This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(12.8, 11%, 11%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EG%3C/text%3E%3C/svg%3E)
Hello,
I try to access a Azure ML Dataset in python and get the following error message:
Execution failed in operation 'to_pandas_dataframe' for Dataset(id='[data.id]', name='[data.name]', version=1, error_code=ScriptExecution.DatabaseConnection.Unexpected,error_message=ScriptExecutionException was caused by DatabaseConnectionException.\n DatabaseConnectionException was caused by UnexpectedException.\n 'MSSQL' encountered unexpected exception of type 'AggregateException' with HResult 'x80131500' while opening connection to server ([REDACTED]), database ([REDACTED]).\n Failed due to inner exception of type: AggregateException\n| session_id=[session-id]) ErrorCode: ScriptExecution.DatabaseConnection.Unexpected
It is a Tabular Dataset created from a Azure SQL database datasource.
For the access from the studio to the database a service principal is used.
I can access all the resources mentioned above in the standard ui.
The "sample usage" in the "consume" tab of the dataset was used for accessing the dataset in python.
Regarding environments i tried python 3.6 and 3.8 locally and in a compute instance of the ML studio.
However the same error keeps coming.
I also tried to use sync-keys as described in this question:
https://learn.microsoft.com/en-us/answers/questions/644562/datasetto-pandas-dataframe-throws-a-scriptexecutio.html
Best Regards,
Gerhard
@GerhardH Have you registered the dataset with Azure ML after you have imported the data from the Azure SQL database?
I believe you have used the import data module to pull the data from the designer which is visible in the output of the module, Ideally if you want to save the data for reuse, rather than importing a new set of data each time you need to register dataset from the Outputs+logs tab in the right panel of the component.
If you have already registered the data then a SQL connection to fetch the data again shouldn't be necessary. What are the exact sequence of steps used to fetch and load the data into the dataframe?
Also, the steps to reset the keys are not necessary because that is only required if you generate new keys for your workspace.
A DevOps-Colleague could find the cause of the problem. The service principal didn't have the correct rights for my use case.
@romungi-MSFT : I'll try your suggestion since I can create a snapshot of the table.
Thanks for your time and support!
Cheers,
Gerhard