Azure ML Workspace cannot create prompt flow compute session

Ruka Tsoi 0 Reputation points
2025-02-07T13:24:15.7633333+00:00

I have a ML workspace with compute instance running, also created connection with Azure OpenAI service. I am doing some tests on the prompt flows and i tried to create a compute session with my compute instance. I am getting the error below:
Session status messages: Custom application deployment failed after retry with new compute sessionXXXXXXXXXXXX, abort session creation

The diagnostics log said it try to install session with config like below:
2025-02-07 11:44:48 +0 Information Try to Install session with config: {"SessionId":"<SESSION ID>","VmSize":"","MaxIdleTimeSeconds":3600,"Identity":"","ComputeName":"<MY COMPUTE INSTANCE NAME>","EnableMultiContainer":false,"ChatGroupExecutionEnvironments":null,"Language":0,"BaseImage":null,"Packages":[],"RequirementsFileName":"requirements.txt","RuntimeImage":null}

Could you please help on fixing this? thanks.

Azure Machine Learning
Azure Machine Learning
An Azure machine learning service for building and deploying models.
3,115 questions
{count} votes

1 answer

Sort by: Most helpful
  1. Manas Mohanty 295 Reputation points Microsoft Vendor
    2025-02-11T14:06:10.9266667+00:00

    Hi Ruka Tsoi!

    Went through the logs again.

    The logs attached says VM SKU, identity and base image were null.

    Here is updated remedials based on your pointers.

    1. Are you sure that you are passing VM SKU in with select compute types as "compute instances" and selecting base image as "mcr.microsoft.com/azureml/promptflow/promptflow-runtime:latest" or the default one showing up. (UI side)
    2. Check for other SKU if you are trying with same SKU. (There are SKU constrains on region wise sometime.
    3. Please enable either system assigned identity or use managed identity on your computes and add them as storage contributor, key Vault contributor, Azure Container Registry (to your respective container registry) and ML workspace contributor. Reference. (Permission)
    4. Please check with prerequisites on storage location and permission (Data scientist role needed, Storage File Data Privileged Contributor, Storage blob contributor) - Prerequisites.
    5. Switch to Azure Managed virtual network for better experience (Test in a dev workspace first). Known limitation.(Virtual network side)
    6. You have enabled outbound access to required service tags mentioned here.

    Please check network isolation for prompt flow for more details

    Hope above pointers fix your issue now.

    Thank you.


Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.