Kubernetes集群中的Nginx配置热更新方案
felixgis
8年前
<p>Nginx已经是互联网IT业界一个无敌的存在,作为反向代理、负载均衡、Web服务器等多种角色的扮演者,Nginx在全球各个互联网公司落地、开花和结果,Ngnix已经成为了支撑全球互联网应用的一个不可获取的组成部分。</p> <p>在我们的平台中,Nginx同样被拿来作为服务接入的最前端的反向代理,并且我们的Nginx也是作为一个Service跑在我们的Kubernetes集群中的。Ngnix背后的服务众多,服务的生生死死都要在Nginx上这些服务路由的配置中有所体现,这就要求部署在Kubernetes集群中的Nginx需要有一个合理的配置热更新方案。</p> <p>Nginx自身是支持配置热更新的,通过nginx -s reload命令可以实现这一点:</p> <pre> <code class="language-groovy"># sudo nginx -s reload # sudo tail -100f /var/log/nginx/error.log 2016/11/18 08:21:03 [notice] 31516#31516: signal process started</code></pre> <p>这也是诸多nginx热更新方案的基础。</p> <p>随着Docker容器以及容器集群/云的出现,Nginx也被Dockerize了,Docker中Nginx的配置热更新方案在 Jason Wilder 的 这篇文章 中有体现,在该方案中,你可以直接使用Jason Wilder开源的 Nginx-proxy 实现容器中Nginx的配置的热更新。但这个方案并不能直接适用于Kubernetes,而且 作者也并没有Plan support k8s 。</p> <p>在Kubernetes集群中部署的Nginx,我其实也找到了一个配置热更新的方案,这是普元的一份技术资料《 微服务动态路由实现:OpenResty与kubernetes 》中提供的,这个方案通过 OpenResty 与K8s的结合实现了配置热更新。由于我对OpenResty并不熟悉,并且我个人更希望通过Kubernetes自身的一些Feature来实现这个方案,于是我开始了我自己的探索。</p> <h3><strong>一、需求场景和方案原理</strong></h3> <p>我们要实现的就是:当Kubernetes集群中的Service发生变化时,比如新创建一个Service或删除了一个Service,这些Service在Nginx反向代理中的路由配置需要同步更新并生效。因此,这个过程的场景大致如下:</p> <ul> <li>管理员通过命令或程序通过API操作K8s集群创建或删除Service;</li> <li>监听API Server Event的某个程序获取该Event,并从API Server读取最新Service数据,重新生成/etc/nginx/conf.d/default.conf;</li> <li>/etc/nginx/conf.d/default.conf文件的变动触发文件变更事件,监听该事件的脚本调用“nginx -s reload”命令实现Nginx的配置热更新。</li> </ul> <p>针对这一需求场景,我这里给出一个实现方案,先上图:</p> <p style="text-align:center"><img src="https://simg.open-open.com/show/916940939cbc28b252a080d8c7a91c7b.png"></p> <p>简答说明一下:</p> <ul> <li>Nginx作为一个Service部署在Kubernetes集群中,可以有多个Pod副本;</li> <li>以一个nginx pod为例,该Pod中包含三个Container,分别是 init container 、nginx container和config-nginx-generator container;</li> <li>三个Container共同挂载且共享一个Pod volume,emptyDir类型即可,无需持久化的存储卷,三个Container的挂载路径均为/etc/nginx/conf.d;</li> <li>Pod启动时,init container首先启动并访问API Server,获取Service列表,按照一定条件过滤后(比如通过label的key和Value值),初始创建/etc/nginx/conf.d/default.conf。创建成功后,Container退出;</li> <li>nginx container启动,加载配置,开始提供反向代理服务,并通过inotify工具监视/etc/nginx/conf.d/default.conf文件状态变化,一般变化,就执行nginx -s reload热加载最新配置。</li> <li>config-nginx-generator container同时也启动起来,监听API Server的service变更Event,一旦有Event出现,就重新读取API Server中的Service list,并重新生成一份新的default.conf,覆盖old版本 default.conf。</li> </ul> <h3><strong>二、环境</strong></h3> <p>由于 Kubernetes 和Docker都在Active Develop的过程中,两个项目的变动都很快,因此,特定的Feature(比如k8s的init container)、操作和说明在某些版本是好用的,但对另外一些版本却是不灵光的。这里先把环境确定清楚,避免误导。</p> <pre> <code class="language-groovy">OS: Ubuntu 14.04.4 LTS Kernel:3.19.0-70-generic #78~14.04.1-Ubuntu SMP Fri Sep 23 17:39:18 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux Docker: # docker version Client: Version: 1.12.2 API version: 1.24 Go version: go1.6.3 Git commit: bb80604 Built: Tue Oct 11 17:00:50 2016 OS/Arch: linux/amd64 Server: Version: 1.12.2 API version: 1.24 Go version: go1.6.3 Git commit: bb80604 Built: Tue Oct 11 17:00:50 2016 OS/Arch: linux/amd64 Kubernetes集群:1.3.7 私有镜像仓库:阿里云镜像仓库</code></pre> <h3><strong>三、实现</strong></h3> <p><strong>1、nginx image的创建</strong></p> <p>nginx image实现了两个功能,一个自然是nginx自身了,另外一个就是监听/etc/nginx/conf.d/default.conf文件的变化,并适时调用nginx -s reload更新nginx配置。