AzureML OutputFileDatasetConfig not useable with destination set

[DominiqueLade]

• Package Name: azureml-core
• Package Version: 1.21.0.post1
• Operating System: MacOS
• Python Version: 3.7.4

Describe the bug
I am not able to use OutputFileDatasetConfig with a destination pointing to a Blobstorage or DataLake Gen 2.

The following error is raised:

File "/.venv/lib/python3.7/site-packages/azureml/data/output_dataset_config.py", line 459, in to_data_location 
datastore_name = destination[0].name

TypeError: 'AzureBlobDatastore' object is not subscriptable

To Reproduce
Steps to reproduce the behavior:

# Connect to Azure ML
interactive_auth = InteractiveLoginAuthentication(tenant_id=tenant_id)
ws = Workspace(subscription_id, resource_group, workspace_name, auth=interactive_auth)
experiment = Experiment(workspace=ws, name="test-experiment")

# Get Dataset
my-output-dataset = OutputFileDatasetConfig(name="attachments", destination=ws.datastores["my-datastore"])

# Set Run Config
pipeline_path = Path(__file__).parent.absolute()
config = ScriptRunConfig(
    source_directory='./',
    script="my-script.py",
    compute_target="default-compute-target",
    arguments=[
        "--path",
        my-output-dataset.as_mount(),
    ],
)

# Execute Script
run = experiment.submit(config)

Expected behavior
Mount Blobstorage and Datalake Gen 2 as output dataset to a compute target

Screenshots

Additional context

[DominiqueLade]

Hi,

unfortunately, I still have the same issue as described above. This is currently a show stopper for us.

Is this the right place to fill the bug report? Is anyone investigating this?

[yeamusic21]

We tried to do this as well, and it didn't work for us either. You can use File Share, which will return a result, but we found other bugs even when we found a workaround for this.

[DominiqueLade]

@yeamusic21 Thanks for sharing your insights. Do you mind to elaborate a bit about the other bugs?

[yeamusic21]

@DominiqueLade - #16568

Also, https://stackoverflow.com/questions/65240603/azure-ml-studio-ml-pipeline-exception-no-temp-file-found

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @azureml-github.

Issue Details

---

• Package Name: azureml-core
• Package Version: 1.21.0.post1
• Operating System: MacOS
• Python Version: 3.7.4

Describe the bug
I am not able to use OutputFileDatasetConfig with a destination pointing to a Blobstorage or DataLake Gen 2.

The following error is raised:

File "/.venv/lib/python3.7/site-packages/azureml/data/output_dataset_config.py", line 459, in to_data_location 
datastore_name = destination[0].name

TypeError: 'AzureBlobDatastore' object is not subscriptable

To Reproduce
Steps to reproduce the behavior:

# Connect to Azure ML
interactive_auth = InteractiveLoginAuthentication(tenant_id=tenant_id)
ws = Workspace(subscription_id, resource_group, workspace_name, auth=interactive_auth)
experiment = Experiment(workspace=ws, name="test-experiment")

# Get Dataset
my-output-dataset = OutputFileDatasetConfig(name="attachments", destination=ws.datastores["my-datastore"])

# Set Run Config
pipeline_path = Path(__file__).parent.absolute()
config = ScriptRunConfig(
    source_directory='./',
    script="my-script.py",
    compute_target="default-compute-target",
    arguments=[
        "--path",
        my-output-dataset.as_mount(),
    ],
)

# Execute Script
run = experiment.submit(config)

Expected behavior
Mount Blobstorage and Datalake Gen 2 as output dataset to a compute target

Screenshots

Additional context

Author:	DominiqueLade
Assignees:	-
Labels:	Client, Machine Learning, Service Attention, bug, customer-reported, needs-triage, question
Milestone:	-

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @azureml-github.

Issue Details

---

• Package Name: azureml-core
• Package Version: 1.21.0.post1
• Operating System: MacOS
• Python Version: 3.7.4

Describe the bug
I am not able to use OutputFileDatasetConfig with a destination pointing to a Blobstorage or DataLake Gen 2.

The following error is raised:

File "/.venv/lib/python3.7/site-packages/azureml/data/output_dataset_config.py", line 459, in to_data_location 
datastore_name = destination[0].name

TypeError: 'AzureBlobDatastore' object is not subscriptable

To Reproduce
Steps to reproduce the behavior:

# Connect to Azure ML
interactive_auth = InteractiveLoginAuthentication(tenant_id=tenant_id)
ws = Workspace(subscription_id, resource_group, workspace_name, auth=interactive_auth)
experiment = Experiment(workspace=ws, name="test-experiment")

# Get Dataset
my-output-dataset = OutputFileDatasetConfig(name="attachments", destination=ws.datastores["my-datastore"])

# Set Run Config
pipeline_path = Path(__file__).parent.absolute()
config = ScriptRunConfig(
    source_directory='./',
    script="my-script.py",
    compute_target="default-compute-target",
    arguments=[
        "--path",
        my-output-dataset.as_mount(),
    ],
)

# Execute Script
run = experiment.submit(config)

Expected behavior
Mount Blobstorage and Datalake Gen 2 as output dataset to a compute target

Screenshots

Additional context

Author:	DominiqueLade
Assignees:	-
Labels:	Client, Machine Learning, Service Attention, bug, customer-reported, needs-triage, question
Milestone:	-