k8s Service(一) ClusterIP、NodePort、LoadBalancer、externalIPs 演示-526互联

Service简介
kube-proxy3种不同的数据调度模式
Service资源的定义格式
示例1: ClusterIP 演示
示例2: NodePort 演示
示例3: LoadBalancer 演示
示例4: externalIPs 演示

Service简介

Service:可以理解为pod的负债均衡器,标准资源类型,Service Controller 为动态的一组Pod提供一个固定的访问入口， kubernetes完成SVC工作的是组件是kube-proxy
Endpoint Controller:管理后端端点与svc的绑定,根据标签选择器,筛选适配的pod,监控就绪的pod 并完成svc与pod的绑定
工作流程:Service Controller---->创建相同标签选择器的 Endpoint Controller根据标签选择器去管理和监听后端Pod状态完成Svc与Pod绑定

Service能够提供负载均衡的能力，但是在使用上有以下限制:

只提供4层负载均衡能力，而没有7层功能，但有时我们可能需要更多的匹配规则来转发请求，这在4层负载均衡上是不支持的

kube-proxy3种不同的数据调度模式

1.Userspace
Userspace模型:Pod-->Service, iptables拦截规则,但自己不做调度工作流程: 用户空间-->ptables(内核)-->kube-proxy-->ptables(内核)-->再调度给用户空间效率低
iptables 用户空间-->ptables(内核完成数据调度)-->调度给用户空间效率高
在iptables模型下kube-proxy的作用不在是数据调度转发,而是监听API server所有service中的定义转为本地的iptables规则
缺点:iptables模式，一个service会生成大量的规则; 如果一个service有50条规则那如果有一万个容器,内核的性能就会受到影响
ipvs代理模式: 在继承iptables优点的情况下,同时改进了iptables产生大量规则的缺点,在大规模集群中serice多的情况下优势更明显,

Service的类型

clusterIP:通过集群内部IP地址暴露服务，但该地址仅在集群内部可见、可达，它无法被集群外部的客户端访问;默认类型;建议由K8S动态指定一个;也支持用户手动明确指定;
NodePort: NodePort是ClusterIP的增强类型，它会于ClusterIP的功能之外，在每个节点上使用一个相同的端口号将外部流量引入到该Service上来。
LoadBalancer: 是NodePort的增强类型，为各节点上的NodePort提供一个外部负载均衡器;需要公有云支持
ExternalName:外部流程引入到K8S内部,借助集群上KubeDNS来实现，服务的名称会被解析为一个CNAME记录，而CNAME名称会被DNS解析为集群外部的服务的TP地址,实现内部服务与外部服务的数据交互 ExternallP 可以与ClusterIP、NodePort一起使用使用其中一个IP做出口IP

ServicePort

Service:被映射进Pod上的应用程序监听的端口; 而且如果后端Pod有多个端口，并且每个端口都想通过Service暴露的话，每个都要单独定义。最终接收请求的是PodIP和ContainerPort;

Service资源规范

Service名称空间级别的资源不能跨名称空间

apiVersion: v1
kind: Service
metadata:
  name: ..
  namespace: ...
  labels:
    key1: value1
    key2: value2
spec:
  type <string>  #Service类型，默认为ClusterIP
  selector <map[string]string> #等值类型的标签选择器，内含“与"逻辑
  ports: # Service的端口对象列表
  - name <string>#端口名称
    protocol <string> #协议，目前仅支持TCP、UDP和SCTP，默认为TCP
    port <integer> # Service的端口号
    targetPort <string> #后端目标进程的端口号或名称，名称需由Pod规范定义
    nodePort <integer> # 节点端口号，仅适用于NodePort和LoadBalancer类型
  clusterIP <string> # Service的集群IP，建议由系统自动分配
  externalTrafficPolicy <string>#外部流量策略处理方式，Local表示由当前节点处理，#Cluster表示向集群范围调度
  loadBalancerIP <string> #外部负载均衡器使用的IP地址，仅适用于LoadBlancer
  externalName <string>  # 外部服务名称，该名称将作为Service的DNS CNAME值

示例1: ClusterIP 演示

[root@k8s-master svc]# cat services-clusterip-demo.yaml 
apiVersion: v1
kind: Service
metadata:
  name: demoapp-svc
  namespace: default
spec: 
  clusterIP: 10.97.72.1   #正式部署不需要指定 会自动生成,手动指定还可能会导致冲突
  selector:               #定义过滤条件
    app: demoapp
  ports:
  - name: http
    protocol: TCP
    port: 80
    targetPort: 80    #后端pod端口

[root@k8s-master svc]# kubectl apply -f services-clusterip-demo.yaml 
service/demoapp-svc created


