标签:其他 lin targe 调度 这一 not 访问 time yml
Kubernetes 的Service定义了这样一种抽象:一个 Pod 的逻辑分组,一种可以访问他们的策略 —— 微服务,这一组Pod能够被Service访问到,通常是通过tabel Selector匹配。
Service 能够提供负载均衡的能力,但是在使用上有以下的限制:
只提供四层负载均衡,不存在七层负载
负载的方式有通过 userspace、iptables转发 和 ipvs 转发三种,如果系统不支持 ipvs 模块,则会自动降级为iptables转发模式
userspace:client端访问Server Pod时需要将请求转发给kube-proxy,再有kube-proxy进行调度,同时apiServer也通过Porxy获取Node节点信息,这种模式下kube-proxy的压力非常大。
iptables:client端访问Server Pod直接通过iptables进行转发,无需经过kube-Proxy
ipvs:client端访问Server Pod直接通过ipvs进行转发,前提ipvs模块必须加载
vim svc-deployment.yml
apiVersion: apps/v1 kind: Deployment metadata: name: myapp-deploy namespace: default spec: replicas: 3 selector: matchLabels: app: myapp release: stabel template: metadata: labels: app: myapp release: stabel version: v1 spec: containers: - name: myapp image: hub.vfancloud.com/test/myapp:v1 imagePullPolicy: IfNotPresent ports: - name: http containerPort: 80 --- apiVersion: v1 kind: Service metadata: name: svc-cluster namespace: default spec: type: ClusterIP selector: app: myapp release: stabel ports: - name: http port: 80 targetPort: 80
[root@Centos8 k8sYaml]# kubectl create -f svc-deployment.yml deployment.apps/myapp-deploy created service/svc-cluster created [root@Centos8 k8sYaml]# kubectl get deployment NAME READY UP-TO-DATE AVAILABLE AGE myapp-deploy 3/3 3 3 2m38s [root@Centos8 k8sYaml]# kubectl get svc NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE kubernetes ClusterIP 10.96.0.1 <none> 443/TCP 49d svc-cluster ClusterIP 10.102.16.24 <none> 80/TCP 78s [root@Centos8 k8sYaml]# kubectl get pod NAME READY STATUS RESTARTS AGE myapp-deploy-b4d49f555-czbb5 1/1 Running 0 2m47s myapp-deploy-b4d49f555-p97dk 1/1 Running 0 2m47s myapp-deploy-b4d49f555-zcjtq 1/1 Running 0 2m47s [root@Centos8 k8sYaml]# curl http://10.102.16.24/hostname.html myapp-deploy-b4d49f555-zcjtq [root@Centos8 k8sYaml]# curl http://10.102.16.24/hostname.html myapp-deploy-b4d49f555-czbb5 [root@Centos8 k8sYaml]# curl http://10.102.16.24/hostname.html myapp-deploy-b4d49f555-p97dk
vim svc-cluNone.yml
apiVersion: v1 kind: Service metadata: name: svc-clunone spec: clusterIP: "None" selector: app: myapp release: stabel ports: - port: 80 targetPort: 80
vim svc-nodeport.yml
apiVersion: v1 kind: Service metadata: name: svc-nodePort namespace: default spec: type: NodePort selector: app: myapp release: stabel ports: - name: http port: 80 # 用于k8s集群内部服务之间相互访问端口 targetPort: 80 # 实际web容器expose的端口 nodePort: 30001 # 用于外部client访问,其会在k8s每个node节点开启30001
[root@Centos8 k8sYaml]# kubectl create -f svc-nodeport.yml service/svc-nodeport created [root@Centos8 k8sYaml]# kubectl get svc NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE kubernetes ClusterIP 10.96.0.1 <none> 443/TCP 49d svc-clunone ClusterIP None <none> 80/TCP 7m37s svc-cluster ClusterIP 10.102.16.24 <none> 80/TCP 29m svc-nodeport NodePort 10.98.165.226 <none> 80:30001/TCP 6s [root@Centos8 k8sYaml]# ipvsadm -ln | grep -A5 10.98.165.226 TCP 10.98.165.226:80 rr -> 10.244.3.78:80 Masq 1 0 0 -> 10.244.3.79:80 Masq 1 0 0 -> 10.244.3.80:80 Masq 1 0 0 ### 外网访问可以在浏览器中输入nodeip+端口即可 ### 若网络不可达,在确认自己的svc创建全部正确的情况下,执行: iptables -P FORWARD ACCEPT ### 执行完毕后,再次测试
此Service类型主要功能为在Node的外部做一个负载均衡器,但是需要收费,了解即可。
apiVersion: v1 kind: Service metadata: name: svc-external namespace: default spec: type: ExternalName externalName: hub.vfancloud.com
当有需求Pod需要访问svc时,可以通过直接访问svc ip和svc name两种方式访问。
但由于svc ip是不固定的,当svc删除或者重建的时候,ip会发生改变,所以如果想固定的访问此svc,要使用coredns的方式,也就是访问svc name。
注意:使用coredns解析svc name时,要在Pod的资源清单中加入dnsPolicy: ClusterFirstWithHostNet 这个配置,该设置是使POD使用的k8s的dns,pod默认使用所在宿主主机使用的DNS(踩过的坑,不知道其他版本如何)
## 当前svc [root@Centos8 http]# kubectl get svc NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE ingress-svc1 ClusterIP 10.102.201.28 <none> 80/TCP 25m ingress-svc2 ClusterIP 10.104.176.12 <none> 8080/TCP 11m ## 当前pod [root@Centos8 http]# kubectl get pod NAME READY STATUS RESTARTS AGE ingress-http1-549f9d6cdf-5d5wf 1/1 Running 0 25m ingress-http1-549f9d6cdf-9bhkx 1/1 Running 0 25m ingress-http1-549f9d6cdf-t75tw 1/1 Running 0 25m ingress-http2-84b79f86c8-55hhw 1/1 Running 0 11m ingress-http2-84b79f86c8-6jhl4 1/1 Running 0 11m ingress-http2-84b79f86c8-7bf8w 1/1 Running 0 11m ### 前往ingress-svc2的pod内,访问ingress-svc1 ## 第一种,访问ip [root@Centos8 http]# kubectl exec -it ingress-http2-84b79f86c8-55hhw -- ping 10.102.201.28 PING 10.102.201.28 (10.102.201.28): 56 data bytes 64 bytes from 10.102.201.28: seq=0 ttl=64 time=0.114 ms 64 bytes from 10.102.201.28: seq=1 ttl=64 time=0.091 ms 64 bytes from 10.102.201.28: seq=2 ttl=64 time=0.082 ms ## 第二种,访问svc name,访问通,并成功解析到ip [root@Centos8 http]# kubectl exec -it ingress-http2-84b79f86c8-55hhw -- ping ingress-svc1 PING ingress-svc1 (10.102.201.28): 56 data bytes 64 bytes from 10.102.201.28: seq=0 ttl=64 time=0.103 ms 64 bytes from 10.102.201.28: seq=1 ttl=64 time=0.149 ms 64 bytes from 10.102.201.28: seq=2 ttl=64 time=0.087 ms
标签:其他 lin targe 调度 这一 not 访问 time yml
原文地址:https://www.cnblogs.com/v-fan/p/13252274.html