8-3 Resources---多维度集群资源管理（下）

如果设置

1 requests==limits # 意味着服务是可靠的

2 不设置（不建议） # 最不可靠 会被最先杀掉

3 limits > requests # 比较可靠的服务 会根据优先级杀掉服务

 

kubernetes 通过requests 和limits 来判断服务等级，以维护服务器的安全性。

 

创建limits-test.yaml

apiVersion: v1
kind: LimitRange
metadata:
  name: test-limits
spec:
  limits:
  - max:
      cpu: 4000m
      memory: 2Gi
    min:
      cpu: 100m
      memory: 100Mi
    maxLimitRequestRatio:
      cpu: 3
      memory: 2
    type: Pod
  - default:
      cpu: 300m
      memory: 200Mi
    defaultRequest:
      cpu: 200m
      memory: 100Mi
    max:
      cpu: 2000m
      memory: 1Gi
    min:
      cpu: 100m
      memory: 100Mi
    maxLimitRequestRatio:
      cpu: 5
      memory: 4
    type: Container

pod不给默认值 是因为pod里面可能有多个容器。

 

 

创建namespace test

kubectl create ns test

创建limitrange limits-test.yaml

apiVersion: v1
kind: LimitRange
metadata:
  name: test-limits
spec:
  limits:
  - max:
      cpu: 4000m
      memory: 2Gi
    min:
      cpu: 100m
      memory: 100Mi
    maxLimitRequestRatio:
      cpu: 3
      memory: 2
    type: Pod
  - default:
      cpu: 300m
      memory: 200Mi
    defaultRequest:
      cpu: 200m
      memory: 100Mi
    max:
      cpu: 2000m
      memory: 1Gi
    min:
      cpu: 100m
      memory: 100Mi
    maxLimitRequestRatio:
      cpu: 5
      memory: 4
    type: Container

kubectl create -f limits-test.yaml  -n test

 

 

查看命名空间下所有的limits

kubectl describe limits -n test

 

创建web-test.yaml如下

#deploy
apiVersion: apps/v1
kind: Deployment
metadata:
  name: web-demo
  namespace: test
spec:
  selector:
    matchLabels:
      app: web-demo
  replicas: 1
  template:
    metadata:
      labels:
        app: web-demo
    spec:
      containers:
      - name: web-demo
        image: harbor.pdabc.com/kubernetes/web:v3
        ports:
        - containerPort: 8080

创建

kubectl apply -f web-test.yaml

查看状态

kubectl get deploy -n test

 

查看它的详细信息。

kubectl get deploy -n test web-demo -o yaml

 

查看pod的状态信息

kubectl  get pods -n test -o yaml

 

发现pod的resources里有配置 说明默认值是加到了pod的配置里面

 

创建web-test.yaml

#deploy
apiVersion: apps/v1
kind: Deployment
metadata:
  name: web-demo
  namespace: test
spec:
  selector:
    matchLabels:
      app: web-demo
  replicas: 1
  template:
    metadata:
      labels:
        app: web-demo
    spec:
      containers:
      - name: web-demo
        image: harbor.pdabc.com/kubernetes/web:v3
        ports:
        - containerPort: 8080
        resources:
          requests:
            memory: 100Mi
            cpu: 100m
          limits:
            memory: 1000Mi
            cpu: 2000m

 

kubectl apply -f web-test.yaml

 

kubectl describe deploy -n test web-demo

kubectl get deploy -n test -o yaml

看到报错信息

 

修改web-test.yaml 并启动

        resources:
          requests:
            memory: 2000Mi
            cpu: 3000m
          limits:
            memory: 3000Mi
            cpu: 4000m

限制生效

 

 

 

如果没有namespace test 则创建新的namespace test

kubectl create namespace test

创建compute-resource.yaml 

apiVersion: v1
kind: ResourceQuota
metadata:
  name: compute-resource
spec:
  hard:
    pods: 4
    requests.cpu: 2000m
    requests.memory: 4Gi
    limits.cpu: 4000m
    limits.memory: 8Gi

创建object-count.yaml

apiVersion: v1
kind: ResourceQuota
metadata:
  name: object-counts
spec:
  hard:
    configmaps: 10
    persistentvolumeclaims: 4
    replicationcontrollers: 20
    secrets: 10
    services: 10

 

创建web-test.yaml

#deploy
apiVersion: apps/v1
kind: Deployment
metadata:
  name: web-demo
  namespace: test
spec:
  selector:
    matchLabels:
      app: web-demo
  replicas: 5
  template:
    metadata:
      labels:
        app: web-demo
    spec:
      containers:
      - name: web-demo
        image: harbor.pdabc.com/kubernetes/web:v3
        ports:
        - containerPort: 8080
        resources:
          requests:
            memory: 100Mi
            cpu: 100m
          # 设置成这样 结果一个pod都起不来。明明刚刚好和resource一样大小。
          #limits:
          #  memory: 1000Mi
          #  cpu: 2000m
          limits:
            memory: 100Mi
            cpu: 100m

教学视屏中是

          limits:
            memory: 100Mi
            cpu: 200m

 

 

启动

kubectl apply -f compute-resource.yaml  -n test

kubectl apply -f object-count.yam1  -n test

kubectl apply -f web-test.yaml -n test

kubectl get deploy -n test

 

查看资源限制

kubectl  describe quota compute-resource -n test

kubectl  describe quota object-count -n  test

 

 

按照限制 只允许4个pod 有一个起不来

 

 

pod驱逐 Eviction

 

常见驱逐策略配置

--eviction-soft=memory.available<1.5Gi

--eviction-soft-grace-period=memory.available=1m30s

--eviction-hard=memory.available<100Mi,nodefs.available<1Gi,nodefs.inodesFree<5% #满足其中一个条件就执行驱逐策略

 

磁盘紧缺

删除死掉的pod 容器

删除没用的镜像

按优先级 资源 占用情况驱逐pod

 

内存紧缺

驱逐不可靠的pod

驱逐基本可靠的pod

驱逐可靠的pod