Skip to content

Local Queue Initialization Failed

Issue Description

When creating a Notebook, training task, or inference service, if the queue is being used for the first time in that namespace, there will be a prompt to initialize the queue with one click. However, the initialization fails.

local-queue-initialization-failed

Issue Analysis

In the intelligent computing environment, the queue management capability is provided by Kueue. Kueue provides two types of queue management resources:

  • ClusterQueue: A cluster-level queue mainly used to manage resource quotas within the queue, including CPU, memory, and GPU.
  • LocalQueue: A namespace-level queue that needs to point to a ClusterQueue for resource allocation within the queue.

In the intelligent computing environment, if a service is created and the specified namespace does not have a LocalQueue, there will be a prompt to initialize the queue.

In rare cases, the LocalQueue initialization might fail due to special reasons.

Solution

Check if Kueue is running normally. If the kueue-controller-manager is not running, you can check it with the following command:

kubectl get deploy kueue-controller-manager -n baize-system

If the kueue-controller-manager is not running properly, fix Kueue first.

References

Comments