HAProxy高可用负载均衡集群部署
基本信息:
系统平台:VMware WorkStation
系统版本: CentOS Linux release 7.2.1511 (Core)
内核版本: 3.10.0-327.el7.x86_64
集群架构:
前端:HAProxy
1、虚拟FQDN:www.simpletime.net
2、VIP:192.168.39.1;DIP:172.16.39.50
3、调度服务器:Varnish1、Varnish2
4、调度算法:URL_Hash_Consistent
5、集群统计页:172.16.39.50:9091/simpletime?admin
缓存服务器:Varnish
1、VarnishServer1:172.16.39.14:9527
2、VarnishServer2:172.16.39.15:9527
3、开启健康状态探测,提供高可用
4、负载均衡后端Web服务器组
5、动静分离后端服务器,并动静都提供负载均衡效果
后端服务器:
StaticServer1:172.16.39.14:80
StaticServer2:172.16.39.15:80
DynamicServer1:172.16.39.151
DynamicServer2:172.16.39.152
Mysql服务器:
MysqlServer:172.16.39.150
思考:
1、负载均衡动静分离后,会话如何保持?
2、负载均衡动静分离后,存储如何解决?
3、该方案适用于什么样的场景?
4、该方案缺陷有哪些?
5、如何改进?
一、部署HAProxy
1、安装HAProxy ~]# yum install HAProxy 2、配置HAProxy #--------------------------------------------------------------------- # main frontend which proxys to the backends #--------------------------------------------------------------------- frontend web *:80 #acl url_static path_beg -i /static /p_w_picpaths /javascript /stylesheets #acl url_static path_end -i .jpg .gif .png .css .js .html .txt .htm #acl url_dynamic path_begin -i .php .jsp #default_backend static_srv if url_static #use_backend dynamic_srv if url_dynamic use_backend varnish_srv #--------------------------------------------------------------------- # round robin balancing between the various backends #--------------------------------------------------------------------- backend varnish_srv balance uri #使用基于URL的一致性哈希调度算法 hash-type consistent server varnish1 172.16.39.14:9527 check server varnish2 172.16.39.15:9527 check listen stats #开启HAProxy图形化Web管理功能 bind :9091 stats enable stats uri /simpletime?admin stats hide-version stats auth admin:abc.123 stats admin if TRUE 3、启动服务 ~]# systemctl start haproxy ~]# systemctl status haproxy #查看状态 ~]# ss -tnlp #查看80和9091端口是否启用 ~]# systemctl enable haproxy #设置开机启动
二、部署Varnish,两台配置一致(172.16.39.14|15)
1、安装及配置 ~]# yum install varnish -y ~]# vim /etc/varnish/varnish.params VARNISH_LISTEN_PORT=9527 #更改默认端口 ~]# systemctl start varnish ~]# systemctl enable varnish ~]# vim /etc/varnish/default.vcl vcl 4.0; ##############启用负载均衡模块############### import directors; ################定义Purge-ACL控制####################### acl purgers { "127.0.0.1"; "172.16.39.0"/16; } # Default backend definition. Set this to point to your content server. ##############配置健康状态探测############## probe HE { #静态检测 .url = "/health.html"; #指定检测URL .timeout = 2s; #探测超时时长 .window = 5; #探测次数 .threshold = 2; #探测次数成功多少次才算健康 .initial = 2; #Varnish启动探测后端主机2次健康后加入主机 .interval = 2s; #探测间隔时长 .expected_response = 200; #期望状态响应码 } probe HC { #动态监测 .url = "/health.php"; .timeout = 2s; .window = 5; .threshold = 2; .initial = 2; .interval = 2s; .expected_response = 200; } #############添加后端主机################ backend web1 { .host = "172.16.39.151:80"; .port = "80"; .probe = HC; } backend web2 { .host = "172.16.39.152:80"; .port = "80"; .probe = HC; } backend