LogisticRegression pipeline does not go through

[thiamhuat2026]

[tracking_store.py:77] Experiment names must be less than 257 characters, begin with a letter or number, and may only include dashes (-) or underscores (_). 2026-05-09:09:48:38,114 ERROR [synapse_mlflow_utils.py:390] [fabric mlflow plugin]: <class 'synapse.ml.mlflow.tracking_store.TridentMLflowTrackingStore'>.check_experiment_name_valid exception Experiment names must be less than 257 characters, begin with a letter or number, and may only include dashes (-) or underscores (_). 2026-05-09:09:48:38,134 ERROR [synapse_mlflow_utils.py:390] [fabric mlflow plugin]: <class 'synapse.ml.mlflow.tracking_store.TridentMLflowTrackingStore'>.get_experiment_by_name exception Experiment names must be less than 257 characters, begin with a letter or number, and may only include dashes (-) or underscores (_). 2026-05-09:09:48:38,136 ERROR [default_experiment_registry.py:55] Error find or created notebook/SJD default experiment, please call mlflow.set_experiment to set a valid one before run Traceback (most recent call last): File "/home/trusted-service-user/cluster-env/trident_env/lib/python3.11/site-packages/synapse/ml/mlflow/default_experiment_registry.py", line 52, in get_experiment_id return mlflow.set_experiment(f"{artifact_name}").experiment_id ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

[v-anbandari · Accepted solution]

Hi @thiamhuat2026,
Thank you for reaching out to the Microsoft community  Forum.

The issue is not caused by the LogisticRegression pipeline itself. Based on the error message , the failure occurs because the notebook environment automatically tries to create an MLflow experiment during "pipeline.fit(train_df)", but the generated experiment name does not satisfy MLflow experiment naming rules. Because of this, the pipeline execution fails before the model training starts.

To resolve the issue, manually set a valid MLflow experiment name before training the pipeline:

import mlflow
mlflow.set_experiment("logistic_regression_experiment")

Then run your existing pipeline code again.

For More Details :
Machine learning experiment - Microsoft Fabric | Microsoft Learn.

I hope this helps. Please feel free to reach out for further queries.
Thank you.