-
Notifications
You must be signed in to change notification settings - Fork 40
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
deploy_model
does not complete when the environment has a CRAN/GitHub package
#414
Comments
This currently occurs even if you do not specify custom environment. Basically just creating experiment with basic "Hello world" script fails on image build step. |
I confirm that deployment fails even for the default environment without additional packages. The deployment hangs and stays in "Transitioning" state indefinitely. There are no deployment logs available. The only way to remove the endpoint is using Azure CLI, the workspace UI does not allow removing endpoints in this state. To debug the issue, I first registered a plain R environment in AML Workspace and then used this in the deployment. The environment build log reveals that the environment build runs into timeout at step 23:
I have previously built custom R environments and noticed that the |
It seems problem is that by default As a workaround, what you can do is build and use custom environment. I used something like this (
And then point to that image when you are setting up environment in script:
|
Yes, the custom Dockerfile approach works for me too. As the R SDK will be deprecated soon, I guess this won't be fixed. Hopefully the new approach will be more stable. |
Describe the bug
deploy_model
does not complete when the environment has a CRAN/GitHub package.To Reproduce
Expected behavior
The model is deployed.
Log
Additional context
azuremlsdk (R): 1.11.0
azureml-sdk (Python): 1.24.0
Both anaconda and miniconda have been tried and neither worked.
I suppose something went wrong with the
-c r
channel inconda install
, probably?The text was updated successfully, but these errors were encountered: