K8S学习之基础五：k8s中node节点亲和性

pod调度具体会调度到哪个节点呢？这里就涉及到node亲和性的问题

node节点亲和性nodeAffinity

硬亲和性：requiredDuringSchedulingIgnoredDuringExecution，必须满足

软亲和性：preferredDuringSchedulingIgnoredDuringExecution，最好满足

# 硬亲和性
vi pod-nodeaffinity-demo.yaml 
kind: Pod
metadata:
  name:  pod-node-affinity-demo
  namespace: default
  labels:
    app: myapp
    tier: frontend
spec:
  affinity:
    nodeAffinity:
     requiredDuringSchedulingIgnoredDuringExecution:
       nodeSelectorTerms:
       - matchExpressions:
         - key: zone
           operator: In
           values:
           - foo
           - bar
  containers:
  - name: myapp
    image: 172.16.80.140/nginx/nginx:1.26
    imagePullPolicy: IfNotPresent
kubectl describe pod pod-node-affinity-demo

# pod处于pending状态，原因是没有符合硬亲和性条件的node节点

# 在node02上打上标签 zone=foo或者zone=bar，那先pod被调度到node02节点

# 软亲和性，如果没有符合条件的就忽略，weight是相对权重，权重越高，pod调度的几率越大
vi pod-nodeaffinity-demo-2.yaml
kind: Pod
metadata:
  name: pod-node-affinity-demo-2
  namespace: default
  labels:
    app: myapp
    tier: frontend
spec:
  containers:
  - name: myapp
    image: 172.16.80.140/nginx/nginx:1.26
    imagePullPolicy: IfNotPresent
  affinity:
    nodeAffinity:
      preferredDuringSchedulingIgnoredDuringExecution:
      - preference:
          matchExpressions:
          - key: zone1
            operator: In
            values:
            - foo1
            - bar1
        weight: 10
      - preference:
          matchExpressions:
          - key: zone2
            operator: In
            values:
            - foo2
            - bar2
        weight: 20
kubectl apply -f pod-nodeaffinity-demo-2.yaml