app1 { .host = "172.16.39.14:80"; .port = "80"; .probe = HE; } backend app2 { .host = "172.16.39.15:80"; .port = "80"; .probe = HE; } #############定义负载均衡及算法############### sub vcl_init { new webcluster = directors.round_robin(); webcluster.add_backend(web1); webcluster.add_backend(web2); new appcluster = directors.round_robin(); appcluster.add_backend(app1); appcluster.add_backend(app2); } ################定义vcl_recv函数段###################### sub vcl_recv { #####ACL未授权,不允许PURGE,并返回405##### if (req.method == "PURGE") { if(!client.ip ~ purgers){ return(synth(405,"Purging not allowed for" + client.ip)); } return (purge); } #####添加首部信息,使后端服务记录访问者的真实IP # if (req.restarts == 0) { # set req.http.X-Forwarded-For = req.http.X-Forwarded-For + ", " + client.ip; # } else { # set req.http.X-Forwarded-For = client.ip; # } # set req.backend_hint = webcluster.backend(); # set req.backend_hint = appcluster.backend(); #注:因为Varnish不是一级代理,配置forward只能取到上级代理IP,而上级代理IP,本身就包含在HAProxy发送过来的Forward里,所以没必要配置,而后端服务器只要日志格式有启用记录Forward信息,并且上级代理没有做限制,那么,就能获取到客户端真实IP; #####动静分离##### if (req.url ~ "(?i)\.(php|asp|aspx|jsp|do|ashx|shtml)($|\?)") { set req.backend_hint = appcluster.backend(); } #####不正常的请求不缓存##### if (req.method != "GET" && req.method != "HEAD" && req.method != "PUT" && req.method != "POST" && req.method != "TRACE" && req.method != "OPTIONS" && req.method != "PATCH" && req.method != "DELETE") { return (pipe); } #####如果请求不是GET或者HEAD,不缓存##### if (req.method != "GET" && req.method != "HEAD") { return (pass); } #####如果请求包含Authorization授权或Cookie认证,不缓存##### if (req.http.Authorization || req.http.Cookie) { return (pass); } #####启用压缩,但排除一些流文件压缩##### if (req.http.Accept-Encoding) { if (req.url ~ "\.(bmp|png|gif|jpg|jpeg|ico|gz|tgz|bz2|tbz|zip|rar|mp3|mp4|ogg|swf|flv)$") { unset req.http.Accept-Encoding; } elseif (req.http.Accept-Encoding ~ "gzip") { set req.http.Accept-Encoding = "gzip"; } elseif (req.http.Accept-Encoding ~ "deflate") { set req.http.Accept-Encoding = "deflate"; } else { unset req.http.Accept-Encoding; } } return (hash); } ####################定义vcl_pipe函数段################# sub vcl_pipe { return (pipe); } sub vcl_miss { return (fetch); } ####################定义vcl_hash函数段################# sub vcl_hash { hash_data(req.url); if (req.http.host) { hash_data(req.http.host); } else { hash_data(server.ip); } if (req.http.Accept-Encoding ~ "gzip") { hash_data ("gzip"); } elseif (req.http.Accept-Encoding ~ "deflate") { hash_data ("deflate"); } } ##############设置资源缓存时长################# sub vcl_backend_response { if (beresp.http.cache-control !~ "s-maxage") { if (bereq.url ~ "(?i)\.