没有准备好的服务副本 - azure aks

Question

I am following the below link to deploy a model to azure Kube.netes service using SDK. https://learn.microsoft.com/en-us/azure/machine-learning/how-to-deploy-azure-kube.netes-service

When I call deployed service, while it is running, either by service.run() or by request.post(). I get this message "No ready replicas for service". Then in a while service goes to transitioning state. Any suggestions?

Answer 1

To figure out why your service is unavailable, you'll need to dig into the logs.

If you are most comfortable using the python interface, you can retrieve the docker container logs for your service through the service.get_logs() method. There are additional troubleshooting steps in that document which might be helpful, in particular making sure you catch exceptions in your run function.

Alternatively, you can use kubectl to interface with your AKS cluster from CLI and pull logs that way.

If you have some specific errors you want help with, please update your question to reflect the steps you are taking as well as a minimal amount of code/config which is needed to reproduce. (ref: https://stackoverflow.com/help/minimal-reproducible-example )