kubernetes集群应用 Service进阶

一、场景

使用kubernetes集群运行工作负载时，由于Pod经常处于用后即焚状态，Pod对应的IP地址也会经常变化，因此我们不能直接访问Pod，可以通过Service对应的端点列表（Endpoints）实现对Pod IP跟踪，进而实现通过Service访问Pod目的。

二、学习目标

• 掌握service三种代理模式
• 了解Service类型
• 了解Service参数
• 掌握Service创建方法
• 了解sessionAffinity配置方法
• 掌握修改为ipvs调度方式的方法
• 掌握DNS应用验证

三、学习步骤

序号	步骤	备注
1	service三种代理模式
2	Service类型
3	Service参数
4	Service创建方法
5	sessionAffinity类型
6	修改为IPVS调度方式的方法
7	DNS应用验证

四、课程内容

4.1 service三种代理模式

• kubernetes集群中有三层网络，一类是真实存在的，例如Node Network、Pod Network,提供真实IP地址;一类是虚拟的，例如Cluster Network或Service Network，提供虚拟IP地址，不会出现在接口上，仅会出现在Service当中

• kube-proxy始终watch（监控）kube-apiserver上关于Service相关的资源变动状态，一旦获取相关信息kube-proxy都要把相关信息转化为当前节点之上的，能够实现Service资源调度到特定Pod之上的规则，近而实现访问Service就能够获取Pod所提供的服务

• Service三种代理模式：UserSpace(用户空间)、内核空间Iptables、内核空间Ipvs

4.1.1 UserSpace

客户端访问Service时，由内核空间Service转发给当前节点用户空间kube-proxy代理后，再转给内核空间Service（iptables规则）进行分发，实现访问与Service关联的Pod，此种工作方式需要在用户空间与内核空间进行二次转化，效率低。

4.1.2 内核空间iptables

客户端直接访问本地内核中的Service IP，由内核Iptables规则直接进行调度，直接转发到Pod。

4.1.3 内核空间ipvs

客户端直接访问本地内核中的Service IP，由内核ipvs规则直接进行调度，直接转发到Pod。

4.1.4 iptables与ipvs对比

• iptables

  • 工作在内核空间

  • 优点

    • 灵活，功能强大（可以在数据包不同阶段对包进行操作）

  • 缺点

    • 表中规则过多时，响应变慢

• ipvs

  • 工作在内核空间

  • 优点

    • 转发效率高

    • 调度算法丰富：rr，wrr，lc，wlc，ip hash...

  • 缺点

    • 内核支持不全,低版本内核不能使用，需要升级到4.0或5.0以上。

• 使用iptables与ipvs时机

  • 1.10版本之前使用iptables(1.1版本之前使用UserSpace进行转发)

  • 1.11版本之后同时支持iptables与ipvs，默认使用ipvs，如果ipvs模块没有加载时，会自动降级至iptables

4.2 service类型

Service类型决定了访问Service的方法

4.2.1 service类型分类

• ClusterIP

  • 默认，分配一个集群内部可以访问的虚拟IP

• NodePort

  • 在每个Node上分配一个端口作为外部访问入口

  • nodePort端口范围为:30000-32767

• LoadBalancer

  • 工作在特定的Cloud Provider上，例如Google Cloud，AWS，OpenStack

• ExternalName

  • 表示把集群外部的服务引入到集群内部中来，即实现了集群内部pod和集群外部的服务进行通信

4.2.2 Service参数

• port 访问service使用的端口

• targetPort Pod中容器端口

• nodePort 通过Node实现外网用户访问k8s集群内service (30000-32767)

4.3 Service创建

Service的创建在工作中有两种方式，一是命令行创建，二是通过资源清单文件YAML文件创建。

4.3.1 ClusterIP类型

ClusterIP根据是否生成ClusterIP又可分为普通Service和Headless Service

Service两类：

• 普通Service:

为Kubernetes的Service分配一个集群内部可访问的固定虚拟IP(Cluster IP), 实现集群内的访问。

• Headless Service:

该服务不会分配Cluster IP, 也不通过kube-proxy做反向代理和负载均衡。而是通过DNS提供稳定的网络ID来访问，DNS会将headless service的后端直接解析为podIP列表。

4.3.1.1 普通ClusterIP Service创建

4.3.1.1.1 命令行创建Service

• 创建Deployment类型的应用

[root@master01 ~]# cat 01_create_deployment_app_nginx.yaml
apiVersion: apps/v1
kind: Deployment
metadata:name: nginx-server1
spec:replicas: 2selector:matchLabels:app: nginxtemplate:metadata:labels:app: nginxspec:containers:- name: c1image: harbor.wego.red/library/nginx:1.9.0imagePullPolicy: IfNotPresentports:- containerPort: 80

• 应用资源清单文件

[root@master01 ~]# kubectl apply -f 01_create_deployment_app_nginx.yaml

• 验证Deployment类型的创建情况

[root@master01 ~]# kubectl get deployment.apps
NAME            READY   UP-TO-DATE   AVAILABLE   AGE
nginx-server1   2/2     2            2           13s

• 创建ClusterIP类型service与Deployment类型应用关联

命令创建service
[root@master01 ~]# kubectl expose deployment.apps nginx-server1 --type=ClusterIP --target-port=80 --port=80
​
输出
service/nginx-server1 exposed
​
说明
expose 创建service
deployment.apps 控制器类型
nginx-server1 应用名称，也是service名称
--type=ClusterIP 指定service类型
--target-port=80 指定Pod中容器端口
--port=80 指定service端口