(jpg|jpeg|png|gif|css|js|html|htm)$") { unset beresp.http.Set-Cookie; set beresp.ttl = 3600s; } } } ################启用Purge##################### sub vcl_purge { return(synth(200,"Purged")); } ###############记录缓存命中状态############## sub vcl_deliver { if (obj.hits > 0) { set resp.http.X-Cache = "HIT from " + req.http.host; set resp.http.X-Cache-Hits = obj.hits; } else { set resp.http.X-Cache = "MISS from " + req.http.host; } unset resp.http.X-Powered-By; unset resp.http.Server; unset resp.http.Via; unset resp.http.X-Varnish; unset resp.http.Age; } 2、加载配置,因为还没有配置后端应用服务器,可以看到后端主机健康检测全部处于Sick状态 ~]# varnishadm -S /etc/varnish/secret -T 127.0.0.1:6082 varnish> vcl.load conf1 default.vcl 200 VCL compiled. varnish> vcl.use conf1 200 VCL 'conf1' now active varnish> backend.list 200 Backend name Refs Admin Probe web1(172.16.39.151,,80) 15 probe Sick 0/5 web2(172.16.39.152,,80) 15 probe Sick 0/5 app1(172.16.39.14,,80) 15 probe Sick 0/5 app2(172.16.39.15,,80) 15 probe Sick 0/5
三、部署Mysql(172.16.39.150)
~]# yum install mariadb.server ~]# rpm -qe mariadb-server mariadb-server-5.5.44-2.el7.centos.x86_64 ~]# vim /etc/my.cnf #数据库基本优化 [mysqld] innodb_file_per_table = ON skip_name_resolve = ON ~]# mysql #创建wordpress数据库并授权该数据库用户 > create database wwwdb; > grant all on wwwdb.* to www@'172.16.39.%' identified by "abc.123"; > exit
四、部署NFS文件系统
1、后端所有主机安装服务 ~]# yum install nfs-utils 2、动态资源主机172.16.39.152设为动态web数据共享服务器 DynamicServer2 ~]# vim /etc/exports /data/web/ 172.16.39.151/16(rw,sync) #rw=可读写,sync=内存及硬盘同步写入数据 3、静态主机172.16.39.15设为静态web数据共享服务器 StaticServer2 ~]# vim /etc/exports /data/web/ 172.16.39.14/16(rw,sync) #rw=可读写,sync=内存及硬盘同步写入数据 ~]# systemctl start nfs-server #启动服务 DynamicServer2 ~]# exportfs -avr #重载配置 exporting 172.16.39.151/16:/data/web StaticServer2 ~]# exportfs -avr #重载配置 exporting 172.16.39.14/16:/data/web 4、两台服务端设为开机启动 ~]# systemctl enable nfs-server 5、客户端同步,动态主机挂载动态服务器共享,静态主机挂载静态服务器共享 ~]# showmount -e 172.16.39.152 Export list for 172.16.39.152: /data/web 172.16.39.151/16 ~]# mount -t nfs 172.16.39.15:/data/web /data/web
五、部署后端主机(注意:已经部署了NFS文件系统)
1、安装及配置(DynamicServer2:172.16.39.152) ~]# yum install nginx php-fpm php-mysql -y ~]# mkdir /data/web/www -pv ~]# vim /etc/nginx/conf.d/www.simple.com.conf server { listen 80; root /data/web/www; server_name www.simple.com; index index.html index.htm index.php; location ~ [^/]\.php(/|$) { try_files $uri = 404; fastcgi_pass 127.0.0.1:9000; fastcgi_index index.php; include fastcgi.conf; #access_log_bypass_if ($uri = '/health.php'); } } 备注:access_log_bypass_if 需添加日志过滤模块,本文主要实现过滤健康状态检测信息; ~]# systemctl start nginx php-fpm 2、部署wordpress应用 ~]# unzip wordpress-4.3.1-zh_CN.zip ~]# mv wordpress/* /data/web/www/ www]# cp wp-config{-sample,}.php www]# vim wp-config.php define('DB_NAME', 'wwwdb'); define('DB_USER', 'www'); define('DB_PASSWORD', 'abc.123'); define('DB_HOST', '172.16.39.150'); 3、设置facl权限 ~]# id apache ~]# setfacl -m u:apache:rwx /data/web/www 4、拷贝web数据至StaticServer2,另两台后端主机挂载的是两台NFS服务端的数据文件,web数据数完成 ~]# tar -jcvf web.tar.gz /data/web/www ~]# scp web.tar.gz 172.16.39.15: ~]# setfacl -m u:apache:rwx /data/web/www StaticServer2 ~]# tar -xf web.tar.gz -C /data/web 5、创建动静资源主机组Varnish健康状态探测页面 DynamicServer2~]# echo "DynamicServer is Health.
