k8s资源之service

关注

原创最新推荐文章于 2026-03-29 10:21:59 发布 373 阅读 AI 写同款· GEO 优化 ›

K8s新手系列之Service资源当创建Service的时候会通过api-server向etcd写入创建的service的信息，而kube-proxy会基于监听的机制发现这种Service的变动，然后它会将最新的Service信息转换成对应的访问规则。服务发现：Kubernetes 中的 Pod 是动态创建和销毁的，它们的 IP 地址也是动态分配的。在kubernetes中，pod是应用程序的载体，我们可以通过pod的ip来访问应用程序，但是pod的ip地址不是固定的，这也就意味着不方便直接采用pod的ip对服务进行访问。阅读详情

欢迎关注我的公众号：

目前刚开始写一个月，一共写了18篇原创文章，文章目录如下：

istio多集群探秘，部署了50次多集群后我得出的结论

istio多集群链路追踪，附实操视频

istio防故障利器，你知道几个,istio新手不要读，太难！

istio业务权限控制，原来可以这么玩

istio实现非侵入压缩，微服务之间如何实现压缩

不懂envoyfilter也敢说精通istio系列-http-rbac-不要只会用AuthorizationPolicy配置权限

不懂envoyfilter也敢说精通istio系列-02-http-corsFilter-不要只会vs

不懂envoyfilter也敢说精通istio系列-03-http-csrf filter-再也不用再代码里写csrf逻辑了

不懂envoyfilter也敢说精通istio系列http-jwt_authn-不要只会RequestAuthorization

不懂envoyfilter也敢说精通istio系列-05-fault-filter-故障注入不止是vs

不懂envoyfilter也敢说精通istio系列-06-http-match-配置路由不只是vs

不懂envoyfilter也敢说精通istio系列-07-负载均衡配置不止是dr

不懂envoyfilter也敢说精通istio系列-08-连接池和断路器

不懂envoyfilter也敢说精通istio系列-09-http-route filter

不懂envoyfilter也敢说精通istio系列-network filter-redis proxy

不懂envoyfilter也敢说精通istio系列-network filter-HttpConnectionManager

不懂envoyfilter也敢说精通istio系列-ratelimit-istio ratelimit完全手册

————————————————

•在k8s集群中，service是一个抽象概念，它通过一个虚拟的IP映射指定的端口，将代理客户端发来的请求转到后端一组pod中的一个上。这是个神马意思呢？pod中的容器经常在不停地销毁和重建，因此pod的IP会不停的改变，这时候客户端就没法访问到pod了，现在有了service作为客户端和pod的中间层，它在这里抽象出一个虚拟IP，然后集群内部都可以通过这个虚拟IP访问到具体的pod。

常用命令：

•kubectl get svc

•kubectl label svc ServiceName type=s1

•kubectl label svc ServiceName type-

•kubectl get svc -l type=s1

•kubectl describe svc ServiceName

•kubectl edit svc ServiceName

•Kubectl delete svc ServiceName

•Kubectl delete svc -l type=s1

•Kubectl delete svc –all –n namespace

•Kubectl annotate svc ServiceName type=s1

•Kubectl annotate svc ServiceName type-

•kubectl patch service nginx-clusterip-svc -p '{"metadata":{"labels":{"aa":"bb"}}}‘

•kubectl get svc nginx-clusterip-svc -o yaml

•kubectl get svc nginx-clusterip-svc -o json

•kubectl get svc -o wide

Service的工作方式：

•Userspace方式

•iptables模型

• ipvs模型

Userspace模式：

Client Pod要访问Server Pod时,它先将请求发给本机内核空间中的service规则，由它再将请求,转给监听在指定套接字上的kube-proxy，kube-proxy处理完请求，并分发请求到指定Server Pod后,再将请求递交给内核空间中的service,由service将请求转给指定的clinet Pod。
　　由于其需要来回在用户空间和内核空间交互通信，因此效率很差