4.3.1.1.2 通过资源清单文件创建Service

[root@master01 ~]# cat 02_create_deployment_app_nginx_with_service.yaml
apiVersion: apps/v1
kind: Deployment
metadata:name: nginx-server1
spec:replicas: 2selector:matchLabels:app: nginxtemplate:metadata:labels:app: nginxspec:containers:- name: nginx-smartimage: harbor.wego.red/library/nginx:1.9.0imagePullPolicy: IfNotPresentports:- containerPort: 80
---
apiVersion: v1
kind: Service
metadata:name: nginx-svc
spec:type: ClusterIPports:- protocol: TCPport: 80targetPort: 80selector:app: nginx
​

[root@master01 ~]# kubectl  apply -f 02_create_deployment_app_nginx_with_service.yaml

• 验证

查看service
[root@master01 ~]# kubectl get service
NAME         TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)    AGE
kubernetes   ClusterIP   10.96.0.1        <none>        443/TCP    4d15h
nginx-svc    ClusterIP   10.101.153.50   <none>        80/TCP    3s

查看endpoints
[root@master01 ~]# kubectl get endpoints
NAME         ENDPOINTS                            AGE
kubernetes   192.168.122.30:6443                  4d15h
nginx-svc    172.16.189.74:80,172.16.235.150:80   8s

查看Pod
[root@master01 ~]# kubectl get pods -l app=nginx
NAME                             READY   STATUS    RESTARTS   AGE
nginx-server1-77d4c485d8-gsrmq   1/1     Running   0          12s
nginx-server1-77d4c485d8-mmc52   1/1     Running   0          12s

4.3.1.1.3 访问

[root@master01 ~]# curl http://10.101.153.50:80
<!DOCTYPE html>
<html>
<head>
<title>Welcome to nginx!</title>
<style>body {width: 35em;margin: 0 auto;font-family: Tahoma, Verdana, Arial, sans-serif;}
</style>
</head>
<body>
<h1>Welcome to nginx!</h1>
<p>If you see this page, the nginx web server is successfully installed and
working. Further configuration is required.</p><p>For online documentation and support please refer to
<a href="http://nginx.org/">nginx.org</a>.<br/>
Commercial support is available at
<a href="http://nginx.com/">nginx.com</a>.</p><p><em>Thank you for using nginx.</em></p>
</body>
</html>

4.3.1.1.4 两个pod里做成不同的主页方便测试负载均衡

[root@master01 ~]# kubectl exec -it nginx-server1-77d4c485d8-gsrmq -- /bin/bash
root@deployment-nginx-6fcfb67547-nv7dn:/# cd /usr/share/nginx/html/
root@deployment-nginx-6fcfb67547-nv7dn:/usr/share/nginx/html# echo web1 > index.html
root@deployment-nginx-6fcfb67547-nv7dn:/usr/share/nginx/html# exit
exit

[root@master01 ~]# kubectl exec -it nginx-server1-77d4c485d8-mmc52 -- /bin/bash
root@deployment-nginx-6fcfb67547-rqrcw:/# cd /usr/share/nginx/html/
root@deployment-nginx-6fcfb67547-rqrcw:/usr/share/nginx/html# echo web2 > index.html
root@deployment-nginx-6fcfb67547-rqrcw:/usr/share/nginx/html# exit
exit

4.3.1.1.5 测试

[root@master01 ~]# curl 10.101.153.50
或
[root@master01 ~]# while true;do curl 10.101.153.50;sleep 1; done

4.3.1.2 Headless Service

普通的ClusterIP service是service name解析为cluster ip,然后cluster ip对应到后面的pod ip,而无头service是指service name 直接解析为后面的pod ip

4.3.1.2.1 编写用于创建Deployment控制器类型的资源清单文件

[root@master01 ~]# cat 03_create_deployment_app_nginx.yaml
apiVersion: apps/v1
kind: Deployment
metadata:name: nginx-server1
spec:replicas: 2selector:matchLabels:app: nginxtemplate:metadata:labels:app: nginxspec:containers:- name: nginx-smartimage: harbor.wego.red/library/nginx:1.9.0imagePullPolicy: IfNotPresentports:- containerPort: 80

4.3.1.2.2 通过资源清单文件创建headless Service

编写YAML文件
命令
[root@master ~]# vim 04_headless-service.yml
apiVersion: v1
kind: Service
metadata:name: headless-servicenamespace: default
spec:type: ClusterIP     # ClusterIP类型,也是默认类型clusterIP: None     # None就代表是无头serviceports:                                # 指定service 端口及容器端口- port: 80                            # service ip中的端口protocol: TCPtargetPort: 80                      # pod中的端口selector:                             # 指定后端pod标签app: nginx                    # 可通过kubectl get pod -l app=nginx查看哪些pod在使用此标签

4.3.1.2.3 应用资源清单文件创建headless Service

命令
[root@master ~]# kubectl apply -f 04_headless_service.yml
输出
service/headless-service created

4.3.1.2.4 查看已创建的headless Service

命令
[root@master ~]# kubectl get svc
输出
NAME               TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)          AGE
headless-service   ClusterIP   None             <none>        80/TCP           2m18s
kubernetes         ClusterIP   10.96.0.1        <none>        443/TCP          5d9h
可以看到headless-service没有CLUSTER-IP,用None表示

4.3.1.2.5 DNS

DNS服务监视Kubernetes API,为每一个Service创建DNS记录用于域名解析

headless service需要DNS来解决访问问题

DNS记录格式为: <service-name>.<namespace-name>.svc.cluster.local.

