pod/deployment使用队列

创建测试队列：

test-queue.yaml

apiVersion: scheduling.volcano.sh/v1beta1
kind: Queue
metadata:
  name: test-queue
spec:
  capability:
    cpu: 10
    memory: 10Gi

通过Pod测试：

example-pod.yaml

apiVersion: v1
kind: Pod
metadata:
  name: example-pod
  annotations:
    # 需要注意，只能使用annotations，不能使用labels指定队列，
    # 否则无法设置队列，生成的PodGroup的队列名称为默认的default
    scheduling.volcano.sh/queue-name: "test-queue"
spec:
  schedulerName: volcano
  containers:
  - name: example-container
    image: nginx

通过Deployment测试：

example-deployment.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: example-deployment
spec:
  replicas: 1
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
      annotations:
        # 只能在annotations中设置队列名称，不能在labels中设置
        # 只能这里设置队列名称，不能在Deployment的annotations中设置
        scheduling.volcano.sh/queue-name: "test-queue"
    spec:
      schedulerName: volcano
      containers:
      - name: nginx
        image: nginx:1.14.2

gang调度

创建一个限制资源上限的队列：

test-gang-queue.yaml

apiVersion: scheduling.volcano.sh/v1beta1
kind: Queue
metadata:
  name: test-gang-queue
spec:
  capability:
    cpu: 2
    memory: 2Gi

创建一个gang调度的Job：

test-gang-job.yaml

apiVersion: batch.volcano.sh/v1alpha1
kind: Job
metadata:
  name: test-gang-job
spec:
  minAvailable: 5
  schedulerName: volcano
  queue: test-gang-queue
  policies:
  - action: CompleteJob
    event: TaskCompleted
  tasks:
  - replicas: 5
    name: worker
    template:
      spec:
        restartPolicy: Never
        containers:
        - image: alpine:latest
          imagePullPolicy: IfNotPresent
          name: worker
          command: ["sh", "-c", "sleep 1d"]
          resources:
            requests:
              cpu: 1
              memory: 100Mi

Job中的所有Pod都不会创建出来，因为队列的CPU资源上限为2，而Job中的Pod需要5个CPU。