在kubernetes的源码目录kubernetes/examples下有一个例子: https-nginx ,这里面已经为我们实现了一个基于 auto-reload-nginx.sh 的Nginx image Dockerfile,我们稍作改造就可以直接使用了:</p> <pre> <code class="language-groovy">//Dockerfile FROM nginx MAINTAINER Tony Bai <bigwhite.cn@aliyun.com> COPY auto-reload-nginx.sh /home/auto-reload-nginx.sh RUN chmod +x /home/auto-reload-nginx.sh # install inotify RUN apt-get update && apt-get install -y inotify-tools</code></pre> <p>基于该Dockefile构建image:</p> <pre> <code class="language-groovy"># docker build -t xxxx/nginx # docker images REPOSITORY TAG IMAGE ID CREATED SIZE xxxx/nginx latest a1503b1c2b70 42 seconds ago 191.9 MB</code></pre> <p>官方nginx image基于debian jessie版本构建,apt-get update & install时需要耐心等待一下。</p> <p>打标签并推送到我们的阿里云私有镜像库:</p> <pre> <code class="language-groovy"># docker tag a1503b1c2b70 registry.cn-hangzhou.aliyuncs.com/xxxx/nginx # docker images REPOSITORY TAG IMAGE ID CREATED SIZE xxxx/nginx latest a1503b1c2b70 12 minutes ago 191.9 MB registry.cn-hangzhou.aliyuncs.com/xxxx/nginx latest a1503b1c2b70 12 minutes ago 191.9 MB # docker push registry.cn-hangzhou.aliyuncs.com/xxxx/nginx</code></pre> <p><strong>2、编写Pod yaml</strong></p> <p>由于init container和config-nginx-generator container在真实场景中都是要与Kubernetes的API Server交互,并生成/etc/nginx/conf.d/default.conf,这需要一个实现过程,在这里我们暂不给出两个Container的具体Dockerfile以及实现功能的实际程序,而是用两个通用docker image,并通过“手动”方式实现它们各自的功能。因此,我们在这一节中就可以给出Nginx Pod的yaml描述文件了:</p> <pre> <code class="language-groovy">//nginx-reload-on-k8s.yaml apiVersion: v1 kind: Pod metadata: name: nginx-reload-on-k8s annotations: pod.beta.kubernetes.io/init-containers: '[ { "name": "nginx-reload-on-k8s-init-1", "image": "busybox", "command": ["wget", "-O", "/etc/nginx/conf.d/index1.html", "http://www.baidu.com"], "volumeMounts": [ { "name": "conf-volume", "mountPath": "/etc/nginx/conf.d" } ] }, { "name": "nginx-reload-on-k8s-init-2", "image": "busybox", "command": ["wget", "-O", "/etc/nginx/conf.d/index2.html", "http://dict.cn"], "volumeMounts": [ { "name": "conf-volume", "mountPath": "/etc/nginx/conf.d" } ] } ]' spec: containers: - name: nginx-config-generator volumeMounts: - mountPath: /etc/nginx/conf.d name: conf-volume image: registry.cn-hangzhou.aliyuncs.com/xxxx/test:latest imagePullPolicy: IfNotPresent command: - "tail" - "-f" - "/var/log/bootstrap.log" - name: nginx-origin volumeMounts: - mountPath: /etc/nginx/conf.d name: conf-volume image: registry.cn-hangzhou.aliyuncs.com/xxxx/nginx:latest imagePullPolicy: IfNotPresent command: ["/home/auto-reload-nginx.sh"] ports: - containerPort: 80 volumes: - name: conf-volume emptyDir: {}</code></pre> <p>Yaml中,我们创建了两个init container,分别用于从baidu.com和dict.cn抓取主页,并存储于/etc/nginx/conf.d的下面备用。nginx-config-generator我们使用image xxxx/test,这就是一个基于ubuntu且安装了诸多网络工具的镜像,用于做目标镜像调试的;nginx container用的就是上面push到私有镜像仓库的那个镜像,command则是执行/home/auto-reload-nginx.sh这个脚本,从而启动nginx和通过inotify监控/etc/nginx/conf.d/default.conf文件。</p> <p>我们来创建这个Pod(注意:只有用kubectl apply命令时,init container才会被创建和执行,如果用kubectl create -f ,那么将忽略init container):</p> <pre> <code class="language-groovy"># kubectl apply -f nginx-reload-on-k8s.yaml pod "nginx-reload-on-k8s" created # kubectl get pod NAME READY STATUS RESTARTS AGE nginx-reload-on-k8s 2/2 Running 0 41s</code></pre> <p>通过describe pod/nginx-reload-on-k8s,我们能看到一些Container创建的详细信息:</p> <pre> <code class="language-groovy"># kubectl describe pod/nginx-reload-on-k8s Name: nginx-reload-on-k8s Namespace: default Node: 10.46.181.146/10.46.181.146 Start Time: Thu, 17 Nov 2016 21:39:55 +0800 Labels: <none> Status: Running IP: 172.16.57.9 ... ... Events: FirstSeen LastSeen Count From SubobjectPath Type Reason Message --------- -------- ----- ---- ------------- -------- ------ ------- 57s 57s 1 {default-scheduler } Normal Scheduled Successfully assigned nginx-reload-on-k8s to 10.46.181.146 39s 39s 1 {kubelet 10.46.181.146} spec.initContainers{nginx-reload-on-k8s-init-1} Normal Created Created container with docker id 0e21afb58eee 39s 39s 1 {kubelet 10.46.181.146} spec.initContainers{nginx-reload-on-k8s-init-1} Normal Started Started container with docker id 0e21afb58eee 56s 38s 2 {kubelet 10.46.181.146} spec.initContainers{nginx-reload-on-k8s-init-1} Normal Pulling pulling image "busybox" 39s 26s 2 {kubelet 10.46.181.146} spec.initContainers{nginx-reload-on-k8s-init-1} Normal Pulled Successfully pulled image "busybox" 26s 26s 1 {kubelet 10.46.181.146} spec.initContainers{nginx-reload-on-k8s-init-2} Normal Created Created container with docker id 85632ff73ea8 26s 26s 1 {kubelet 10.46.181.146} spec.initContainers{nginx-reload-on-k8s-init-2} Normal Started Started container with docker id 85632ff73ea8 25s 25s 1 {kubelet 10.46.181.146} spec.containers{nginx-config-generator} Normal Pulled Container image "registry.cn-hangzhou.aliyuncs.com/xxxx/test:latest" already present on machine 25s 25s 1 {kubelet 10.46.181.146} spec.containers{nginx-config-generator} Normal Created Created container with docker id 1ce8c6d8a8af 25s 25s 1 {kubelet 10.46.181.146} spec.containers{nginx-config-generator} Normal Started Started container with docker id 1ce8c6d8a8af 25s 25s 1 {kubelet 10.46.181.146} spec.containers{nginx-origin} Normal Pulled Container image "registry.cn-hangzhou.aliyuncs.com/xxxx/nginx:latest" already present on machine 25s 25s 1 {kubelet 10.46.181.146} spec.containers{nginx-origin} Normal Created Created container with docker id 0c692ec28acd 25s 25s 1 {kubelet 10.46.181.146} spec.containers{nginx-origin} Normal Started Started container with docker id 0c692ec28acd ... ...</code></pre> <p>可以看到四个container依次被pull and create。</p> <h3><strong>四、测试</strong></h3> <p>现在我们就来测试一下nginx的reload。