Azure ML - Error starting compute instance
A couple of days ago, when I tried to start the Azure ML compute instance linked to my user, I started receiving the following error message:
Services not ready for connections
Timed out waiting for Jupyter to become ready
My role in this resource is AzureML Data Scientist, meaning I can perform all actions within an Azure Machine Learning workspace, except for creating or deleting compute resources and modifying the workspace itself. In other words, I can't delete the current instance and replace it with a new one.
In the past few days, I have tried to restart, force restart, and follow any tutorial remotely related to this issue I could find online, but I wasn’t able to find a solution for this problem.
My question: Does anyone know how I can fix the compute instance booting process, or why I'm encountering this issue (preferably without having to open a ticket and wait weeks for the IT department to help me)?
Additional Context
The day before the error started happening, I tried to install a custom package I am developing to the compute instance. The package has an extensive requirements list. I do not know if this could be somehow related, but is there a chance that the package installation is messing with the compute instance nodes?
Additionally, the SSH access to the instance is disabled, therefore I cannot use SSH to directly access its nodes files.
Workspace diagnostics
Running the workspace diagnostics also returned the following error message:
ImageBuildComputeNotValid: If Container Registry is behind the virtual network,
Container Registry cannot build your image. Set the imageBuildCompute property
to build your image.
See https://learn.microsoft.com/azure/machine-learning/how-to-secure-workspace-vnet#enable-azure-container-registry-acr
It also pointed out that the following categories have no problems:
User defined routing
- Network security group
- Resource lock
- DNS resolution
- Storage account
- Key vault
- Application Insights
- Other
The following image shows the error I am getting when trying to start the compute instance:
Any help would be appreciated!
The error is because of having different subscription dependencies to the workspace. The compute instances will be having different choices to make while creating. Like standard V2 version. Based on the subscription the options will be enabled. Kindly check with the dependencies available and make them to enable in your work.