A dedicated_cores pod do not have an exclusive CPU

[flpanbin]

What happened?

I created a dedicated_cores pod, but it not have an exclusive CPU.

dedicated_cores_pod.yaml:

apiVersion: v1
kind: Pod
metadata:
  annotations:
    "katalyst.kubewharf.io/qos_level": dedicated_cores
    "katalyst.kubewharf.io/memory_enhancement": '{
      "numa_binding": "true",
      "numa_exclusive": "true"
    }'
  name: numa-dedicated-normal-pod
  namespace: default
spec:
  containers:
    - name: stress
      image: joedval/stress:latest
      command:
        - stress
        - -c
        - "1"
      imagePullPolicy: IfNotPresent
      resources:
        requests:
          cpu: "1"
          memory: 1Gi
        limits:
          cpu: "1"
          memory: 1Gi
  schedulerName: katalyst-scheduler

check the cpuset for the pod:

root@ubuntu:~# ./get_cpuset.sh numa-dedicated-normal-pod
Wed 05 Jun 2024 03:03:08 AM UTC
0-47

What did you expect to happen?

The dedicated_cores pod should be allocated an exclusive CPU core.

How can we reproduce it (as minimally and precisely as possible)?

Create a dedicated cores pod, like dedicated_cores_pod.yaml, as mentioned above.

Software version

root@ubuntu:~/katalyst/examples# helm list -A
NAME               	NAMESPACE       	REVISION	UPDATED                                	STATUS  	CHART                        	APP VERSION
katalyst-colocation	katalyst-system 	1       	2024-05-24 09:28:44.44903291 +0000 UTC 	deployed	katalyst-colocation-orm-0.5.0	v0.5.0
malachite          	malachite-system	1       	2024-05-24 09:16:19.208333849 +0000 UTC	deployed	malachite-0.1.0              	0.1.0