Azure ML Workspace - Export Data Component

Question

Azure ML Workspace - Export Data Component

Dominic Parkes/IT/CREDITSAFE 0

I have a job in Azure ML workspace that has worked nicely for the last 18 months - now I can't export data to an Azure SQL DB. The component doesn't fail - but no data is written. Using the Export Data component.

This is in the log: Failure while loading azureml_run_type_providers. Failed to load entrypoint azureml.PipelineRun = azureml.pipeline.core.run:PipelineRun._from_dto with exception (cryptography 38.0.3 (/azureml-envs/azureml_5d077abbd07db0adce33a31f89d135fc/lib/python3.8/site-packages), Requirement.parse('cryptography!=1.9,!=2.0.,!=2.1.,!=2.2.*,<38.0.0'), {'azureml-core'}).

Failure while loading azureml_run_type_providers. Failed to load entrypoint azureml.ReusedStepRun = azureml.pipeline.core.run:StepRun._from_reused_dto with exception (cryptography 38.0.3 (/azureml-envs/azureml_5d077abbd07db0adce33a31f89d135fc/lib/python3.8/site-packages), Requirement.parse('cryptography!=1.9,!=2.0.,!=2.1.,!=2.2.*,<38.0.0'), {'azureml-core'}).

Failure while loading azureml_run_type_providers. Failed to load entrypoint azureml.StepRun = azureml.pipeline.core.run:StepRun._from_dto with exception (cryptography 38.0.3 (/azureml-envs/azureml_5d077abbd07db0adce33a31f89d135fc/lib/python3.8/site-packages), Requirement.parse('cryptography!=1.9,!=2.0.,!=2.1.,!=2.2.*,<38.0.0'), {'azureml-core'}).

Failure while loading azureml_run_type_providers. Failed to load entrypoint azureml.scriptrun = azureml.core.script_run:ScriptRun._from_run_dto with exception (cryptography 38.0.3 (/azureml-envs/azureml_5d077abbd07db0adce33a31f89d135fc/lib/python3.8/site-packages), Requirement.parse('cryptography!=1.9,!=2.0.,!=2.1.,!=2.2.*,<38.0.0')).

Any help much appreciated.

Anonymous

2024-10-21T16:00:27.2466667+00:00

Hi Dominic Parkes/IT/CREDITSAFE,

Thanks for Reaching the Microsoft Q&A Forum.

The error message indicates that the cryptography library version 38.0.3 is installed, but the required version should be less than 38.0.0. You might need to downgrade the cryptography library to a compatible version.

Ensure that you are using the latest version of the Azure ML SDK. Sometimes, updating the SDK can resolve compatibility issues. You can update the SDK by running.

Make sure that the export data component is correctly configured.

When previewing data in Azure Machine Learning studio, ensure that the workspace Managed Service Identity (MSI) is enabled. A SQL server user needs to be created for the service principal and workspace MSI, and that user must be granted classic database permissions.

Thankyou!
Dominic Parkes/IT/CREDITSAFE 0 Reputation points

2024-10-21T16:09:10.8833333+00:00

Hi Kothapally,

Yes that was my first thought - I was using a compute cluster, I switched this to a compute instance and down graded the cryptography library - then got 10 lines of other libraries now made incompatible by the downgrading. Nothing has changed on this job in the last 18 months and all of a sudden it doesn't write to SQL but without erroring. The component reports success - but the table remains empty. I can browse the data source and the data is there so just not sure why it won't write the data with the 'component'. Think I will just have to run this job outside of the ML workspace, it seem to add to complexity rather than assist with it.

Thanks for responding though appreciate it.

Dominic
Anonymous

2024-10-22T16:19:18.0766667+00:00

Hi Dominic Parkes/IT/CREDITSAFE,

Thank you for your follow-up query.

Might it’s possible something has changed in the environment. Check your Azure ML workspace's permissions that have the correct access to your SQL Database. Sometimes, permissions or roles can get altered without obvious notice. If any changes in the database schema or connection details could cause this issue. Running the job outside the workspace and handling the data export directly with a SQL connection might simplify things.

Thank you!
Anonymous

2024-10-23T16:18:42.1233333+00:00

Hi Dominic Parkes/IT/CREDITSAFE,

Following up to see if the above response was helpful.

Thank you!
Anonymous

2024-10-24T08:49:58.1733333+00:00

Hi Dominic Parkes/IT/CREDITSAFE,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thank you!
Dominic Parkes/IT/CREDITSAFE 0 Reputation points

2024-10-24T09:49:49.7566667+00:00

Hi Kothapally. Its not resovled - I just won't be using ML Workspace for my ML jobs in future, which is a shame, but when it adds to admin overhead rather than reduce it what options are left. Data Export Component - that doesn't fail but doesn't export any data, and just cryptic clues in the log file. I am sure eventually I could resurrect it but is it worth the effort when a simple python script running in docker can do the same. Thanks for the comments though.
Anonymous

2024-10-24T11:15:11.0066667+00:00

Hi Dominic Parkes/IT/CREDITSAFE,

Sorry for the trouble you're facing. I completely understand why you'd prefer to simplify things. If running a Python script in Docker provides the same result with less overhead, it seems like the more efficient choice for now. Also, I recommend raising a support request through the Azure portal. They can investigate the issue further and provide a more targeted solution.

Thank you.!