In the same queue, a low-priority job preempts resources of a high-priority job

[wangyang0616]

What happened:
Create two jobs in the same queue. Set the priority of job-a to 100 and that of job-b to 10.

Create job-a, set the number of copies to 12, and set minAvailable to 2 to use up idle CPUs in the cluster. After all pods of job-a are running, create job-b with replicas set to 6 and minAvailable set to 2. Job-b preempts job-a resources.

What you expected to happen:
The pod of job-b is in the pending state, and the pod of job-a is not preempted.

How to reproduce it (as minimally and precisely as possible):

• job-a

apiVersion: scheduling.k8s.io/v1
kind: PriorityClass
metadata:
  name: share-1-priority
value: 100
globalDefault: false
description: "This priority class should be used for volcano job only."
---
apiVersion: batch.volcano.sh/v1alpha1
kind: Job
metadata:
  name: job-a
spec:
  priorityClassName: share-1-priority
  schedulerName: volcano
  minAvailable: 2
  tasks:
    - replicas: 12
      name: "test"
      template:
        spec:
          containers:
            - image: alpine
              command: ["/bin/sh", "-c", "sleep 1000"]
              imagePullPolicy: IfNotPresent
              name: running
              resources:
                requests:
                  cpu: "1"
          restartPolicy: OnFailure

• job-b

apiVersion: scheduling.k8s.io/v1
kind: PriorityClass
metadata:
  name: share-2-priority
value: 10
globalDefault: false
description: "This priority class should be used for volcano job only."
---
apiVersion: batch.volcano.sh/v1alpha1
kind: Job
metadata:
  name: job-b
spec:
  priorityClassName: share-2-priority
  schedulerName: volcano
  minAvailable: 2
  tasks:
    - replicas: 6
      name: "test"
      template:
        spec:
          containers:
            - image: alpine
              command: ["/bin/sh", "-c", "sleep 1000"]
              imagePullPolicy: IfNotPresent
              name: running
              resources:
                requests:
                  cpu: "1"
          restartPolicy: OnFailure

Anything else we need to know?:

apiVersion: v1
data:
  volcano-scheduler-ci.conf: |
    actions: "enqueue, allocate, backfill, reclaim, preempt"
    tiers:
    - plugins:
      - name: priority
      - name: gang
      - name: conformance
      - name: sla
    - plugins:
      - name: overcommit
      - name: drf
      - name: predicates
      - name: proportion
      - name: nodeorder
      - name: binpack

Environment:

• Volcano Version:
• Kubernetes version (use kubectl version):
• Cloud provider or hardware configuration:
• OS (e.g. from /etc/os-release):
• Kernel (e.g. uname -a):
• Install tools:
• Others: