Azure Machine Learning Studio environment fails to build with context

David-3633 131 Reputation points
2022-06-02T14:07:50.087+00:00

I'm trying to build an environment for ML Studio which requires some private compiled binaries. I packed these files into a Tar (about 2.8 MB, if it's relevant) and added them to the context tab of an existing environment (I want to update it). I then added this Tar in my Dockerfile and built the environment, but it failed instantly. The log tab also doesn't open (it just sits loading).

I tried changing things around to eliminate as many variables as possible:

  • Instead of using ADD I tried COPY: No change
  • I tested building the Dockerfile locally in a folder containing only the Dockerfile and the tar: It worked
  • I tested clicking the "Download Content" button and building the image in the downloaded folder: It worked
  • I tried removing all references to the Tar from the Dockerfile but still uploading the file: It failed instantly
  • I tried uploading an equivalent Zip file: It failed in the same way
  • I uploaded a single text file instead of the Tar: It worked

How can I get this to work?

Azure Container Registry
Azure Container Registry
An Azure service that provides a registry of Docker and Open Container Initiative images.
508 questions
Azure Machine Learning
Azure Machine Learning
An Azure machine learning service for building and deploying models.
3,337 questions
{count} vote

1 answer

Sort by: Most helpful
  1. David-3633 131 Reputation points
    2022-06-07T19:09:21.2+00:00

    @romungi-MSFT I managed to get the environment to run by manual building it in the registry, then registering a new environment with this registry image. However, it's annoying to have to redo this every time I need to change something. Anything else I should try to get the portal to work properly?


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.