看懂本文要具备一下知识点:
-
Service实现原理和会应用
-
知道反向代理原理,了解nginx和apache的vhost概念
-
了解service的几种类型(Nodeport、clusterip、LB)
-
四层和七层区别(不明白就这样去理解,七层最常见就是应用层的http,也就是url,四层是传输层,为tcp/udp端口)
-
域名解析,/etc/hosts等基础知识
Ingress Controller介绍
Ingress Controller
-
Ingress NGINX: Kubernetes 官方维护的方案,也是本次安装使用的 Controller。
-
F5 BIG-IP Controller: F5 所开发的 Controller,它能够让管理员通过 CLI 或 API 让 Kubernetes 与 OpenShift 管理 F5 BIG-IP 设备。
-
Ingress Kong: 著名的开源 API Gateway 方案所维护的 Kubernetes Ingress Controller。
-
Traefik: 是一套开源的 HTTP 反向代理与负载均衡器,而它也支援了 Ingress。
-
Voyager: 一套以 HAProxy 为底的 Ingress Controller。
Ingress Controller 的实现不只上面这些方案,还有很多可以在网络上找到这里不一一列出来了。
我们部署在集群里的服务的svc想暴露出来的时候,从长久眼光看和易于管理维护都是用的Ingress Controller
来处理,clusterip非集群主机无法访问,Nodeport不方便长久管理和效率,LB服务多了不方便因为需要花费额外的钱,externalIPS不好用(后面有空写文章会说它)。
我们跑的大多服务都是应用层http(s),Ingress Controller使用service或者pod的网络将它暴露在集群外,然后它反向代理集群内的七层服务,通过vhost子域名那样路由到后端的服务,Ingress Controller
工作架构如下,借用traefik官方的图。
你可以将api.domain.com
进来的流量路由到集群里api的pod,你可以将backoffice.domain.com
流量路由到backoffice的一组pod上,虽说我们可以自己搭建一个nginx来代替掉Ingress Controller
,但是要增加代理的service长期来看维护很不方便,在使用上Ingress Controller
后可以用一种抽象的对象告诉controller添加对应的代理,也就是kind: Ingress
。它里面描述了从Ingress Controller访问进来的ServerName和web的url要代理到集群里哪个service(以及service的port)等等具体信息。
而官方的Ingress Nginx
可以视为一个魔改的nginx,拥有集群赋予的RBAC权限后,能够有监听集群Ingress相关的变化能力,用户创建了kind: Ingress
, 例如上面trafik图里的Ingress大致就是下面这样:
apiVersion: extensions/v1beta1kind: Ingressmetadata: name: my-ingress annotations: nginx.ingress.kubernetes.io/use-regex: "true"spec: rules: - host: api.mydomain.com http: paths: - backend: serviceName: api servicePort: 80 - host: domain.com http: paths: - path: /web/* backend: serviceName: web servicePort: 8080 - host: backoffice.domain.com http: paths: - backend: serviceName: backoffice servicePort: 8080
只要创建了上面的Ingress后,ingress controller里会监听到从而生成对应的配置段后动态reload配置文件。
Ingress Controller部署
部署非常简单,一条命令创建即可,yml来源于 。
kubectl apply -f https://raw.githubusercontent.com/kubernetes/ingress-nginx/master/deploy/mandatory.yaml
该yaml缺少向群外暴露的方式,ingress-controller需要开启 hostNetwork: true便于暴漏ingress的80端口和其他ingress-controller的nginx.conf暴漏的端口
下面提供一个修改好验证可用的yaml:
apiVersion: v1kind: Namespacemetadata: name: ingress-nginx labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginx---kind: ConfigMapapiVersion: v1metadata: name: nginx-configuration namespace: ingress-nginx labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginx---kind: ConfigMapapiVersion: v1metadata: name: tcp-services namespace: ingress-nginx labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginx---kind: ConfigMapapiVersion: v1metadata: name: udp-services namespace: ingress-nginx labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginx---apiVersion: v1kind: ServiceAccountmetadata: name: nginx-ingress-serviceaccount namespace: ingress-nginx labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginx---apiVersion: rbac.authorization.k8s.io/v1beta1kind: ClusterRolemetadata: name: nginx-ingress-clusterrole labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginxrules: - apiGroups: - "" resources: - configmaps - endpoints - nodes - pods - secrets verbs: - list - watch - apiGroups: - "" resources: - nodes verbs: - get - apiGroups: - "" resources: - services verbs: - get - list - watch - apiGroups: - "extensions" resources: - ingresses verbs: - get - list - watch - apiGroups: - "" resources: - events verbs: - create - patch - apiGroups: - "extensions" resources: - ingresses/status verbs: - update---apiVersion: rbac.authorization.k8s.io/v1beta1kind: Rolemetadata: name: nginx-ingress-role namespace: ingress-nginx labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginxrules: - apiGroups: - "" resources: - configmaps - pods - secrets - namespaces verbs: - get - apiGroups: - "" resources: - configmaps resourceNames: # Defaults to "- " # Here: " - " # This has to be adapted if you change either parameter # when launching the nginx-ingress-controller. - "ingress-controller-leader-nginx" verbs: - get - update - apiGroups: - "" resources: - configmaps verbs: - create - apiGroups: - "" resources: - endpoints verbs: - get---apiVersion: rbac.authorization.k8s.io/v1beta1kind: RoleBindingmetadata: name: nginx-ingress-role-nisa-binding namespace: ingress-nginx labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginxroleRef: apiGroup: rbac.authorization.k8s.io kind: Role name: nginx-ingress-rolesubjects: - kind: ServiceAccount name: nginx-ingress-serviceaccount namespace: ingress-nginx---apiVersion: rbac.authorization.k8s.io/v1beta1kind: ClusterRoleBindingmetadata: name: nginx-ingress-clusterrole-nisa-binding labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginxroleRef: apiGroup: rbac.authorization.k8s.io kind: ClusterRole name: nginx-ingress-clusterrolesubjects: - kind: ServiceAccount name: nginx-ingress-serviceaccount namespace: ingress-nginx---apiVersion: apps/v1kind: Deploymentmetadata: name: nginx-ingress-controller namespace: ingress-nginx labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginxspec: replicas: 2 selector: matchLabels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginx template: metadata: labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginx annotations: prometheus.io/port: "10254" prometheus.io/scrape: "true" spec: serviceAccountName: nginx-ingress-serviceaccount hostNetwork: true containers: - name: nginx-ingress-controller image: hejianlai/nginx-ingress-controller:0.23.0 args: - /nginx-ingress-controller - --default-backend-service=$(POD_NAMESPACE)/default-http-backend - --configmap=$(POD_NAMESPACE)/nginx-configuration - --tcp-services-configmap=$(POD_NAMESPACE)/tcp-services - --udp-services-configmap=$(POD_NAMESPACE)/udp-services - --publish-service=$(POD_NAMESPACE)/ingress-nginx - --annotations-prefix=nginx.ingress.kubernetes.io securityContext: allowPrivilegeEscalation: true capabilities: drop: - ALL add: - NET_BIND_SERVICE # www-data -> 33 runAsUser: 33 env: - name: POD_NAME valueFrom: fieldRef: fieldPath: metadata.name - name: POD_NAMESPACE valueFrom: fieldRef: fieldPath: metadata.namespace ports: - name: http containerPort: 80 - name: https containerPort: 443 livenessProbe: failureThreshold: 3 httpGet: path: /healthz port: 10254 scheme: HTTP initialDelaySeconds: 10 periodSeconds: 10 successThreshold: 1 timeoutSeconds: 10 readinessProbe: failureThreshold: 3 httpGet: path: /healthz port: 10254 scheme: HTTP periodSeconds: 10 successThreshold: 1 timeoutSeconds: 10---
上面的yaml里后面详细解释我们需要关注的配置项,先来创建ingress对象试试。
测试http 7层负载
部署了官方的ingress nginx后,我部署了一个nginx的pod,为它创建了一个名为nginx的service:
apiVersion: extensions/v1beta1kind: Deploymentmetadata: name: nginxspec: template: metadata: labels: app: nginx spec: containers: - image: nginx name: nginx ports: - containerPort: 80---apiVersion: v1kind: Servicemetadata: name: nginxspec: selector: app: nginx ports: - protocol: TCP port: 80 targetPort: 80
然后创建对应的一个ingress对象来暴露集群里这个nginx的http服务:
apiVersion: extensions/v1beta1kind: Ingressmetadata: name: nginx-ingressspec: rules: - host: nginx.testdomain.com http: paths: - backend: serviceName: nginx servicePort: 80
查看ingress资源
[root@master k8s_yaml]# kubectl get ingressNAME HOSTS ADDRESS PORTS AGEapp-nginx-ingress nginx.testdomain.com 80 3d
找到ingress nginx的pod名字后通过命令查看里面nginx配置文件能找到有对应的配置段生成:
$ kubectl -n ingress-nginx exec nginx-ingress-controller-6cdcfd8ff9-t5sxl -- cat /etc/nginx/nginx.conf... ## start server nginx.testdomain.com server { server_name nginx.testdomain.com ; listen 80; set $proxy_upstream_name "-"; location / { set $namespace "default"; set $ingress_name "nginx-ingress"; set $service_name "nginx"; set $service_port "80"; set $location_path "/"; ........ ## end server nginx.testdomain.com ...
找一台非集群的Windows机器(也可以mac,主要是有图形界面且非集群内机器),设置hosts文件把域名nginx.testdomain.com
设置到对service的那个node的ip上,打开浏览器访问nginx.testdomain.com
即可发现集群内的nginx已经暴露在集群外。
注意:Ingress Controller虽然调用的是service,看起来按照nginx来理解转发是client–nginx–svc–pod; 实际上转发是client–nginx–pod,因为已经魔改了不能按照nginx的来理解,是直接负载到svc的endpoint上面的。
另外低版本的ingress nginx的args参数--default-backend-service=$(POD_NAMESPACE)/default-http-backend
,该参数指定ingress nginx的同namespace下名为default-http-backend
的service作为默认访问的时候页面,通常那个时候是创建一个404页面的的pod和对应service,如果ingress nginx启动的时候没找到这个service会无法启动,新版本不是必须了,好像也自带404页面了。
下面是default-http-backend.yaml:
apiVersion: extensions/v1beta1kind: Deploymentmetadata: name: default-http-backend labels: app: default-http-backend namespace: ingress-nginxspec: replicas: 1 template: metadata: labels: app: default-http-backend spec: terminationGracePeriodSeconds: 60 containers: - name: default-http-backend # Any image is permissable as long as: # 1. It serves a 404 page at / # 2. It serves 200 on a /healthz endpoint image: gcr.io/google_containers/defaultbackend:1.4 livenessProbe: httpGet: path: /healthz port: 8080 scheme: HTTP initialDelaySeconds: 30 timeoutSeconds: 5 ports: - containerPort: 8080 resources: limits: cpu: 10m memory: 20Mi requests: cpu: 10m memory: 20Mi---apiVersion: v1kind: Servicemetadata: name: default-http-backend namespace: ingress-nginx labels: app: default-http-backendspec: ports: - port: 80 targetPort: 8080 selector: app: default-http-backend
另外ingress也能多路径,如下:
spec: rules: - host: xxxx.xxxx.xxx http: paths: - backend: serviceName: service-index servicePort: 80 path: / - backend: serviceName: service-test-api servicePort: 80 path: /api/
如何来4层负载
我们可以看到ingress nginx的args里有这两行:
- --tcp-services-configmap=$(POD_NAMESPACE)/tcp-services - --udp-services-configmap=$(POD_NAMESPACE)/udp-services
从选项和值可以猜测出,要想代理四层(tcp/udp),得写同namespace里一个名为tcp-service
和udp-service
的两个configmap的数据 四层的话这边我们创建一个mysql的pod,来代理3306端口到集群外面,则需要写tcp-services这个configmap:
kind: ConfigMapapiVersion: v1metadata: name: tcp-services namespace: ingress-nginxdata: 3306: "default/mysql:3306"
四层写这两个ConfigMap的data即可,按照这样去写即可out_port: namespaces/svc_name:port
,要给每个ingress加一些nginx里的配置可以查看官方的annotation字段以及值(traefik同理)。
Ingress Controller高可用
这里来讨论下Ingress Controller
的高可用。
Ingress Controller到集群内的路径这部分都有负载均衡了,我们比较关注部署了Ingress Controller后,外部到它这段路怎么高可用?