4.3.1.2.5.1 查看kube-dns服务的IP

命令
[root@master1 ~]# kubectl get svc -n kube-system输出
NAME             TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)                  AGE
kube-dns         ClusterIP   10.96.0.10      <none>        53/UDP,53/TCP,9153/TCP   5d9h
metrics-server   ClusterIP   10.105.219.44   <none>        443/TCP                  45h
查看到coreDNS的服务地址是10.96.0.10

4.3.1.2.5.2 在集群主机通过DNS服务地址查找无头服务的dns解析

命令
[root@master01 ~]# dig -t A headless-service.default.svc.cluster.local. @10.96.0.10输出
; <<>> DiG 9.11.4-P2-RedHat-9.11.4-16.P2.el7_8.2 <<>> -t A headless-service.default.svc.cluster.local. @10.96.0.10
;; global options: +cmd
;; Got answer:
;; WARNING: .local is reserved for Multicast DNS
;; You are currently testing what happens when an mDNS query is leaked to DNS
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 31371
;; flags: qr aa rd; QUERY: 1, ANSWER: 1, AUTHORITY: 0, ADDITIONAL: 1
;; WARNING: recursion requested but not available;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 4096
;; QUESTION SECTION:
;headless-service.default.svc.cluster.local. IN A #被解析域名;; ANSWER SECTION:
headless-service.default.svc.cluster.local. 30 IN A 10.224.235.147 #注意这里IP;; Query time: 0 msec
;; SERVER: 10.96.0.10#53(10.96.0.10)
;; WHEN: Sun May 17 10:58:50 CST 2020
;; MSG SIZE  rcvd: 129

4.3.1.2.5.3 验证pod的IP

命令
[root@master ~]# kubectl get pod -o wide
输出
NAME                                READY   STATUS             RESTARTS   AGE   IP               NODE      NOMINATED NODE   READINESS GATES
nginx-deployment-56bf6c9c8c-jmk7r   1/1     Running            0          35m   10.224.235.147   worker1   <none>           <none>

4.3.1.2.5.4 在集群中创建一个pod验证

创建一个镜像为busyboxplus:curl的pod，pod名称为bb2,用来解析域名

命令
[root@master01 ~]# kubectl run bbp --image=busyboxplus:curl -it输出
If you don't see a command prompt, try pressing enter.解析域名
命令
[ root@bbp:/ ]$ curl http://headless-service.default.svc.cluster.local.输出
<!DOCTYPE html>
<html>
<head>
<title>Welcome to nginx!</title>
<style>body {width: 35em;margin: 0 auto;font-family: Tahoma, Verdana, Arial, sans-serif;}
</style>
</head>
<body>
<h1>Welcome to nginx!</h1>
<p>If you see this page, the nginx web server is successfully installed and
working. Further configuration is required.</p><p>For online documentation and support please refer to
<a href="http://nginx.org/">nginx.org</a>.<br/>
Commercial support is available at
<a href="http://nginx.com/">nginx.com</a>.</p><p><em>Thank you for using nginx.</em></p>
</body>
</html>
[ root@bbp:/ ]$ exit
Session ended, resume using 'kubectl attach bbp -c bbp -i -t' command when the pod is running

4.3.2 NodePort类型

• 创建资源清单文件

[root@master01 ~]# cat 05_create_nodeport_service_app.yaml
apiVersion: apps/v1
kind: Deployment
metadata:name: nginx-applabels:app: nginx-app
spec:replicas: 2selector:matchLabels:app: nginx-apptemplate:metadata:labels:app: nginx-appspec:containers:- name: c1image: harbor.wego.red/library/nginx:1.9.0imagePullPolicy: IfNotPresentports:- containerPort: 80
---
apiVersion: v1
kind: Service
metadata:name: nginx-app
spec:type: NodePortselector:app: nginx-appports:- protocol: TCPnodePort: 30001port: 8060targetPort: 80

• 应用资源清单文件

[root@master01 ~]# kubectl apply -f 05_create_nodeport_service_app.yaml
deployment.apps/nginx-app created
service/nginx-app created

• 验证service创建

[root@master01 ~]# kubectl get deployment.apps
NAME         READY   UP-TO-DATE   AVAILABLE   AGE
nginx-app    2/2     2            2           26s[root@master01 ~]# kubectl get svc
NAME         TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)          AGE
kubernetes   ClusterIP   10.96.0.1        <none>        443/TCP          2d22h
nginx-app    NodePort    10.104.157.20    <none>        8060:30001/TCP   36s[root@master01 ~]# kubectl get endpoints
NAME         ENDPOINTS                       AGE
kubernetes   192.168.122.10:6443             2d22h
nginx-app    172.16.1.24:80,172.16.2.20:80   2m10s[root@master01 ~]# ss -anput | grep ":30001"
tcp    LISTEN     0      128      :::30001                :::*                   users:(("kube-proxy",pid=5826,fd=9))[root@worker01 ~]# ss -anput | grep ":30001"
tcp    LISTEN     0      128      :::30001                :::*                   users:(("kube-proxy",pid=4937,fd=11))[root@worker02 ~]# ss -anput | grep ":30001"
tcp    LISTEN     0      128      :::30001                :::*                   users:(("kube-proxy",pid=5253,fd=11))

[root@master01 ~]# kubectl get pods
NAME                          READY   STATUS    RESTARTS   AGE
nginx-app-ffd5ccc78-cnwbx    1/1     Running   0          8m59s
nginx-app-ffd5ccc78-mz77g    1/1     Running   0          8m59s[root@master01 ~]# kubectl exec -it nginx-app-ffd5ccc78-cnwbx -- bash
root@nginx-app-ffd5ccc78-cnwbx:/# echo "nginx-app-1" > /usr/share/nginx/html/index.html
root@nginx-app-ffd5ccc78-cnwbx:/# exit
exit
[root@master01 ~]# kubectl exec -it nginx-app-ffd5ccc78-mz77g -- bash
root@nginx-app-ffd5ccc78-mz77g:/# echo "nginx-app-2" > /usr/share/nginx/html/index.html
root@nginx-app-ffd5ccc78-mz77g:/# exit
exit

• 在与kubernetes 节点同一网络主机中访问k8s集群内service

[root@bogon ~]# curl http://192.168.122.11:30001
nginx-app-2
[root@bogon ~]# curl http://192.168.122.21:30001
nginx-app-1
[root@bogon ~]# curl http://192.168.122.22:30001
nginx-app-1
[root@bogon ~]# curl http://192.168.122.23:30001
nginx-app-2

4.3.3 LoadBalancer

4.3.3.1 集群外访问过程

• 用户

• 域名

• 云服务提供商提供LB服务

• NodeIP:Port(service IP)

• Pod IP：端口

4.3.3.2 自建Kubernetes的LoadBalancer类型服务方案-MetalLB

MetalLB可以为kubernetes集群中的Service提供网络负载均衡功能。

MetalLB两大功能为:

• 地址分配，类似于DHCP

• 外部通告，一旦MetalLB为服务分配了外部IP地址，它就需要使群集之外的网络意识到该IP在群集中“存在”。MetalLB使用标准路由协议来实现此目的：ARP，NDP或BGP。

4.3.3.2.1 拉取镜像

# docker pull metallb/speaker:v0.9.3
# docker pull metallb/controller:v0.9.3

4.3.3.2.2 参考资料

参考网址：
https://metallb.universe.tf/installation/资源清单文件下载：
# kubectl apply -f https://raw.githubusercontent.com/metallb/metallb/v0.9.3/manifests/namespace.yaml
# kubectl apply -f https://raw.githubusercontent.com/metallb/metallb/v0.9.3/manifests/metallb.yaml
# On first install only
# kubectl create secret generic -n metallb-system memberlist --from-literal=secretkey="$(openssl rand -base64 128)"

4.3.3.2.3 资源清单及其应用

创建namespace查看文件内容
[root@nginx metallb]# cat namespace.yaml
apiVersion: v1
kind: Namespace
metadata:name: metallb-systemlabels:app: metallb在master01应用资源清单文件
[root@master01 ~]# kubectl apply -f http://nginx.wego.red/service/metallb/namespace.yaml