[root@k8s-master svc]# kubectl get svc -o wide
NAME          TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)        AGE   SELECTOR
demoapp-svc   ClusterIP   10.97.72.1       <none>        80/TCP         11s   app=demoapp
kubernetes    ClusterIP   10.96.0.1        <none>        443/TCP        30d   <none>
my-grafana    NodePort    10.96.4.185      <none>        80:30379/TCP   27d   app.kubernetes.io/instance=my-grafana,app.kubernetes.io/name=grafana
myapp         NodePort    10.106.116.205   <none>        80:31532/TCP   30d   app=myapp,release=stabel

[root@k8s-master svc]# curl 10.97.72.1  #通过访问svc IP访问到后端节点
iKubernetes demoapp v1.0 !! ClientIP: 10.244.0.0, ServerName: demoapp-66db74fcfc-9wkgj, ServerIP: 10.244.2.97!
[root@k8s-master svc]# curl 10.97.72.1
iKubernetes demoapp v1.0 !! ClientIP: 10.244.0.0, ServerName: demoapp-66db74fcfc-vzb4f, ServerIP: 10.244.1.98!

[root@k8s-master svc]# kubectl describe svc demoapp-svc
Name:              demoapp-svc
Namespace:         default
Labels:            <none>
Annotations:       <none>
Selector:          app=demoapp
Type:              ClusterIP
IP:                10.97.72.1
Port:              http  80/TCP
TargetPort:        80/TCP
Endpoints:         10.244.1.98:80,10.244.2.97:80   #后端节点
Session Affinity:  None
Events:            <none>

[root@k8s-master svc]# kubectl get pod -o wide --show-labels   #匹配到前1、2个
NAME                          READY   STATUS    RESTARTS   AGE   IP             NODE        NOMINATED NODE   READINESS GATES   LABELS
demoapp-66db74fcfc-9wkgj      1/1     Running   0          39m   10.244.2.97    k8s-node2   <none>           <none>            app=demoapp,pod-template-hash=66db74fcfc,release=stable
demoapp-66db74fcfc-vzb4f      1/1     Running   0          39m   10.244.1.98    k8s-node1   <none>           <none>            app=demoapp,pod-template-hash=66db74fcfc,release=stable,track=daily
liveness-httpget-demo         1/1     Running   3          29m   10.244.1.99    k8s-node1   <none>           <none>            app=liveness
liveness-tcpsocket-demo       1/1     Running   3          29m   10.244.1.100   k8s-node1   <none>           <none>            <none>
my-grafana-7d788c5479-kpq9q   1/1     Running   4          27d   10.244.1.84    k8s-node1   <none>           <none>            app.kubernetes.io/instance=my-grafana,app.kubernetes.io/name=grafana,pod-template-hash=7d788c5479

[root@k8s-master svc]# kubectl get ep  #实际管理后端端点与svc的绑定是Endpoints
NAME          ENDPOINTS                       AGE
demoapp-svc   10.244.1.98:80,10.244.2.97:80   2m33s
kubernetes    192.168.4.170:6443              30d
my-grafana    10.244.1.84:3000                27d
myapp         <none>                          30d

[root@k8s-master svc]#  kubectl scale deployment demoapp  --replicas=4  #修改deployment副本数为4
deployment.apps/demoapp scaled

[root@k8s-master svc]# kubectl get pod --show-labels
NAME                          READY   STATUS    RESTARTS   AGE    LABELS
demoapp-66db74fcfc-9jzs5      1/1     Running   0          18s    app=demoapp,pod-template-hash=66db74fcfc,release=stable
demoapp-66db74fcfc-9wkgj      1/1     Running   0          100m   app=demoapp,pod-template-hash=66db74fcfc,release=stable
demoapp-66db74fcfc-dw9w2      1/1     Running   0          18s    app=demoapp,pod-template-hash=66db74fcfc,release=stable
demoapp-66db74fcfc-vzb4f      1/1     Running   0          100m   app=demoapp,pod-template-hash=66db74fcfc,release=stable,track=daily
liveness-httpget-demo         1/1     Running   3          90m    app=liveness
liveness-tcpsocket-demo       1/1     Running   3          90m    <none>
my-grafana-7d788c5479-kpq9q   1/1     Running   4          27d    app.kubernetes.io/instance=my-grafana,app.kubernetes.io/name=grafana,pod-template-hash=7d788c5479

[root@k8s-master svc]# kubectl get ep  #已实时添加到ep与svc绑定
NAME          ENDPOINTS                                                   AGE
demoapp-svc   10.244.1.101:80,10.244.1.98:80,10.244.2.97:80 + 1 more...   63m
kubernetes    192.168.4.170:6443                                          30d
my-grafana    10.244.1.84:3000                                            27d
myapp         <none>                                                      30d