上面的例子里service我使用的externalIPs,但是代理四层的时候会新加端口,需要每次人为去介入增加暴露端口?
流量从入口到Ingress Controller
的pod有下面几种方式:
-
type为
LoadBalancer
的时候手写externalIPs
很鸡肋,后面会再写文章去讲它 -
type为
LoadBalancer
的时候只有云厂商支持分配公网ip来负载均衡,LoadBalancer 公开的每项服务都将获得自己的 IP 地址,但是需要收费,且自己建立集群无法使用 -
不创建service,pod直接用hostport,效率等同于
hostNetwork
,如果不代理四层端口还好,代理了需要修改pod的template来滚动更新来让nginx bind的四层端口能映射到宿主机上 -
Nodeport
,端口不是web端口(但是可以修改Nodeport的范围改成web端口),如果进来流量负载到Nodeport上可能某个流量路线到某个node上的时候因为Ingress Controller
的pod不在这个node上,会走这个node的kube-proxy转发到Ingress Controller的pod上,多走一趟路 -
不创建service,效率最高,也能四层负载的时候不修改pod的template,唯一要注意的是
hostNetwork
下pod会继承宿主机的网络协议,也就是使用了主机的dns,会导致svc的请求直接走宿主机的上到公网的dns服务器而非集群里的dns server,需要设置pod的dnsPolicy: ClusterFirstWithHostNet
即可解决
写在最后
部署方式没多大区别开心就好。
-
DaemonSet + nodeSeletor
-
deploy设置replicas数量 + nodeSeletor + pod互斥
-
所以可以一个vip飘在拥有存活的controller的宿主机上,云上的话就用slb来负载代替vip
-
最后说说域名请求指向它,如果部署在内网或者办公室啥的,内网有dns server的话把ing的域名全部解析到ingress controller的宿主机ip上,否则要有人访问每个人设置/etc/hosts才能把域名解析来贼麻烦,如果没有dns server可以跑一个external-dns,它的上游dns是公网的dns服务器,办公网内机器的dns server指向它即可,云上的话把域名请求解析到对应ip即可
-
traefik和ingress nginx类似,不过它用go实现的
-
在一些老版本的ingress nginx的log里会一直刷找不到ingress-nginx的svc,不处理的话会狂刷log导致机器load过高,创建一个同名的svc即可解决,例如创建一个不带选择器clusterip为null的即可。非要创建port的svc的话参照下面:
apiVersion: v1kind: Servicemetadata: name: ingress-nginx namespace: ingress-nginx labels: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginxspec: type: ClusterIP ports: - name: http port: 80 targetPort: 80 protocol: TCP - name: https port: 443 targetPort: 443 protocol: TCP - name: metrics port: 10254 targetPort: 10254 protocol: TCP selector: app.kubernetes.io/name: ingress-nginx app.kubernetes.io/part-of: ingress-nginx