iptables：

直接由内核中的iptables规则，接受Client Pod的请求，并处理完成后，直接转发给指定ServerPod

ipvs：

接有内核中的ipvs规则来接受Client Pod请求，并处理该请求,再有内核封包后，直接发给指定的Server Pod

以上不论哪种，kube-proxy都通过watch的方式监控着kube-APIServer写入etcd中关于Pod的最新状态信息,它一旦检查到一个Pod资源被删除了或新建，它将立即将这些变化，反应再iptables 或 ipvs规则中，以便iptables和ipvs在调度Clinet Pod请求到Server Pod时，不会出现Server Pod不存在的情况

启用方法：

以service方式启动kube-proxy

• –proxy-mode= userspace

• – proxy-mode= iptables

• – proxy-mode= ipvs

以pod方式启动kube-proxy

kubectl edit cm kube-proxy -n kube-system

mode: “ipvs” or iptables or userspace

之后重启各个节点上的kube-proxy pod

Service类型：

•ExternalName

•ClusterIP

•NodePort

•LoadBalancer

clusterip：

[root@master01 service]# cat nginx-clusterIp-svc.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-clusterip-svc
spec:
 selector:  
   app: nginx
 type: ClusterIP
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

[root@master01 service]# cat nginx-clusterIp-svc-assign-address.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-clusterip-svc-with-ip
spec:
 selector:  
   app: nginx
 clusterIP: 10.68.100.100
 type: ClusterIP
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

[root@master01 service]# cat nginx-clusterIp-svc-headless.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-clusterip-svc-headless
spec:
 selector:  
   app: nginx
 type: ClusterIP
 clusterIP: None
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

[root@master01 service]# cat nginx-clusterIp-svc-with-externalip.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-clusterip-svc-with-externalip
spec:
 selector:  
   app: nginx
 type: ClusterIP
 externalIPs: 
 - 192.168.198.111
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

exteranlName：

[root@master01 service]# cat nginx-exteranlName-svc.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-externalname-svc
spec:
 externalName: www.baidu.com
 type: ExternalName

[root@master01 service]# cat nginx-exteranlName-svc-with-externalips.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-externalname-svc-with-externalips
spec:
 externalName: www.baidu.com
 type: ExternalName
 externalIPs:
 - 192.168.198.16
 - 192.168.198.17

loadblancer：

[root@master01 service]# cat nginx-loadblancer-svc.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-loadblancer-svc
spec:
 selector:  
   app: nginx
 type: LoadBalancer
 loadBalancerIP: 1.2.3.4
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

[root@master01 service]# cat nginx-loadblancer-svc-without-ip.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-loadblancer-withoutip-svc
spec:
 selector:  
   app: nginx
 type: LoadBalancer
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

[root@master01 service]# cat nginx-loadblancer-svc-with-externalips.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-loadblancer-with-externalips
spec:
 selector:  
   app: nginx
 type: LoadBalancer
 externalIPs:
 - 192.168.198.13
 - 192.168.198.14
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

nodePort：

[root@master01 service]# cat nginx-nodePort-svc.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-nodeport-svc
spec:
 selector:  
   app: nginx
 type: NodePort
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

[root@master01 service]# cat nginx-nodePort-svc-with-port.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-nodeport-svc-with-port
spec:
 selector:  
   app: nginx
 type: NodePort
 ports:
 -  name: http
    port: 8000      
    targetPort: 80
    nodePort: 31000

[root@master01 service]# cat nginx-nodePort-svc-with-externalip.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-nodeport-svc-with-externalips
spec:
 selector:  
   app: nginx
 type: NodePort
 externalIPs:
 - 192.168.198.10
 - 192.168.198.11
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

sessionaffinity：

[root@master01 service]# cat nginx-clusterIp-svc-sessionAffinity.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-clusterip-svc-with-sessionaffinity
spec:
 selector:  
   app: nginx
 type: ClusterIP
 sessionAffinity: ClientIP
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

