fix: deployment and pod, high priority cannot preempt low priority resources

[wangyang0616]

Signed-off-by: wangyang wangyang289@huawei.com

fix: #2629

The test results are as follows:

[root@ecs-4b42-0002 volcano]# kubectl get pod 
NAME                           READY   STATUS    RESTARTS   AGE
deploy-high-587b867944-9qdvm   1/1     Running   0          40m
deploy-high-587b867944-dzfsr   0/1     Pending   0          40m
deploy-high-587b867944-msvxv   0/1     Pending   0          40m
deploy-high-587b867944-qtfbb   0/1     Pending   0          40m
deploy-low-67b97db94b-7wt2h    1/1     Running   0          40m
deploy-low-67b97db94b-scsx5    1/1     Running   0          40m
deploy-low-67b97db94b-vw8fv    1/1     Running   0          40m
deploy-low-67b97db94b-z44fq    0/1     Pending   0          40m

deploy-high can preempt the resources of deploy-low, but only one copy has preempted resources.

Because when the deployment type creates a podgroup, the default setting minAvailble is 1. Under the control of the gang plugin, only one copy will be preempted.

If you want the behavior to be exactly the same as kube-scheduler and trigger the preemption action of all replicas, you can configure the following information in the volcano-test-scheduler-configmap file

- name: gang
   enableJobStarving: false

[jiangkaihua]

/lgtm

[william-wang]

/approve

[volcano-sh-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: william-wang

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• pkg/scheduler/plugins/OWNERS [william-wang]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment