MLTable Register Data Asset from Compute Instance Azure Machine Learning

[ilhamfadhil14]

• Package Name: mltable==1.4.1, azure-ai-ml==1.8.0
• Operating System: Linux
• Python Version: 3.10.11

Describe the bug I attempted to create an MLTable from local files stored on an Azure ML compute instance and to register it as a Data Asset. However, after registration, it could not be loaded

To Reproduce

import mltable

FILE_PATH = [{"pattern" : "./data/parquet-file.parquet"}] # ganti untuk menyimpan nama file yang sesuai

tbl = mltable.from_parquet_files(FILE_PATH)

from azure.ai.ml import MLClient
from azure.ai.ml.entities import Data
from azure.ai.ml.constants import AssetTypes
from azure.identity import DefaultAzureCredential

# set VERSION variable
VERSION="1"
SUBSCRIPTION_ID=<subs-id>
RESOURCE_GROUP=<rg-name>
WORKSPACE_NAME=<ws-name>

CREDENTIAL = DefaultAzureCredential()

# connect to the AzureML workspace
# NOTE: the subscription_id, resource_group, workspace variables are set
# in the previous code snippet.
ml_client = MLClient(
    credential=CREDENTIAL, 
    subscription_id=SUBSCRIPTION_ID, 
    resource_group_name=RESOURCE_GROUP, 
    workspace_name=WORKSPACE_NAME
)

my_data = Data(
    path=MLTABLE_PATH,
    type=AssetTypes.MLTABLE,
    description="data asset from parquet",
    name="data_asset_name",
    version=VERSION,
)

ml_client.data.create_or_update(my_data)

data_asset = ml_client.data.get("data_asset_name", version="1")

tbl = mltable.load(data_asset.path)

Screenshot

Additional context I have tried change path naming to load MLTable when create it not change the final outcome

[swathipil]

Hi @ilhamfadhil14 - Thanks for the detailed report. We'll take a look asap!

@azureml-github

[github-actions[bot]]

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

[github-actions[bot]]

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

[tawan0109]

@ilhamfadhil14 thanks for reporting this issue, would you please try the latest version of mltable (1.6.0) and see if the issue still persists?

[sutharzan-ch]

I am also getting the same issue when trying to register MLTable as a data asset from Azure ML Studio web interface. I created the ML table in my workspace using Notebook and once I try to load the data asset its still pointing to the ML table in my local workspace rather than pointing to the blob I uploaded it to. I am using Python 3.10 - SDK v2 Kernel in the ML studio web interface. Not sure how to set mltable latest version. Since, I am not managing the compute should it already be the latest version?