先建立一个pod4
kubectl apply -f pod4.yaml
这个pod有node亲和性,指定到了cluster-3
apiVersion: v1 kind: Pod metadata: name: myapp-pod4 labels: app: myapp-pod4 version: v1 spec: containers: - name: myapp-pod4 image: nginx affinity: nodeAffinity: requiredDuringSchedulingIgnoredDuringExecution: nodeSelectorTerms: - matchExpressions: - key: kubernetes.io/hostname operator: In values: - xgcloud-ops-k8s-cluster-3.novalocal preferredDuringSchedulingIgnoredDuringExecution: - weight: 1 preference: matchExpressions: - key: kubernetes.io/hostname operator: In values: - xgcloud-ops-k8s-cluster-5.novalocal
kubectl get pod -o wide
给node打污点:
# 设置污点 kubectl taint nodes node1 key1=value1:NoSchedule # 节点说明中,查找 Taints 字段 kubectl describe pod pod-name # 去除污点 kubectl taint nodes node1 key1:NoSchedule-
加污点:
kubectl taint nodes xgcloud-ops-k8s-cluster-3.novalocal check=wtl:NoExecute
删除污点:
kubectl taint node xgcloud-ops-k8s-cluster-3.novalocal check=wtl:NoExecute-
查看污点:
kubectl describe node/xgcloud-ops-k8s-cluster-3.novalocal
因为不是deployment,所以pod被驱逐之后没有在建立,就没了
容忍(tolerations)
如果 operator 的值是 Exists,则 value 属性可省略
如果 operator 的值是 Equal,则表示其 key 与 value 之间的关系是 equal(等于)
如果不指定 operator 属性,则默认值为 Equal
另外,还有两个特殊值: 空的 key 如果再配合 Exists 就能匹配所有的 key 与 value,也就是是能容忍所有节点的所有 Taints 空的 effect 匹配所有的 effect
给pod4设置容忍
apiVersion: v1 kind: Pod metadata: name: myapp-pod4 labels: app: myapp-pod4 version: v1 spec: containers: - name: myapp-pod4 image: nginx affinity: nodeAffinity: requiredDuringSchedulingIgnoredDuringExecution: nodeSelectorTerms: - matchExpressions: - key: kubernetes.io/hostname operator: In values: - xgcloud-ops-k8s-cluster-3.novalocal preferredDuringSchedulingIgnoredDuringExecution: - weight: 1 preference: matchExpressions: - key: kubernetes.io/hostname operator: In values: - xgcloud-ops-k8s-cluster-5.novalocal tolerations: - key: "check" operator: "Equal" value: "wtl" effect: "NoExecute" tolerationSeconds: 3600
kubectl get pod -o wide