创建metallb查看内容
[root@nginx metallb]# cat metallb.yaml
apiVersion: policy/v1beta1
kind: PodSecurityPolicy
metadata:labels:app: metallbname: controllernamespace: metallb-system
spec:allowPrivilegeEscalation: falseallowedCapabilities: []allowedHostPaths: []defaultAddCapabilities: []defaultAllowPrivilegeEscalation: falsefsGroup:ranges:- max: 65535min: 1rule: MustRunAshostIPC: falsehostNetwork: falsehostPID: falseprivileged: falsereadOnlyRootFilesystem: truerequiredDropCapabilities:- ALLrunAsUser:ranges:- max: 65535min: 1rule: MustRunAsseLinux:rule: RunAsAnysupplementalGroups:ranges:- max: 65535min: 1rule: MustRunAsvolumes:- configMap- secret- emptyDir
---
apiVersion: policy/v1beta1
kind: PodSecurityPolicy
metadata:labels:app: metallbname: speakernamespace: metallb-system
spec:allowPrivilegeEscalation: falseallowedCapabilities:- NET_ADMIN- NET_RAW- SYS_ADMINallowedHostPaths: []defaultAddCapabilities: []defaultAllowPrivilegeEscalation: falsefsGroup:rule: RunAsAnyhostIPC: falsehostNetwork: truehostPID: falsehostPorts:- max: 7472min: 7472privileged: truereadOnlyRootFilesystem: truerequiredDropCapabilities:- ALLrunAsUser:rule: RunAsAnyseLinux:rule: RunAsAnysupplementalGroups:rule: RunAsAnyvolumes:- configMap- secret- emptyDir
---
apiVersion: v1
kind: ServiceAccount
metadata:labels:app: metallbname: controllernamespace: metallb-system
---
apiVersion: v1
kind: ServiceAccount
metadata:labels:app: metallbname: speakernamespace: metallb-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:labels:app: metallbname: metallb-system:controller
rules:
- apiGroups:- ''resources:- servicesverbs:- get- list- watch- update
- apiGroups:- ''resources:- services/statusverbs:- update
- apiGroups:- ''resources:- eventsverbs:- create- patch
- apiGroups:- policyresourceNames:- controllerresources:- podsecuritypoliciesverbs:- use
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:labels:app: metallbname: metallb-system:speaker
rules:
- apiGroups:- ''resources:- services- endpoints- nodesverbs:- get- list- watch
- apiGroups:- ''resources:- eventsverbs:- create- patch
- apiGroups:- policyresourceNames:- speakerresources:- podsecuritypoliciesverbs:- use
---
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:labels:app: metallbname: config-watchernamespace: metallb-system
rules:
- apiGroups:- ''resources:- configmapsverbs:- get- list- watch
---
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:labels:app: metallbname: pod-listernamespace: metallb-system
rules:
- apiGroups:- ''resources:- podsverbs:- list
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:labels:app: metallbname: metallb-system:controller
roleRef:apiGroup: rbac.authorization.k8s.iokind: ClusterRolename: metallb-system:controller
subjects:
- kind: ServiceAccountname: controllernamespace: metallb-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:labels:app: metallbname: metallb-system:speaker
roleRef:apiGroup: rbac.authorization.k8s.iokind: ClusterRolename: metallb-system:speaker
subjects:
- kind: ServiceAccountname: speakernamespace: metallb-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: RoleBinding
metadata:labels:app: metallbname: config-watchernamespace: metallb-system
roleRef:apiGroup: rbac.authorization.k8s.iokind: Rolename: config-watcher
subjects:
- kind: ServiceAccountname: controller
- kind: ServiceAccountname: speaker
---
apiVersion: rbac.authorization.k8s.io/v1
kind: RoleBinding
metadata:labels:app: metallbname: pod-listernamespace: metallb-system
roleRef:apiGroup: rbac.authorization.k8s.iokind: Rolename: pod-lister
subjects:
- kind: ServiceAccountname: speaker
---
apiVersion: apps/v1
kind: DaemonSet
metadata:labels:app: metallbcomponent: speakername: speakernamespace: metallb-system
spec:selector:matchLabels:app: metallbcomponent: speakertemplate:metadata:annotations:prometheus.io/port: '7472'prometheus.io/scrape: 'true'labels:app: metallbcomponent: speakerspec:containers:- args:- --port=7472- --config=configenv:- name: METALLB_NODE_NAMEvalueFrom:fieldRef:fieldPath: spec.nodeName- name: METALLB_HOSTvalueFrom:fieldRef:fieldPath: status.hostIP- name: METALLB_ML_BIND_ADDRvalueFrom:fieldRef:fieldPath: status.podIP- name: METALLB_ML_LABELSvalue: "app=metallb,component=speaker"- name: METALLB_ML_NAMESPACEvalueFrom:fieldRef:fieldPath: metadata.namespace- name: METALLB_ML_SECRET_KEYvalueFrom:secretKeyRef:name: memberlistkey: secretkeyimage: harbor.wego.red/library/metallb/speaker:v0.9.3 修改imagePullPolicy: Alwaysname: speakerports:- containerPort: 7472name: monitoringresources:limits:cpu: 100mmemory: 100MisecurityContext:allowPrivilegeEscalation: falsecapabilities:add:- NET_ADMIN- NET_RAW- SYS_ADMINdrop:- ALLreadOnlyRootFilesystem: truehostNetwork: truenodeSelector:beta.kubernetes.io/os: linuxserviceAccountName: speakerterminationGracePeriodSeconds: 2tolerations:- effect: NoSchedulekey: node-role.kubernetes.io/master
---
apiVersion: apps/v1
kind: Deployment
metadata:labels:app: metallbcomponent: controllername: controllernamespace: metallb-system
spec:revisionHistoryLimit: 3selector:matchLabels:app: metallbcomponent: controllertemplate:metadata:annotations:prometheus.io/port: '7472'prometheus.io/scrape: 'true'labels:app: metallbcomponent: controllerspec:containers:- args:- --port=7472- --config=configimage: harbor.wego.red/library/metallb/controller:v0.9.3 修改imagePullPolicy: Alwaysname: controllerports:- containerPort: 7472name: monitoringresources:limits:cpu: 100mmemory: 100MisecurityContext:allowPrivilegeEscalation: falsecapabilities:drop:- allreadOnlyRootFilesystem: truenodeSelector:beta.kubernetes.io/os: linuxsecurityContext:runAsNonRoot: truerunAsUser: 65534serviceAccountName: controllerterminationGracePeriodSeconds: 0	在master01应用资源清单文件
[root@master01 ~]# kubectl apply -f http://nginx.wego.red/service/metallb/metallb.yaml	

创建secret在master01节点执行如下命令
# On first install only
# kubectl create secret generic -n metallb-system memberlist --from-literal=secretkey="$(openssl rand -base64 128)"

创建metallb配置文件查看内容
[root@nginx metallb]# cat metallb-conf.yaml
apiVersion: v1
kind: ConfigMap
metadata:namespace: metallb-systemname: config
data:config: |address-pools:- name: defaultprotocol: layer2addresses:- 192.168.122.90-192.168.122.100192.168.122.90-192.168.122.100是集群节点服务器IP同一段。在master01节点应用资源清单文件
[root@master01 ~]# kubectl apply -f http://nginx.wego.red/service/metallb/metallb-conf.yaml	

