kubernetes之资源调度的方式

1. 容器资源的限制

表示最多使用多少cpu和内存资源

• resources.limits.cpu
• resources.limits.memory
  表示最少使用的CPU和内存资源
• resources.requests.cpu
• resources.requests.memory

[root@master ~]# kubectl describe nodes node1.example.com  //节点的信息
省略N行
Allocated resources:
  (Total limits may be over 100 percent, i.e., overcommitted.)
  Resource           Requests   Limits
  --------           --------   ------
  cpu                100m (5%)  100m (5%)  //cpu使用率
  memory             50Mi (1%)  50Mi (1%)  //内存使用情况
  ephemeral-storage  0 (0%)     0 (0%)
  hugepages-1Gi      0 (0%)     0 (0%)
  hugepages-2Mi      0 (0%)     0 (0%)


[root@master test]# cat test.yml 
apiVersion: v1
kind: Pod
metadata:
  name: test
spec:
  containers:
  - name: b1
    image: busybox
    command: ["/bin/sh","-c","sleep 9000"]
    resources:
      requests:
        memory: 100Mi
        cpu: 0.01
    env:
    - name: HN
      valueFrom:
        fieldRef:
          fieldPath: status.podIPs

[root@master test]# kubectl apply -f test.yml  //创建一个pod进行测试
pod/test created

[root@master test]# kubectl get pod -o wide  //运行再node1上
NAME   READY   STATUS    RESTARTS   AGE   IP            NODE                NOMINATED NODE   READINESS GATES
test   1/1     Running   0          57s   10.244.1.60   node1.example.com   <none>           <none>

[root@master haproxy]# kubectl describe nodes node1.example.com
  Resource           Requests    Limits
  --------           --------    ------
  cpu                110m (5%)   100m (5%)  //cpu变成了110，这是再原来的基础上增加的，原来是100，而我们设置requests为0.01就是增加十，因为默认100才能让node1运行，所以设置最少0.01就是在此基础上加10
  memory             150Mi (5%)  50Mi (1%)// 内存变为150，同理
  ephemeral-storage  0 (0%)      0 (0%)
  hugepages-1Gi      0 (0%)      0 (0%)
  hugepages-2Mi      0 (0%)      0 (0%)


[root@master test]# cat test.yml 
---
apiVersion: v1
kind: Pod
metadata:
  name: test
spec:
  containers:
  - name: b1
    image: busybox
    command: ["/bin/sh","-c","sleep 9000"]
    resources:
      requests:
        memory: 30Mi
        cpu: 0.001
      limits:
        memory: 40Mi
        cpu: 0.01
    env:
    - name: HN
      valueFrom:
        fieldRef:
          fieldPath: status.podIPs

2. nodeSelector节点选择器

nodeSelector：用于将Pod调度到匹配Label的Node上，如果没有匹配的标签会调度失败。

作用：

• 约束Pod到特定的节点运行
• 完全匹配节点标签
• 应用场景：
• 专用节点：根据业务线将Node分组管理
• 配备特殊硬件：部分Node配有SSD硬盘、GPU

// 查看所有节点的标签
[root@master test]# kubectl get nodes --show-labels
NAME                 STATUS   ROLES                  AGE    VERSION   LABELS
master.example.com   Ready    control-plane,master   6d3h   v1.20.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=master.example.com,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=
node1.example.com    Ready    <none>                 6d2h   v1.20.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node1.example.com,kubernet