[Autoscaler] Print the value of WorkerGroupSpec.Replicas when the number of worker Pods is not as expected in autoscaling-related tests. #2980
Description
Search before asking
- I had searched in the issues and found no similar feature requirement.
Description
Some autoscaling tests are flaky because they fail to scale up or down to the correct number of Pods. There are two possible reasons:
- KubeRay doesn't scale up or down correctly in accordance with the value of
WorkerGroupSpec.Replicas. - The autoscaler doesn't make the correct autoscaling decision to update
WorkerGroupSpec.Replicas.
Hence, print the value of WorkerGroupSpec.Replicas when the number of worker Pods is not as expected in autoscaling-related tests so that we can know whether it is KubeRay's issue or the Autoscaler's issue.
Use case
No response
Related issues
No response
Are you willing to submit a PR?
- Yes I am willing to submit a PR!