4.3.3.2.4发布Service类型为LoadBalancer的Deployment控制器类型应用

创建Deployment控制器类型应用nginx-metallb及service，service类型为LoadBalancer[root@master01 ~]# vim 02_nginx-metabllb.yaml
apiVersion: apps/v1
kind: Deployment
metadata:name: nginx-metallb
spec:selector:matchLabels:app: nginxtemplate:metadata:labels:app: nginxspec:containers:- name: nginx-metallb1image: harbor.wego.red/library/nginx:1.9.0imagePullPolicy: IfNotPresentports:- containerPort: 80---
apiVersion: v1
kind: Service
metadata:name: nginx-metallb
spec:ports:- port: 8090protocol: TCPtargetPort: 80selector:app: nginxtype: LoadBalancer[root@master01 ~]# kubectl apply -f http://nginx.wego.red/service/metallb/nginx.yaml

4.3.3.2.2 验证

[root@master01 ~]# kubectl get ns
NAME                   STATUS   AGE
default                Active   16d
kube-node-lease        Active   16d
kube-public            Active   16d
kube-system            Active   16d
kubernetes-dashboard   Active   13d
metallb-system         Active   130m
test1                  Active   12d
[root@master01 ~]# kubectl get pods -n metallb-system
NAME                         READY   STATUS    RESTARTS   AGE
controller-64f8f944d-qdf8m   1/1     Running   0          110m
speaker-cwzq7                1/1     Running   0          110m
speaker-qk5fb                1/1     Running   0          110m
speaker-wsllb                1/1     Running   0          110m
speaker-x4bwt                1/1     Running   0          110m[root@master01 ~]# kubectl get svc
NAME            TYPE           CLUSTER-IP      EXTERNAL-IP      PORT(S)          AGE
kubernetes      ClusterIP      10.96.0.1       <none>           443/TCP          16d
nginx-metallb   LoadBalancer   10.105.239.69   192.168.122.90   8090:31372/TCP   106m[root@master01 ~]# ping 192.168.122.90
PING 192.168.122.90 (192.168.122.90) 56(84) bytes of data.
From 192.168.122.23: icmp_seq=2 Redirect Host(New nexthop: 192.168.122.90)
From 192.168.122.23: icmp_seq=3 Redirect Host(New nexthop: 192.168.122.90)

4.3.3.2.3 访问

[root@master01 ~]# curl http://192.168.122.90:8090
<!DOCTYPE html>
<html>
<head>
<title>Welcome to nginx!</title>
<style>body {width: 35em;margin: 0 auto;font-family: Tahoma, Verdana, Arial, sans-serif;}
</style>
</head>
<body>
<h1>Welcome to nginx!</h1>
<p>If you see this page, the nginx web server is successfully installed and
working. Further configuration is required.</p><p>For online documentation and support please refer to
<a href="http://nginx.org/">nginx.org</a>.<br/>
Commercial support is available at
<a href="http://nginx.com/">nginx.com</a>.</p><p><em>Thank you for using nginx.</em></p>
</body>
</html>

如果在IPVS模式下使用kube-proxy，从Kubernetes v1.14.2开始，必须启用ARP模式。可以通过在当前集群中编辑kube-proxy配置来实现：
# kubectl edit configmap -n kube-system kube-proxy并设置：
apiVersion: kubeproxy.config.k8s.io/v1alpha1
kind: KubeProxyConfiguration
mode: "ipvs"
ipvs:strictARP: true

4.3.4 ExternalName

4.3.4.1 ExternalName作用

• 把集群外部的服务引入到集群内部中来，实现了集群内部pod和集群外部的服务进行通信

• ExternalName 类型的服务适用于外部服务使用域名的方式，缺点是不能指定端口

• 还有一点要注意: 集群内的Pod会继承Node上的DNS解析规则。所以只要Node可以访问的服务，Pod中也可以访问到, 这就实现了集群内服务访问集群外服务

4.3.4.2 创建ExternalName Service

• 编写YAML文件

[root@master01 ~]# vim externalname.yaml
apiVersion: v1
kind: Service
metadata:name: my-service					#对应的服务是my-servicenamespace: default
spec:type: ExternalNameexternalName: www.baidu.com		#对应的外部域名为www.baidu.com

• 应用YAML文件

[root@master01 ~]# kubectl apply -f externalname.yaml
service/my-service configured

• 查看my-service的dns解析

[root@master01 ~]# kubectl get svc -n kube-system
NAME             TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)                  AGE
kube-dns         ClusterIP   10.96.0.10      <none>        53/UDP,53/TCP,9153/TCP   18d
[root@master01 ~]# dig -t A my-service.default.svc.cluster.local. @10.96.0.10; <<>> DiG 9.9.4-RedHat-9.9.4-72.el7 <<>> -t A my-service.default.svc.cluster.local. @10.96.0.10
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 43624
;; flags: qr aa rd; QUERY: 1, ANSWER: 1, AUTHORITY: 0, ADDITIONAL: 1
;; WARNING: recursion requested but not available;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 4096
;; QUESTION SECTION:
;my-service.default.svc.cluster.local. IN A;; ANSWER SECTION:
my-service.default.svc.cluster.local. 30 IN CNAME www.baidu.com.	注意这里;; Query time: 2001 msec
;; SERVER: 10.2.0.10#53(10.2.0.10)
;; WHEN: Mon Aug 05 21:23:38 CST 2019
;; MSG SIZE  rcvd: 133从上面看到把外部域名做了一个别名过来

