k8s亲和性调度-nodeAffinity _k8s

简介nodeAffinity就是节点亲和性，相对应的是Anti-Affinity，就是反亲和性。

文章插图
【k8s亲和性调度-nodeAffinity】
这种方法比nodeSelector灵活，可以有一些简单的逻辑组合。
调度可以分成软策略和硬策略两种方式：

软策略就是如果你没有满足调度要求的节点的话，POD 就会忽略这条规则，继续完成调度过程。
没有满足条件的节点的话，就不断重试直到满足条件为止。

文章插图

例子

apiVersion: Apps/v1kind: Deploymentmetadata:labels:app: hostnamename: hostnamenamespace: defaultspec:progressDeadlineSeconds: 600replicas: 2revisionHistoryLimit: 10selector:matchLabels:app: hostnamestrategy:rollingUpdate:maxSurge: 10%maxUnavailable: 1type: RollingUpdatetemplate:metadata:creationTimestamp: nulllabels:app: hostnamespec:affinity:nodeAffinity:requiredDuringSchedulingIgnoredDuringExecution:nodeSelectorTerms:- matchExpressions:- key: kube.NETes.io/hostnameoperator: Invalues:- k8s-node.example.com- centos7.example.compreferredDuringSchedulingIgnoredDuringExecution:- weight: 1preference:matchExpressions:- key: node-typeoperator: Invalues:- testingcontainers:- image: tosomeone/hostname:v1.0.3imagePullPolicy: IfNotPresentresources:requests:memory: 20Micpu: 100mlimits:memory: 30Micpu: 150mlivenessProbe:failureThreshold: 5httpGet:path: /port: 8080scheme: HTTPinitialDelaySeconds: 1periodSeconds: 2successThreshold: 1timeoutSeconds: 2name: hostnameports:- containerPort: 8080protocol: TCPreadinessProbe:failureThreshold: 3httpGet:path: /port: 8080scheme: HTTPinitialDelaySeconds: 2periodSeconds: 1successThreshold: 1timeoutSeconds: 5lifecycle:postStart:exec:command:- echo- "started"preStop:exec:command:- sleep- "3"terminationMessagePath: /dev/termination-logterminationMessagePolicy: FileDNSPolicy: ClusterFirstrestartPolicy: AlwaysschedulerName: default-schedulersecurityContext: {}terminationGracePeriodSeconds: 30

preferredDuringSchedulingIgnoredDuringExecution：指定的就是软策略
requiredDuringSchedulingIgnoredDuringExecution：指定的就是硬策略
例子中优先选择node上有node-type=testing标签的node，且这些node需要满足主机名k8s-node.example.com或者centos7.example.com
操作符有下面的几种：