Datastore workspaceblobstore access failed, ErrorCode: ResourceNotFound using AutoML.

Question

Datastore workspaceblobstore access failed, ErrorCode: ResourceNotFound using AutoML.

Caesarus 21

Why remote run is getting "Datastore workspaceblobstore access failed, ErrorCode: ResourceNotFound" when using automl?

I've attached datastore to workspace and I'm able to create Tabular dataset using blob URL.

But it crashes when submittting experiment with auto ml.

Ramr-msft 17,826 Reputation points

2021-10-14T08:30:05.743+00:00

@Caesarus Thanks for the question. Can you please add more details about the region that you are trying. Is the workspaceblobstore behind vnet?

This doc will be helpful: https://learn.microsoft.com/en-us/azure/machine-learning/how-to-enable-studio-virtual-network
Caesarus 21 Reputation points

2021-10-14T09:58:47.263+00:00

Hello @Ramr-msft ,

Thank you for your response.
I'm currently trying on westeurope and workspaceblobstore is not behind vnet. I'm using remote run for my experiment.
I tried to change the way to point on data from TabularDatasetFactory to Dataset.Tabular and now I'm getting error: UnboundLocalError: local variable 'resp' referenced before assignment.

Maybe the compute cluster is not updated or is it possible to down grade azureml?

Regards.
Ramr-msft 17,826 Reputation points

2021-10-14T13:11:42.287+00:00

@Caesarus Thanks for the details. Can you please share the code that you are trying.

Here is the link to get the data: https://learn.microsoft.com/en-us/azure/machine-learning/how-to-access-data#get-datastores-from-your-workspace

Caesarus 21

Yes sure:

from azureml.core.workspace import Workspace
from azureml.core.experiment import Experiment
from azureml.core.compute import ComputeTarget
from azureml.core import Datastore

ws = Workspace.from_config()
experiment = Experiment(ws, "myexperiment")
compute_target = ComputeTarget(workspace = ws, name="cluster-name")


datastore = Datastore.get(ws, datastore_name="workspacetsstore")
data = Dataset.Tabular.from_delimited_files((datastore, "myfile.csv"))

from azureml.automl.core.forecasting_parameters import ForecastingParameters
from azureml.train.automl import AutoMLConfig

forecasting_parameters = ForecastingParameters(
    time_column_name="Date",
    forecast_horizon=fcst_length,
    country_or_region_for_holidays="FR",
    use_stl = True,
    seasonality='auto',
    freq = "H"
)
config = AutoMLConfig(task = 'forecasting', 
                        n_cross_validations=5,
                        primary_metric="normalized_mean_absolute_error",
                        forecasting_parameters=forecasting_parameters,
                        enable_dnn=False,
                        compute_target=compute_target,
                        training_data = data,
                        label_column_name="my_label_column _name"
                        )

experiment.submit(config, show_output = True)

Your answer

Ramr-msft 17,826 Reputation points

2021-10-14T08:30:05.743+00:00

@Caesarus Thanks for the question. Can you please add more details about the region that you are trying. Is the workspaceblobstore behind vnet?

This doc will be helpful: https://learn.microsoft.com/en-us/azure/machine-learning/how-to-enable-studio-virtual-network
Caesarus 21 Reputation points

2021-10-14T09:58:47.263+00:00

Hello @Ramr-msft ,

Thank you for your response.
I'm currently trying on westeurope and workspaceblobstore is not behind vnet. I'm using remote run for my experiment.
I tried to change the way to point on data from TabularDatasetFactory to Dataset.Tabular and now I'm getting error: UnboundLocalError: local variable 'resp' referenced before assignment.

Maybe the compute cluster is not updated or is it possible to down grade azureml?

Regards.
Ramr-msft 17,826 Reputation points

2021-10-14T13:11:42.287+00:00

@Caesarus Thanks for the details. Can you please share the code that you are trying.

Here is the link to get the data: https://learn.microsoft.com/en-us/azure/machine-learning/how-to-access-data#get-datastores-from-your-workspace
Caesarus 21 Reputation points

2021-10-14T13:30:44.367+00:00

Yes sure:

from azureml.core.workspace import Workspace from azureml.core.experiment import Experiment from azureml.core.compute import ComputeTarget from azureml.core import Datastore ws = Workspace.from_config() experiment = Experiment(ws, "myexperiment") compute_target = ComputeTarget(workspace = ws, name="cluster-name") datastore = Datastore.get(ws, datastore_name="workspacetsstore") data = Dataset.Tabular.from_delimited_files((datastore, "myfile.csv")) from azureml.automl.core.forecasting_parameters import ForecastingParameters from azureml.train.automl import AutoMLConfig forecasting_parameters = ForecastingParameters( time_column_name="Date", forecast_horizon=fcst_length, country_or_region_for_holidays="FR", use_stl = True, seasonality='auto', freq = "H" ) config = AutoMLConfig(task = 'forecasting', n_cross_validations=5, primary_metric="normalized_mean_absolute_error", forecasting_parameters=forecasting_parameters, enable_dnn=False, compute_target=compute_target, training_data = data, label_column_name="my_label_column _name" ) experiment.submit(config, show_output = True)

Share via

Datastore workspaceblobstore access failed, ErrorCode: ResourceNotFound using AutoML.

Your answer