[root@master01 service]# cat nginx-nodePort-svc-with-sessionaffinity.yal 
apiVersion: v1
kind: Service
metadata:
  name: nginx-nodeport-svc-with-sessionaffinity
spec:
 selector:  
   app: nginx
 type: NodePort
 sessionAffinity: ClientIP
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

[root@master01 service]# cat nginx-loadblancer-svc-with-sessionaffinity.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-loadblancer-with-sessionaffinity
spec:
 selector:  
   app: nginx
 type: LoadBalancer
 sessionAffinity: ClientIP
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

[root@master01 service]# cat nginx-exteranlName-svc-with-sessionaffinity.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-externalname-svc-with-sessionaffinity
spec:
 externalName: www.baidu.com
 type: ExternalName
 sessionAffinity: ClientIP

externalTrafficPolicy：

• LoadBalancer and Nodeport type services

如果服务需要将外部流量路由到本地节点或者集群级别的端点，即service type 为LoadBalancer或NodePort，那么需要指明该参数。存在两种选项：”Cluster”（默认）和 “Local”。 “Cluster” 隐藏源 IP 地址，可能会导致第二跳（second hop）到其他节点，但是全局负载效果较好。”Local” 保留客户端源 IP 地址，避免 LoadBalancer 和 NodePort 类型服务的第二跳，但是可能会导致负载不平衡

root@master01 service]# cat nginx-nodePort-svc-externalTrafficPolicy.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-nodeport-svc-externaltrafficpolicy
spec:
 selector:  
   app: nginx
 type: NodePort
 externalTrafficPolicy: Local
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

[root@master01 service]# cat nginx-loadblancer-svc-externalTrafficPolicy.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-loadblancer-svc-externaltrafficpolicy
spec:
 selector:  
   app: nginx
 type: LoadBalancer
 externalTrafficPolicy: Local
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

healthCheckNodePort：

只有当类型被设置成 “LoadBalancer” 并且 externalTrafficPolicy 被设置成 “Local” 时，才会生效

ipFamily：

[root@master01 service]# cat nginx-clusterIp-svc-ipFamily.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-clusterip-svc-ipfamily
spec:
 selector:  
   app: nginx
 type: ClusterIP
 ipFamily: IPv6
 ports:
 -  name: http
    port: 8000      
    targetPort: 80

• To enable IPv4/IPv6 dual-stack, enable the IPv6DualStack feature gate for the relevant components of your cluster, and set dual-stack cluster network assignments:

• kube -controller-manager:

– --feature-gates="IPv6DualStack=true"

– --cluster- cidr =<IPv4 CIDR>,<IPv6 CIDR> eg . --cluster- cidr =10.244.0.0/16,fc00::/24

– --service-cluster- ip -range=<IPv4 CIDR>,<IPv6 CIDR>

– --node-cidr-mask-size-ipv4|--node-cidr-mask-size-ipv6 defaults to /24 for IPv4 and /64 for IPv6

• kubelet :

– --feature-gates="IPv6DualStack=true"

• kube -proxy:

– --proxy-mode= ipvs

– --cluster- cidrs =<IPv4 CIDR>,<IPv6 CIDR>

– --feature-gates="IPv6DualStack=true"

loadBalancerSourceRanges：

• This feature is currently supported on Google Compute Engine, Google Kubernetes Engine, AWS Elastic Kubernetes Service, Azure Kubernetes Service, and IBM Cloud Kubernetes Service.

sessionAffinityConfig：

[root@master01 service]# cat nginx-clusterIp-svc-sessionAffinityConfig.yaml 
apiVersion: v1
kind: Service
metadata:
  name: nginx-clusterip-svc-sessionaffinityconfig
spec:
 selector:  
   app: nginx
 type: ClusterIP
 sessionAffinity: ClientIP
 sessionAffinityConfig:
   clientIP:
     timeoutSeconds: 2
 ports:
 -  name: http
    port: 8000      
    targetPort: 80