</p> <p>之前的两个init container分别在/etc/nginx/conf.d下创建了index1.html和index2.html,我们就用这两个文件分别作为配置变更前和变更后的首页。</p> <p>注意:这时我们还没有/etc/nginx/conf.d/default.conf文件,我们在Pod内访问localhost:80将会得到失败结果:</p> <pre> <code class="language-groovy"># curl localhost:80 curl: (7) Failed to connect to localhost port 80: Connection refused</code></pre> <p>我们进入nginx-config-generator,创建/etc/nginx/conf.d/default.conf文件,与此同时,通过docker logs -f 监控nginx-origin容器的日志:</p> <pre> <code class="language-groovy">//default.conf server { listen 80; server_name localhost; #charset koi8-r; #access_log /var/log/nginx/log/host.access.log main; location / { root /etc/nginx/conf.d; index index1.html index1.htm; } #error_page 404 /404.html; # redirect server error pages to the static page /50x.html # error_page 500 502 503 504 /50x.html; location = /50x.html { root /usr/share/nginx/html; } }</code></pre> <p>我们把/etc/nginx/conf.d/index1.html作为服务站点的首页了。文件创建完毕后,我们同时就可以从nginx-origin容器的日志能看到如下内容:</p> <pre> <code class="language-groovy">At 14:07 on 17/11/16, config file update detected. 2016/11/17 14:07:25 [notice] 20#20: signal process started</code></pre> <p>我们再从Pod中访问localhost:80(注意:Pod中的多个container共享network namespace,通过localhost就可以进行互访):</p> <pre> <code class="language-groovy">root@nginx-reload-on-k8s:/etc/nginx# curl localhost:80 <!DOCTYPE html> <!--STATUS OK--><html> <head><meta http-equiv=content-type content=text/html;charset=utf-8><meta http-equiv=X-UA-Compatible content=IE=Edge><meta content=always name=referrer><link rel=stylesheet type=text/css href=http://s1.bdstatic.com/r/www/cache/bdorz/baidu.min.css><title>百度一下,你就知道</title></head> .... </html></code></pre> <p>我们顺利得到index1.html的内容,这说明配置实时生效了。</p> <p>我们再来“触发”一次配置变更。我们将default.conf中的:</p> <pre> <code class="language-groovy">location / { root /etc/nginx/conf.d; index index1.html index1.htm; }</code></pre> <p>改为:</p> <pre> <code class="language-groovy">location / { root /etc/nginx/conf.d; index index2.html index2.htm; }</code></pre> <p>保存!</p> <p>从nginx-origin容器日志可以看到如下输出:</p> <pre> <code class="language-groovy">At 14:17 on 17/11/16, config file update detected. 2016/11/17 14:17:46 [notice] 32#32: signal process started</code></pre> <p>在Pod中再次访问站点首页:</p> <pre> <code class="language-groovy"># curl localhost:80 <!DOCTYPE HTML> <html> <head> <meta name="renderer" content="webkit"/> <meta http-equiv="X-UA-Compatible" content="IE=EmulateIE7" /> <meta http-equiv="Content-Type" content="text/html; charset=utf-8" /> <title>海词词典_在线词典_在线翻译_海量正版权威词典官方网站</title> ... ...</code></pre> <p>可以看到配置更新成功,首页换成了dict.cn的首页。</p> <h3><strong>五、测试</strong></h3> <p>通过上述这些“手动”的触发和测试,可以看出这个方案是可行的。并且我们可以看出,这个方案是有一些好处的:</p> <ul> <li>不需要依赖外部持久化存储卷;</li> <li>通过k8s api server获取当前所有 service列表,通过service label来过滤,无需依赖额外的redis server或etcd服务;</li> </ul> <p>剩下的就是具体init container以及config-generator的实现了。这个留给我以及大家后续去完成^_^。</p> <p>© 2016,bigwhite. 版权所有.</p> <p> </p> <p>来自:http://tonybai.com/2016/11/17/nginx-config-hot-reloading-approach-for-kubernetes-cluster/</p> <p> </p>