[root@master01 ~]# kubectl run busyboxapp --image=harbor.wego.red/library/busybox:1.28.4 -it
If you don't see a command prompt, try pressing enter./ # nslookup my-service.default.svc.cluster.local.
Server:    10.96.0.10
Address 1: 10.96.0.10 kube-dns.kube-system.svc.cluster.localName:      my-service.default.svc.cluster.local.
Address 1: 2409:8c00:6c21:104f:0:ff:b03f:3ae
Address 2: 2409:8c00:6c21:104c:0:ff:b02c:156c/ # nslookup www.baidu.com
Server:    10.96.0.10
Address 1: 10.96.0.10 kube-dns.kube-system.svc.cluster.localName:      www.baidu.com
Address 1: 2409:8c00:6c21:104f:0:ff:b03f:3ae
Address 2: 2409:8c00:6c21:104c:0:ff:b02c:156c

4.4 sessionAffinity

会话粘贴

设置sessionAffinity为Clientip (类似nginx的ip_hash算法,lvs的sh算法)

[root@nginx ~]# cat 02_create_deployment_app_nginx_with_service.yaml
apiVersion: apps/v1
kind: Deployment
metadata:name: nginx-server1
spec:replicas: 2selector:matchLabels:app: nginxtemplate:metadata:labels:app: nginxspec:containers:- name: c1image: harbor.wego.red/library/nginx:1.9.0imagePullPolicy: IfNotPresentports:- containerPort: 80
---
apiVersion: v1
kind: Service
metadata:name: nginx-svc
spec:type: ClusterIPports:- protocol: TCPport: 80targetPort: 80selector:app: nginx

[root@master01 ~]# kubectl apply -f http://nginx.wego.red/service/clusterip/02_create_deployment_app_nginx_with_service.yaml
deployment.apps/nginx-server1 created
service/nginx-svc created

[root@master01 ~]# kubectl get pods
NAME                             READY   STATUS    RESTARTS   AGE
nginx-server1-58845f75f4-9zlnw   1/1     Running   0          2m11s
nginx-server1-58845f75f4-ffqdt   1/1     Running   0          2m11s
[root@master01 ~]# kubectl exec -it nginx-server1-58845f75f4-9zlnw bash
kubectl exec [POD] [COMMAND] is DEPRECATED and will be removed in a future version. Use kubectl kubectl exec [POD] -- [COMMAND] instead.
root@nginx-server1-58845f75f4-9zlnw:/# echo web1 > /usr/share/nginx/html/index.html
root@nginx-server1-58845f75f4-9zlnw:/# exit
exit
[root@master01 ~]# kubectl exec -it nginx-server1-58845f75f4-ffqdt bash
kubectl exec [POD] [COMMAND] is DEPRECATED and will be removed in a future version. Use kubectl kubectl exec [POD] -- [COMMAND] instead.
root@nginx-server1-58845f75f4-ffqdt:/# echo web2 > /usr/share/nginx/html/index.html
root@nginx-server1-58845f75f4-ffqdt:/# exit
exit

[root@master01 ~]# kubectl get svc
NAME         TYPE        CLUSTER-IP     EXTERNAL-IP   PORT(S)   AGE
kubernetes   ClusterIP   10.96.0.1      <none>        443/TCP   16d
nginx-svc    ClusterIP   10.100.53.31   <none>        80/TCP    3m53s
[root@master01 ~]# curl http://10.100.53.31
web1
[root@master01 ~]# curl http://10.100.53.31
web2
或
[root@master01 ~]# while true;do curl 10.100.53.31;sleep 1; done

[root@master01 ~]# kubectl patch svc nginx-svc -p '{"spec":{"sessionAffinity":"ClientIP"}}'
service/nginx-svc patched[root@master01 ~]# curl 10.100.53.31
web1
多次访问,会话粘贴

设置回sessionAffinity为None
[root@master01 ~]# kubectl patch svc nginx-svc -p '{"spec":{"sessionAffinity":"None"}}'
service/my-service patched

测试
[root@master01 ~]# curl 10.100.53.31
web1
多次访问,回到负载均衡
或
[root@master01 ~]# while true;do curl 10.100.53.31;sleep 1; done
web1
多次访问,会话粘贴

4.5 修改为ipvs调度方式

从kubernetes1.8版本开始，新增了kube-proxy对ipvs的支持，在kubernetes1.11版本中被纳入了GA.

