lakehouse-retail-c360 Job fails at register_churn_model run

[jasonscalia]

Issue:

lakehouse-retail-c360 - register_churn_model run job fails with:

Invalid experiment format or no experiment avvailable. Please re-run the previous notebook.

Log Excerpt:

---------------------------------------------------------------------------
Exception                                 Traceback (most recent call last)
File <command-69050624446694>, line 3
      1 import mlflow
      2 #Added for the demo purpose
----> 3 xp = get_last_experiment("lakehouse-retail-c360")
      4 # Use MLflow to track experiments
      5 mlflow.set_experiment(xp["path"])

File <command-69050624446796>, line 27, in get_last_experiment(demo_name, experiment_path)
     25         raise Exception(f"It looks like the last experiment {last_xp} is too old ({days} days). Please re-run the previous notebook to make sure you have the latest version.")
     26 else:
---> 27     raise Exception("Invalid experiment format or no experiment avvailable. Please re-run the previous notebook.")
     28 return last_xp

Exception: Invalid experiment format or no experiment avvailable. Please re-run the previous notebook.

Re-running prior job, destroying and recreating demo including compute does not resolve the issue.

Steps to recreate:

1. %pip install dbdemos
2. dbutils.library.restartPython()

3. import dbdemos
   dbdemos.install('lakehouse-retail-c360')

4. register_churn_model job fails

[QuentinAmbard]

alright I found the issue, I think it's specific to your workspace where you have folders conflicting. You can cleanup your /Shared/dbdemos/experiment/lakehouse-retail-c360 and rerun it, it'll fix the issue. New version will handle this properly.