on documentation https://www.mlflow.org/docs/latest/tracking.html#scenario-5-mlflow-tracking-server-enabled-with-proxied-artifact-storage-access there is a scenario when tracking server should act as proxy artifact:

I am running the server using this command:

mlflow server --backend-store-uri postgresql+psycopg2://myuser:mypassword@db:5432/mlflow_db --artifacts-destination s3://mybucker/mlflow/prod --serve-artifacts --host 0.0.0.0

But unfortunately, I can clearly see that this code on the client side

mlflow.set_tracking_uri("http://my_server_on_amazon")
mlflow.set_experiment("test")
...
log_artifacts("outputs")

tries to access S3 directly (instead of sending all requests to the http://my_server_on_amazon, as I expect it to do).

May I have a question: is documentation is bad and this feature of artifact proxying is still not implemented? Or I am doing something wrong with “mlflow server … (parameters)” command?

MLflow version: 1.24.0. Thanks in advance.

[njanopoulos]

I am having similar issue to what was posted here at the beginning of the conversation. Unfortunately, the solution provided hasn’t worked for me.

When running the same script on the Remote host, the artifacts get stored in S3 properly. When I run from localhost, I get:

botocore.exceptions.NoCredentialsError: Unable to locate credentials

Anybody have more suggestions of things to try or look for?

[ibobak]

Feel free to close the issue, if needed. Or change status to “problem in documentation”, whatever.

Thanks again!