Azure / MachineLearningNotebooks

Python notebooks with ML and deep learning examples with Azure Machine Learning Python SDK | Microsoft
https://docs.microsoft.com/azure/machine-learning/service/
MIT License
4.06k stars 2.51k forks source link

Error Running Sample Notebook - Azure Machine Learning Pipeline with NotebookRunnerStep #1781

Open Tee-A opened 2 years ago

Tee-A commented 2 years ago

Currently Trying to execute this Sample Notebook.

I get the following error when Creating Notebook run configuration and setting parameter values as shown below

`handler = AzureMLNotebookHandler(timeout=600, progress_bar=False, log_output=True)

cfg = NotebookRunConfig(source_directory=source_directory, notebook="training_notebook.ipynb", handler = handler, parameters={"arg1": "Machine Learning"}, run_config=conda_run_config)

print("Notebook Run Config is created.")`

Screen Shot 2022-07-14 at 20 26 36

l

Screen Shot 2022-07-14 at 20 27 20
lindsayevanslee commented 11 months ago

I am also having the same issue. Disappointing to see no response on this in all the time since the issue was opened.

Attempting to run this script on compute instance of size Standard_DS11_v2, software version Linux 23.06.30, using kernel "Python 3.8 - AzureML", SDK version 1.51.0

Trying to dig into the error messages, it looks like the error is stemming from the handler AzureMLNotebookHandler(), which has this __init__ method:

    def __init__(self, history=True, timeout=None, **kwargs):
        """Construct PapermillExecutionHandler object."""
        super().__init__(os.path.join(os.path.dirname(__file__),
                                      "_handlers",
                                      "papermill_notebook_run_handler.py"))

        for key, val in kwargs.items():
            if key not in self.supported_options:
                raise Exception("Unsupported option: {}".format(key))

        kwargs["engine_name"] = "azureml_engine"

        self.papermill_args = kwargs
        self.execution_args = {"history": history, "timeout": timeout}
        self.dependencies.append("azureml-contrib-notebook[full]")

It's trying to look for a file papermill_notebook_run_handler.py, but it doesn't exist. What is the expected behavior here and how can we fix this?

lindsayevanslee commented 11 months ago

With the help of Microsoft support, got this sample to run by cloning the entire repository that this sample script is in: https://github.com/Azure/MachineLearningNotebooks/tree/master/how-to-use-azureml/machine-learning-pipelines/intro-to-pipelines

This papermill_notebook_run_handler.py script that the notebook is looking for is in the notebook_runner folder at the link above.