4.5.1 修改为IPVS调度方式前升级内核

现使用Centos7u6发布版本，默认内核版本为3.10.0，使用kubernetes为1.18.0时，可升级内核版本至4.18.0或5.6.0版本。

在所有节点中安装,需要重启操作系统更换内核。以下升级方法供参考。

[root@localhost ~]# yum -y install perl[root@localhost ~]# rpm --import https://www.elrepo.org/RPM-GPG-KEY-elrepo.org[root@localhost ~]# yum -y install https://www.elrepo.org/elrepo-release-7.0-4.el7.elrepo.noarch.rpm[root@localhost ~]# yum  --enablerepo="elrepo-kernel"  -y install kernel-ml.x86_64 
此处升级为5.0以上版本。[root@localhost ~]# grub2-set-default 0[root@localhost ~]# grub2-mkconfig -o /boot/grub2/grub.cfg[root@localhost ~]# reboot

4.5.2 修改kube-proxy的配置文件

[root@master01 ~]# kubectl edit configmap kube-proxy -n kube-system26     iptables:27       masqueradeAll: false28       masqueradeBit: 1429       minSyncPeriod: 0s30       syncPeriod: 30s31     ipvs:32       excludeCIDRs: null33       minSyncPeriod: 0s34       scheduler: ""	  # 可以在这里修改ipvs的算法,默认为rr轮循算法35       strictARP: false36       syncPeriod: 30s37     kind: KubeProxyConfiguration38     metricsBindAddress: 127.0.0.1:1024939     mode: "ipvs"	  # 默认""号里为空,加上ipvs

4.5.3 查看kube-system的namespace中kube-proxy有关的pod

[root@master01 ~]# kubectl get pods -n kube-system |grep kube-proxy
kube-proxy-69mv6                           1/1     Running   6          2d18h
kube-proxy-jpc6c                           1/1     Running   4          4d16h
kube-proxy-kq65l                           1/1     Running   4          4d16h
kube-proxy-lmphf                           1/1     Running   5          4d16h

4.5.4 验证kube-proxy-xxx的pod中的信息

[root@master01 ~]# kubectl logs kube-proxy-jpc6c -n kube-system
W0517 00:55:10.914754       1 server_others.go:559] Unknown proxy mode "", assuming iptables proxy
I0517 00:55:10.923228       1 node.go:136] Successfully retrieved node IP: 192.168.122.32
I0517 00:55:10.923264       1 server_others.go:186] Using iptables Proxier.
I0517 00:55:10.923567       1 server.go:583] Version: v1.18.2
I0517 00:55:10.923965       1 conntrack.go:100] Set sysctl 'net/netfilter/nf_conntrack_max' to 131072
I0517 00:55:10.924001       1 conntrack.go:52] Setting nf_conntrack_max to 131072
I0517 00:55:10.924258       1 conntrack.go:83] Setting conntrack hashsize to 32768
I0517 00:55:10.927041       1 conntrack.go:100] Set sysctl 'net/netfilter/nf_conntrack_tcp_timeout_established' to 86400
I0517 00:55:10.927086       1 conntrack.go:100] Set sysctl 'net/netfilter/nf_conntrack_tcp_timeout_close_wait' to 3600
I0517 00:55:10.927540       1 config.go:315] Starting service config controller
I0517 00:55:10.927556       1 shared_informer.go:223] Waiting for caches to sync for service config
I0517 00:55:10.927576       1 config.go:133] Starting endpoints config controller
I0517 00:55:10.927594       1 shared_informer.go:223] Waiting for caches to sync for endpoints config
I0517 00:55:11.027749       1 shared_informer.go:230] Caches are synced for service config
I0517 00:55:11.027858       1 shared_informer.go:230] Caches are synced for endpoints config

4.5.5 重新启动kube-proxy

删除kube-proxy-xxx的所有pod，让它重新拉取新的kube-proxy-xxx的pod

[root@master01 ~]# kubectl delete pod kube-proxy-69mv6 -n kube-system
pod "kube-proxy-69mv6" deleted[root@master01 ~]# kubectl delete pod kube-proxy-jpc6c -n kube-system
pod "kube-proxy-jpc6c" deleted[root@master01 ~]# kubectl delete pod kube-proxy-kq65l -n kube-system
pod "kube-proxy-kq65l" deleted[root@master01 ~]# kubectl delete pod kube-proxy-lmphf -n kube-system
pod "kube-proxy-lmphf" deleted

[root@master01 ~]# kubectl get pods -n kube-system |grep kube-proxy
kube-proxy-2mk2b                           1/1     Running   0          2m23s
kube-proxy-5bj87                           1/1     Running   0          30s
kube-proxy-7qq9l                           1/1     Running   0          52s
kube-proxy-tjtqf                           1/1     Running   0          80s

随意查看其中1个或3个kube-proxy-xxx的pod,验证是否为IPVS方式了[root@master1 ~]# kubectl logs kube-proxy-tjtqf -n kube-system
I0517 02:32:26.557696       1 node.go:136] Successfully retrieved node IP: 192.168.122.32
I0517 02:32:26.557745       1 server_others.go:259] Using ipvs Proxier.
W0517 02:32:26.557912       1 proxier.go:429] IPVS scheduler not specified, use rr by default
I0517 02:32:26.560008       1 server.go:583] Version: v1.18.2
I0517 02:32:26.560428       1 conntrack.go:52] Setting nf_conntrack_max to 131072
I0517 02:32:26.561094       1 config.go:315] Starting service config controller
I0517 02:32:26.562251       1 shared_informer.go:223] Waiting for caches to sync for service config
I0517 02:32:26.561579       1 config.go:133] Starting endpoints config controller
I0517 02:32:26.562271       1 shared_informer.go:223] Waiting for caches to sync for endpoints config
I0517 02:32:26.662541       1 shared_informer.go:230] Caches are synced for service config
I0517 02:32:26.662566       1 shared_informer.go:230] Caches are synced for endpoints config

五、学习总结

六、课程预约

深入学习kubernetes，可以预约《kubernetes集群从入门到企业应用实战》相关课程。