[kubernetes集群调度]

Ⅰ、Pod.spec.nodeName 将 Pod 直接调度到指定的 Node 节点上，会跳过 Scheduler 的调度策略，该匹配规则是强制匹配

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

apiVersion: extensions/v1beta1 kind: Deployment metadata:   name: myweb spec:   replicas: 7   template:     metadata:       labels:         app: myweb     spec:       nodeName: k8s-node01       containers:       - name: myweb         image: hub.atguigu.com/library/myapp:v1         ports:         - containerPort: 80

　　

Ⅱ、Pod.spec.nodeSelector：通过 kubernetes 的 label-selector 机制选择节点，由调度器调度策略匹配 label，而后调度 Pod 到目标节点，该匹配规则属于强制约束

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

apiVersion: extensions/v1beta1 kind: Deployment metadata:   name: myweb spec:   replicas: 2   template:     metadata:       labels:         app: myweb     spec:       nodeSelector:         type: backEndNode1       containers:       - name: myweb         image: harbor/tomcat:8.5-jre8         ports:         - containerPort: 80