Taints and Tolerations

https://kubernetes.io/docs/concepts/scheduling-eviction/taint-and-toleration/

Taints

Node affinity 使得pod有选择node的能力.

Taints 正好相反– 它允许Node排斥特定的Pod。也就是用在node上的。

Tolerations

Tolerations are applied to pods. Tolerations allow the scheduler to schedule pods with matching taints

Demo

给Node添加一个taint, 一个taint包含一个key和value，以及效果。 比如下面这个意味着没有任何一个pod能够schedule到这个node上，除非这个pod有一个matching toleration.

$ kubectl taint nodes k8s-worker1 key1=value1:NoSchedule   # to add
$ kubectl taint nodes k8s-worker1 key1=value1:NoSchedule-  # to delete

create a deployment with replica=3 (no pod will be scheduled on k8s-worker1)

$ kubectl taint nodes k8s-worker1 color=red:NoSchedule
$ kubectl create deployment web --image=nginx --replicas=3

add Tolerations

apiVersion: apps/v1
kind: Deployment
metadata:
  name: web
spec:
  replicas: 3
  selector:
    matchLabels:
      app: web
  template:
    metadata:
      labels:
        app: web
    spec:
      containers:
      - image: nginx
        name: nginx
      tolerations:
      - key: "color"
        operator: "Equal"
        value: "red"
        effect: "NoSchedule"