endpoints是一种kubernetes资源类型，它是namespace scoped，简写为ep:

kubectl api-resources|grep endpointsendpoints                         ep                                          true         Endpoints

endpoints有两种应用办法：

  • 随service被创立：

    • 创立service对象的同时，kubernetes会创立同名的endpoints的对象；
    • service的podSelector会将指标pod的endpoint信息(ip/port)，增加到endpoints对象中；

  • 独立创立endpoints：

    • enpdoints作为一种资源类型，反对用户创立；
    • 将内部服务的endpoint增加到endpoints对象，能够应用kubernetes service拜访内部服务；

service与endpoints

以一个简略的例子，来演示endpoints随service被主动创立进去。
创立nginx deploy，含2个pod:

apiVersion: apps/v1kind: Deploymentmetadata:  labels:    app: nginx  name: nginxspec:  replicas: 2  selector:    matchLabels:      app: nginx  template:    metadata:      labels:        app: nginx    spec:      containers:      - name: nginx        image: nginx:latest        imagePullPolicy: IfNotPresent        command:        - nginx        - -g        - "daemon off;"        workingDir: /usr/share/nginx/html        ports:        - name: http          containerPort: 80          protocol: TCP

创立nginx service:

apiVersion: v1kind: Servicemetadata:  name: nginx-svcspec:  ports:  - port: 80    targetPort: 80  selector:    app: nginx

查看其service和endpoints资源：有一个同名的endpoints被创立进去

# kubectl get svc,endpointsNAME                 TYPE        CLUSTER-IP     EXTERNAL-IP   PORT(S)   AGEservice/nginx-svc    ClusterIP   10.233.26.39   <none>        80/TCP    3m2sNAME                   ENDPOINTS                                                       AGEendpoints/nginx-svc    10.233.96.113:80,10.233.97.95:80                3m2s

其endpoints的内容，理论是nginx pod的ip+port:

# kubectl get pod -owideNAME                    READY   STATUS    RESTARTS   AGE    IP              NODE      NOMINATED NODE   READINESS GATESnginx-6dcdfd5b8-8q7l5   1/1     Running   0          4m3s   10.233.97.95    master1   <none>           <none>nginx-6dcdfd5b8-xsl96   1/1     Running   0          4m3s   10.233.96.113   master3   <none>           <none>

内部服务与endpoints

手动创立endpoints，增加内部服务的endpoint，能够将内部服务作为kubernetes的service应用。
etcd有3节点部署在内部，为其创立service: etcd-k8s，不指定podSelector：

apiVersion: v1kind: Servicemetadata:  name: etcd-k8s  namespace: kube-systemspec:  type: ClusterIP  ports:  - name: port    port: 2379    protocol: TCP

创立同名的endpoints资源: etcd-k8s，将etcd节点的ip+port填进去：

apiVersion: v1kind: Endpointsmetadata:  name: etcd-k8s  namespace: kube-systemsubsets:- addresses:  - ip: 178.104.163.38    nodeName: etc-master1  - ip: 178.104.163.187    nodeName: etc-master2  - ip: 178.104.163.243    nodeName: etc-master3  ports:  - name: port    port: 2379    protocol: TCP

这样，通过serviceName.namespace:port就能够在kubernetes集群内拜访内部的etcd服务：

etcd-k8s.kube-system:2379