[Python] MLFlow doesnt copy properly my artifacts to the mlruns folder when used in a container

I am logging a few models using MLflow. The corresponding scripts are executed in a docker container.

The command mlflow.log_artifacts doesn't seem to work properly since I can t see the corresponding files under the mlruns folder. The command doesn't return any error though.

EDIT1: after some further investigations, it seems that the problem arises whenever I mount the folder containing mlruns/ as a docker volume. I did some tests with the example provided by the docstring of log_artifacts

import os
import json
import mlflow
# Create some files to preserve as artifacts
features = "rooms, zipcode, median_price, school_rating, transport"
data = {"state": "TX", "Available": 25, "Type": "Detached"}
# Create couple of artifact files under the directory "data"
os.makedirs("data", exist_ok=True)
with open("data/data.json", 'w', encoding='utf-8') as f:
json.dump(data, f, indent=2)
with open("data/features.txt", 'w') as f:
f.write(features)
# Write all files in "data" to root artifact_uri/states
with mlflow.start_run():
mlflow.log_artifacts("data", artifact_path="states")


If I run this in a container without volume mounting, it runs just fine, ie the artifacts appear under mlruns/<exp-id>/<run-id>/artifacts/state

However If I run this in a container with the folder containing mlruns/ mounted, it doesn't work, ie the folder mlruns/<exp-id>/<run-id>/artifacts/ is empty.

Continue reading...