> /data/web/www/health.php StaticServer2~]# echo "StaticServer is Health.
" > /data/web/www/health.html 6、在Varnish主机上查看健康状态(172.16.39.14|15,也就是StaticServer主机) StaticServer2~]# varnishadm -S /etc/varnish/secret -T 127.0.0.1:6082 varnish> backend.list #后端Web主机正常 200 Backend name Refs Admin Probe web1(172.16.39.151,,80) 15 probe Healthy 5/5 web2(172.16.39.152,,80) 15 probe Healthy 5/5 app1(172.16.39.14,,80) 15 probe Healthy 5/5 app2(172.16.39.15,,80) 15 probe Healthy 5/5
7、web访问172.16.39.50完成wordpress配置
8、创建图文测试,发现动静分离下,图片上传后无法显示
9、分析原因,动静分离后,用户可以上传图片至动态资源主机上,但静态资源却没有该图片,初步解决方案:使用文件同步软件,实时同步图片等流格式资源。
六、部署Rsync+inodify,实现静态资源主机同步上传资源图片目录,解决图片不能显示问题
1、将StaticServer2配置为备份端,同步动态资源资源DynamicServer2下的web图片上传目录文件 ~]# rpm -qa rsync ~]# vim /etc/rsyncd.conf #尾行追加 uid = root #指定执行备份的用户及组权限为root用户 gid = root user chroot = no #关闭chroot max connections = 200 #最大连接数 timeout = 600 pid file = /var/run/rsyncd.pid lock file = /var/run/rsyncd.lock log file = /var/log/rsyncd.log [web] path=/data/web/www/wp-content/uploads ignore errors read only = no list = no hosts allow = 172.16.39.0/255.255.0.0 #只允许39网段进行同步 auth users = rsync #认证的用户名 secrets file = /etc/rsyncd.password #指定密码文件 2、创建rsync同步用户密码文件 ~]# echo “rsync:abc.123” > /etc/rsyncd.password ~]# chmod 600 /etc/rsyncd.password 3、启动服务 ~]# systemctl start rsyncd ~]# ss -tnlp |grep rsync #查看是否开启873端口 ~]# systemctl enable rsyncd #设置rsync开机启动 4、部署Rsync同步端DynamicServer2,实时同步本机web图片数据至StaticServer2,并保持以DynamicServer2为主两台服务器web图片资源一致性。 ~]# yum install rsync inodify-tools inodify-tools-devel 如果yum源没有找到inodify,就直接下载源码包进行编译 ~]# tar -xf inodify-tools-3.13.tar.gz ~]# ./configure --prefix=/usr/local/inotify ~]# ]# make -j 2 && make install ~]# ln -sv /usr/local/inotify/bin/inotifywait /usr/bin/ ~]# echo “abc.123” > /etc/rsyncd.password ~]# chmod 600 /etc/rsyncd.password #创建实时同步脚本 ~]# vim /etc/rsync.sh #!/bin/bash host=172.16.39.15 #数据备份主机 src=/data/web/www/wp-content/uploads/ #想要备份的目录 des=web user=rsync /usr/bin/rsync -vzrtopg --delete --progress --password-file=/etc/rsyncd.password $src $user@$host::$des /usr/bin/inotifywait -mrq --timefmt '%d/%m/%y %H:%M' --format '%T %w%f%e' -e modify,delete,create,attrib $src \ | while read files do /usr/bin/rsync -vzrtopg --delete --progress --password-file=/etc/rsyncd.password $src $user@$host::$des echo "${files} was rsynced" >>/tmp/rsync.log 2>&1 done 5、添加开机自动运行脚本 ~]# chmod 764 /etc/rsync.sh ~]# echo "/usr/bin/bash /etc/rsync.sh &" >> /etc/rc.local #开机后台自动运行该脚本 ~]# chmod 750 /etc/rc.d/rc.local #Centos7,rc.local没有执行权限,要给予权限才能执行 ~]# sh /etc/rsync.sh & #执行同步脚本
