default queue with a pod has pvc will be overused

[lowang-bh]

What happened:

What you expected to happen:

pod scheduled succcessfully with pvc

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

apiVersion: apps/v1
kind: Deployment
metadata:
  name: deployment-volcano-pvc
spec:
  selector:
    matchLabels:
      app: deployment-volcano-pvc
  replicas: 1
  template:
    metadata:
      annotations:
        scheduling.k8s.io/group-name: pg-volcano-pvc
      labels:
        app: deployment-volcano-pvc
    spec:
      schedulerName: volcano
      terminationGracePeriodSeconds: 1
      containers:
        - name: nginx
          image: nginx:1.14.2
          ports:
            - containerPort: 8080
          resources:
            requests:
              cpu: '100m'
              memory: 100Mi
            limits:
              cpu: '100m'
              memory: 100Mi
          volumeMounts:
            - mountPath: /mnt/claim
              name: local-vol
      volumes:
        - name: local-vol
          persistentVolumeClaim:
            claimName: claim-for-deployment-volcano-pvc
---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: claim-for-deployment-volcano-pvc
spec:
  accessModes:
  - ReadWriteOnce
  resources:
    requests:
      storage: 100Mi
  # storageClassName: csi-lvm # null to use default storage class
  volumeMode: Filesystem
---
apiVersion: scheduling.volcano.sh/v1beta1
kind: PodGroup
metadata:
  name: pg-volcano-pvc
  namespace: default
spec:
  minMember: 1
  queue: default
---
apiVersion: scheduling.volcano.sh/v1beta1
kind: Queue
metadata:
  name: default
spec:
  guarantee:
    resource:
      cpu: 2
      memory: 1Gi
  reclaimable: true
  weight: 1

Anything else we need to know?:

Environment:

• Volcano Version: master
• Kubernetes version (use kubectl version): client v1.27.1, server: v1.28.0
• Cloud provider or hardware configuration:
• OS (e.g. from /etc/os-release):
• Kernel (e.g. uname -a):
• Install tools:
• Others:

[Monokaix]

Hi, please paste cisnode resource yaml output.

[lowang-bh]

This is the storage class, and I checked it works on v1.7.0

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  annotations:
    kubectl.kubernetes.io/last-applied-configuration: |
      {"apiVersion":"storage.k8s.io/v1","kind":"StorageClass","metadata":{"annotations":{"storageclass.kubernetes.io/is-default-class":"true"},"name":"standard"},"provisioner":"rancher.io/local-path","reclaimPolicy":"Delete","volumeBindingMode":"WaitForFirstConsumer"}
    storageclass.kubernetes.io/is-default-class: "true"
  creationTimestamp: "2023-12-20T14:53:36Z"
  name: standard
  resourceVersion: "267"
  uid: fac2253f-22bb-4d32-a211-39896c8bb829
provisioner: rancher.io/local-path
reclaimPolicy: Delete
volumeBindingMode: WaitForFirstConsumer

[lowang-bh]

v1.8.0 also works

[Monokaix]

When set csi resource, volcano scheduler will check spec.driver of csinode, if spec.driver is null, the csi resource will not be added to node.Allocatable, and the csi reource will be missing in cluster resources.

Then deserved resource of csi resource computed in proportion will be zero, and task's request is 1, so it's not allocatable because of not enough csi resource.