示例2: NodePort 演示

[root@k8s-master svc]# cat services-nodeport-demo.yaml 
apiVersion: v1
kind: Service
metadata:
  name: demoapp-nodeport-svc
  namespace: default
spec: 
  type: NodePort
  clusterIP: 10.97.56.1   #正式部署不需要指定 会自动生成手动指定还可能会导致冲突
  selector:
    app: demoapp
  ports:
  - name: http
    protocol: TCP
    port: 80
    targetPort: 80    #后端pod端口
    nodePort: 31399  #正式部署不需要指定 会自动生成   默认生成端口在30000-32768之间

[root@k8s-master svc]# kubectl apply -f services-nodeport-demo.yaml 
service/demoapp-nodeport-svc created
[root@k8s-master svc]# kubectl get pod
NAME                          READY   STATUS    RESTARTS   AGE
demoapp-66db74fcfc-9jzs5      1/1     Running   0          8m47s
demoapp-66db74fcfc-9wkgj      1/1     Running   0          109m
demoapp-66db74fcfc-dw9w2      1/1     Running   0          8m47s
demoapp-66db74fcfc-vzb4f      1/1     Running   0          109m
liveness-httpget-demo         1/1     Running   3          98m
liveness-tcpsocket-demo       1/1     Running   3          98m
my-grafana-7d788c5479-kpq9q   1/1     Running   4          27d
[root@k8s-master svc]# kubectl get svc
NAME                   TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)        AGE
demoapp-nodeport-svc   NodePort    10.97.56.1       <none>        80:31399/TCP   11s    #可以看到两个prot 其中31399就是nodeport端口
demoapp-svc            ClusterIP   10.97.72.1       <none>        80/TCP         72m

[root@k8s-master svc]# while true;do curl 192.168.4.171:31399;sleep 1;done  #通过节点IP:prot访问
iKubernetes demoapp v1.0 !! ClientIP: 10.244.2.1, ServerName: demoapp-66db74fcfc-9wkgj, ServerIP: 10.244.2.97!
iKubernetes demoapp v1.0 !! ClientIP: 10.244.2.0, ServerName: demoapp-66db74fcfc-dw9w2, ServerIP: 10.244.1.101!
iKubernetes demoapp v1.0 !! ClientIP: 10.244.2.0, ServerName: demoapp-66db74fcfc-vzb4f, ServerIP: 10.244.1.98!
iKubernetes demoapp v1.0 !! ClientIP: 10.244.2.1, ServerName: demoapp-66db74fcfc-9wkgj, ServerIP: 10.244.2.97!

可以看到上面虽然是通过节点2访问,但通过IP地址发现还是会轮询到节点1上的pod
这时就要提到 externalTrafficPolicy <string>#外部流量策略处理方式，
Local表示由当前节点处理
Cluster表示向集群范围调度

[root@k8s-master ~]# kubectl edit svc demoapp-nodeport-svc
...
spec:
  clusterIP: 10.97.56.1
  externalTrafficPolicy: Local  #把默认的Cluster改成Local
...

[root@k8s-master svc]# kubectl scale deployment demoapp  --replicas=1  #调整deployment副本数为1
deployment.apps/demoapp scaled

[root@k8s-master ~]# kubectl get pod -o wide  #可以看到唯一的pod运行node2节点上
NAME                          READY   STATUS    RESTARTS   AGE    IP             NODE        NOMINATED NODE   READINESS GATES
demoapp-66db74fcfc-9wkgj      1/1     Running   0          123m   10.244.2.97    k8s-node2   <none>           <none>
liveness-httpget-demo         1/1     Running   3          112m   10.244.1.99    k8s-node1   <none>           <none>

[root@k8s-master ~]# curl 192.168.4.171:31399  #通过节点1  失败

^C
[root@k8s-master ~]# curl 192.168.4.172:31399  #通过节点2
iKubernetes demoapp v1.0 !! ClientIP: 192.168.4.170, ServerName: demoapp-66db74fcfc-9wkgj, ServerIP: 10.244.2.97!

示例3: LoadBalancer 演示

[root@k8s-master svc]# cat services-loadbalancer-demo.yaml 
apiVersion: v1
kind: Service
metadata:
  name: demoapp-loadbalancer-svc
  namespace: default
spec: 
  type: LoadBalancer
  selector:
    app: demoapp
  ports:
  - name: http
    protocol: TCP
    port: 80
    targetPort: 80    #后端pod端口
#  loadBalancerIP: 1.2.3.4    #这里应该不是在Iaas平台上,无法创建ELB,所以无法创建

[root@k8s-master svc]# kubectl get svc
NAME                       TYPE           CLUSTER-IP       EXTERNAL-IP   PORT(S)        AGE
demoapp-loadbalancer-svc   LoadBalancer   10.110.155.70    <pending>     80:31619/TCP   31s    #可以看到因为不是Iaas平台上 EXTERNAL-IP一直为pending状态,表示一直在申请资源而挂起,依然可以通过NodePort的方式访问
demoapp-nodeport-svc       NodePort       10.97.56.1       <none>        80:31399/TCP   30m
demoapp-svc                ClusterIP      10.97.72.1       <none>        80/TCP         102m