七、测试
1、图文博客发布
2、PURGE清空Varnish缓存测试 ~]# curl -X PURGE www.simpletime.net/200 Purged Error 200 Purged
Purged
Guru Meditation:
XID: 100925448
Varnish cache server
3、压力测试 #存在缓存服务器,单机测试只做参考 ~]# ab -c 20000 -n 20000 -r www.simpletime.net/ Requests per second: 670.86 [#/sec] (mean) Time per request: 29812.421 [ms] (mean) Time per request: 1.491 [ms] (mean, across all concurrent requests) ~]# ab -c 20000 -n 20000 -r www.simpletime.net/ Requests per second: 521.78 [#/sec] (mean) Time per request: 38330.212 [ms] (mean) Time per request: 1.917 [ms] (mean, across all concurrent requests) ~]# ab -c 20000 -n 20000 -r www.simpletime.net/ Requests per second: 521.78 [#/sec] (mean) Time per request: 38330.212 [ms] (mean) Time per request: 1.917 [ms] (mean, across all concurrent requests) 4、HAProxy调度器挂了,整个架构崩溃... 5、Varnish挂一台,没事,有冗余,高可用实现,另一台Varnish压力增大,尽快修复故障机器即可。 6、后端主机,理论上动静态群集组能冗余各组一台服务器故障,可因为NFS文件系统懒人配置,NFS客户端挂了没事,NFS服务端挂了,GameOver... 7、文件服务器NFS服务端挂了,(NFS服务器没有做冗余)GameOver... 8、Rsync+Inodify文件实时同步系统挂了,服务器还能访问,只是图片等流格式文件不显示了... 9、MysqlServer挂了,无冗余,(GameOver)... 10、后端应用服务器Nginx日志示例: ~]# tail /var/log/nginx/access.log 172.16.39.15(VarnishIP) - - [26/Aug/2016:17:09:56 +0800] "POST /wp-admin/admin-ajax.php HTTP/1.1" 200 32 "http://172.16.39.50/wp-admin/post-new.php" "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/49.0.2623.110 Safari/537.36" "172.16.39.5(客户端IP), 172.16.39.50(HAProxyIP)"
八、总结
1、架构优点:
(1)HAProxy为Varnish缓存服务器提供了高可用的负载均衡,其使用的算法,也为Varnish提升了缓存命中率;
(2)HAProxy提供Web图形化管理界面,节约了学习成本;
(3)Varnish的高效缓存机制极大的提升了Web应用的性能,降低了Web服务器的压力;
(4)Varnish给后端Web服务器提供了高可用负载均衡,并使用了动态分离技术,保障了后端主机的冗余,显著提升了其性能;
(5)NFS+Rsync+inodify基本解决了该架构存储的问题,该架构文件存储因服务器有限,并没有设计好,储存服务器因有单独的网络存储服务器来提供;
2、冗余不足:
(1)Nginx负载均衡调度器,没有冗余能力,易出现单点故障。
解决方案:增加服务器,使用Keepalive做高可用。
(2)MysqlServer,没有冗余能力。
解决方案:主从复制、MHA。
(3)NFS文件服务,没有冗余能力。
解决方案:可以增加服务器,使用rsync+inodify实现数据实时同步,再用keepalive做高可用。
3、性能瓶颈:
(1)NFS文件系统,因网络IO能力与磁盘本身性能,多台主机同时挂载执行读取读写操作,势必带来性能下降,根据木桶原理,此短板已成为该群集方案性能瓶颈。
(2)在大量的读写访问下,数据库的压力会非常大,从而影响性能。
解决方案:数据库主从复制,读写分离;
友情链接:
1、Nginx日志过滤模块官网下载链接
https://github.com/cfsego/ngx_log_if/
2、Varnish4.0官方帮助指南
http://www.varnish-cache.org/docs/4.0/users-guide/
以上均为个人观点,本架构还有许多不足之处,仅作学习交流之用;