AzureMLCompute job failed: container registry failed unexpectedly: container setup task failed

Question

Hi,

Could you please help me with running python script in azureml environment? I created the workspace and azure container registry and pushed docker image to the container. This is the example of dockerfile:

FROM python:3.7

RUN pip install --upgrade pip

RUN pip install virtualenv

ENV VIRTUAL_ENV=/venv

RUN virtualenv venv -p python3

ENV PATH="VIRTUAL_ENV/bin:$PATH"

WORKDIR /app

ADD . /app

ENV PYTHON_PACKAGES="\
     numpy \
 pandas \
 seaborn \
 matplotlib \
 sklearn \
 scipy \
 imbalanced-learn \
 xgboost \
 joblib \
" 

RUN pip install --no-cache-dir $PYTHON_PACKAGES

ENTRYPOINT ["python3","train.py"]

When I run the experiment I get this error:

"Message": "AzureMLCompute job failed.\nJobContainerConfigFailed: Container configuration failed unexpectedly\n\tJobContainerConfigFailed: Container configuration failed unexpectedly\n\terr: container setup task failed: exit status 1\n\tReason: container setup task failed: exit status 1\n\tInfo: Failed to prepare an environment for the job execution: Job environment preparation failed on 10.0.0.5 with err exit status 1."

I do not understand what this error mean.

Thank you!

Answer 1

@Momir Beljic Thanks for the question. Could you please try the following solution given below.
Solution:
Updated storage account key with below command.
Change storage account access keys - Azure Machine Learning | Microsoft Learn

az ml workspace sync-keys -w myworkspace -g myresourcegroup

This message occurs when the AML storage account restricts access to specific VNETs and the Compute Cluster isn’t in that VNET.