[root@k8s-master svc]# while true;do curl 192.168.4.171:31399;sleep 1;done  #通过NodePort的方式访问
iKubernetes demoapp v1.0 !! ClientIP: 10.244.1.0, ServerName: demoapp-66db74fcfc-9wkgj, ServerIP: 10.244.2.97!
iKubernetes demoapp v1.0 !! ClientIP: 10.244.1.1, ServerName: demoapp-66db74fcfc-2jf49, ServerIP: 10.244.1.103!
iKubernetes demoapp v1.0 !! ClientIP: 10.244.1.0, ServerName: demoapp-66db74fcfc-9wkgj, ServerIP: 10.244.2.97!
iKubernetes demoapp v1.0 !! ClientIP: 10.244.1.1, ServerName: demoapp-66db74fcfc-5dp5n, ServerIP: 10.244.1.102!

示例4: externalIPs 演示

NodePort 实际应用中还需要在前面加一层负载均衡，以起到统一入口和高可用，而且后端新增的节点也不会自动添加到负载上
externalIPs 在只有1个或多个节点暴露IP的情况下，可通过虚拟IP，实现高可用

[root@k8s-master ~]# ip addr add 192.168.100.100/16 dev eth0
[root@k8s-master ~]# ip addr show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host 
       valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
    link/ether 52:54:00:44:16:16 brd ff:ff:ff:ff:ff:ff
    inet 192.168.4.170/24 brd 192.168.4.255 scope global noprefixroute eth0
       valid_lft forever preferred_lft forever
    inet 192.168.100.100/16 scope global eth0
       valid_lft forever preferred_lft forever
       
[root@k8s-master svc]# cat services-
services-clusterip-demo.yaml     services-externalip-demo.yaml    services-loadbalancer-demo.yaml  services-nodeport-demo.yaml
[root@k8s-master svc]# cat services-externalip-demo.yaml 
apiVersion: v1
kind: Service
metadata:
  name: demoapp-externalip-svc
  namespace: default
spec: 
  type: ClusterIP
  selector:
    app: demoapp
  ports:
  - name: http
    protocol: TCP
    port: 80
    targetPort: 80    #后端pod端口
  externalIPs:
  - 192.168.100.100  #实际应用中，可以通过过haproxy等实现虚拟IP 达到高可用
  
[root@k8s-master svc]# kubectl apply -f services-externalip-demo.yaml 
service/demoapp-externalip-svc created
[root@k8s-master svc]# kubectl get svc
NAME                       TYPE           CLUSTER-IP       EXTERNAL-IP       PORT(S)        AGE
demoapp-externalip-svc     ClusterIP      10.110.30.133    192.168.100.100   80/TCP         16s
demoapp-loadbalancer-svc   LoadBalancer   10.110.155.70    <pending>         80:31619/TCP   3h6m
demoapp-nodeport-svc       NodePort       10.97.56.1       <none>            80:31399/TCP   3h36m
demoapp-svc                ClusterIP      10.97.72.1       <none>            80/TCP         4h47m

#访问测试
[root@k8s-master svc]# curl 192.168.100.100
iKubernetes demoapp v1.0 !! ClientIP: 10.244.0.0, ServerName: demoapp-66db74fcfc-9wkgj, ServerIP: 10.244.2.97!
[root@k8s-master svc]# while true;do curl 192.168.100.100;sleep 1;done
iKubernetes demoapp v1.0 !! ClientIP: 10.244.0.0, ServerName: demoapp-66db74fcfc-z682r, ServerIP: 10.244.2.99!
iKubernetes demoapp v1.0 !! ClientIP: 10.244.0.0, ServerName: demoapp-66db74fcfc-5dp5n, ServerIP: 10.244.1.102!
iKubernetes demoapp v1.0 !! ClientIP: 10.244.0.0, ServerName: demoapp-66db74fcfc-5dp5n, ServerIP: 10.244.1.102!
iKubernetes demoapp v1.0 !! ClientIP: 10.244.0.0, ServerName: demoapp-66db74fcfc-9wkgj, ServerIP: 10.244.2.97!
iKubernetes demoapp v1.0 !! ClientIP: 10.244.0.0, ServerName: demoapp-66db74fcfc-5dp5n, ServerIP: 10.244.1.102!

loadbalancer externalips clusterip nodeport

targetport clusterip hostport nodeport

集群kubernetes clusterip service

nodeport service模式ipvs

loadbalancer组件openfeign原理

loadbalancer spring cloud