Azure ML Endpoint stuck in "Transitioning" state

Pranav Lakshminarayanan 6 Reputation points
2020-06-23T20:01:19.5+00:00

Hello,

I am trying to deploy several Azure ML models as webservice endpoints using an AKS cluster. I have scripted the deployment process and created a new inference cluster using the Azure ML Studio interface. The first 2-3 endpoints deploy successfully and quickly reach a "Healthy" state (in under a minute), but any subsequent deployments are stuck in the "Transitioning" state endlessly (several hours before just deleting the endpoint).

Any idea why this might be happening, or what I can do to fix this?
Is there a limit to the number of endpoints available on an inference cluster?

I noticed another question with a similar problem about 12 days ago, but that seems to have been resolved by deploying a fix to infrastructure.

Thank you in advance.

Azure Machine Learning
Azure Machine Learning
An Azure machine learning service for building and deploying models.
2,563 questions

3 answers

Sort by: Most helpful
Most helpful Newest Oldest
  1. romungi-MSFT 42,116 Reputation points Microsoft Employee
    2020-07-01T15:30:20.057+00:00

    @Pranav Lakshminarayanan Our team has deployed a fix in all the regions yesterday. Could you please check if you are able to deploy the endpoint now?

    0 comments
  2. Saumya Bahukhandi 6 Reputation points
    2020-07-29T21:27:06.36+00:00

    Hi @romungi-MSFT , I am also facing similar issue while deploying a classification model. Is this latest release with the fix globally available? This was my first deployment of an autoML trained model in AML studio.

    Also I opened another thread for this issue, before coming across this discussion
    https://learn.microsoft.com/en-us/answers/questions/54834/not-able-to-deploy-a-classification-model-using-au.html

    Thanks

    0 comments
  3. Sudeshna Sen 0 Reputation points
    2023-09-03T05:46:14.7733333+00:00

    Facing the same issue for SEA cluster.

    0 comments