之前详细介绍了haproxy的基础知识点, 下面记录下Haproxy Heartbeat高可用web集群方案实现过程, 以加深理解.
架构草图如下:
1) 基本环境准备 (centos6.9系统)
代码语言:javascript复制172.16.60.208(eth0) HA主节点(ha-master) haproxy,heartbeat
172.16.60.207(eth0) HA备节点(ha-slave) haproxy,heartbeat
172.16.60.229 VIP地址
172.16.60.204(eth0) 后端节点1(rs-204) nginx/tomcat
172.16.60.205(eth0) 后端节点2(rs-205) nginx/tomcat
1) 关闭防火墙和selinux (四台节点机都操作)
[root@ha-master ~]# /etc/init.d/iptables stop
[root@ha-master ~]# setenforce 0
[root@ha-master ~]# vim /etc/sysconfig/selinux
SELINUX=disabled
2) 设置主机名和绑定hosts (两台HA节点机器都操作)
主节点操作
[root@ha-master ~]# hostname ha-master
[root@ha-master ~]# vim /etc/sysconfig/network
HOSTNAME=ha-master
[root@ha-master ~]# vim /etc/hosts
172.16.60.208 ha-master
172.16.60.207 ha-slave
备节点操作
[root@ha-slave ~]# hostname ha-slave
[root@ha-slave ~]# vim /etc/sysconfig/network
HOSTNAME=ha-slave
[root@ha-slave ~]# vim /etc/hosts
172.16.60.208 ha-master
172.16.60.207 ha-slave
2) 安装后端两个realserver节点的web环境 (即172.16.60.204/205两台机器都要安装nginx)
代码语言:javascript复制采用yum方式在两台realserver节点上安装nginx (先安装nginx的yum源)
[root@rs-204 ~]# rpm -ivh http://nginx.org/packages/centos/6/noarch/RPMS/nginx-release-centos-6-0.el6.ngx.noarch.rpm
[root@rs-204 ~]# yum install -y nginx
rs-204的nginx配置
[root@rs-204 ~]# cd /etc/nginx/conf.d/
[root@rs-204 conf.d]# cat default.conf
[root@rs-204 conf.d]# >/usr/share/nginx/html/index.html
[root@rs-204 conf.d]# vim /usr/share/nginx/html/index.html
this is test page of realserver01:172.16.60.204
[root@rs-204 conf.d]# /etc/init.d/nginx start
Starting nginx: [ OK ]
[root@rs-204 conf.d]# lsof -i:80
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
nginx 31944 root 6u IPv4 91208 0t0 TCP *:http (LISTEN)
nginx 31945 nginx 6u IPv4 91208 0t0 TCP *:http (LISTEN)
rs-205的nginx配置
[root@rs-205 src]# cd /etc/nginx/conf.d/
[root@rs-205 conf.d]# cat default.conf
[root@rs-205 conf.d]# >/usr/share/nginx/html/index.html
[root@rs-205 conf.d]# vim /usr/share/nginx/html/index.html
this is test page of realserver02:172.16.60.205
[root@rs-205 conf.d]# /etc/init.d/nginx start
Starting nginx: [ OK ]
[root@rs-205 conf.d]# lsof -i:80
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
nginx 20839 root 6u IPv4 289527645 0t0 TCP *:http (LISTEN)
nginx 20840 nginx 6u IPv4 289527645 0t0 TCP *:http (LISTEN)
访问http://172.16.60.204/, 访问结果为"this is test page of realserver01:172.16.60.204"
访问http://172.16.60.205/, 访问结果为"this is test page of realserver02:172.16.60.205"
3) 安装配置Haproxy (两台HA节点机进行同样操作)
代码语言:javascript复制1) 先安装haproxy
[root@ha-master ~]# yum install gcc gcc-c make openssl-devel kernel-devel
[root@ha-master ~]# cd /usr/local/src/ #下载haproxy软件到/usr/local/src目录下
[root@ha-master src]# ls haproxy-1.8.12.tar.gz
haproxy-1.8.12.tar.gz
[root@ha-master src]# tar -zvxf haproxy-1.8.12.tar.gz
[root@ha-master src]# cd haproxy-1.8.12
[root@ha-master haproxy-1.8.12]# make TARGET=linux26 CPU=x86_64 PREFIX=/usr/local/haprpxy USE_OPENSSL=1 ADDLIB=-lz
参数说明:
TARGET=linux26 #使用 uname -r 查看内核,如:2.6.32-642.el6.x86_64,此时该参数就为linux26
CPU=x86_64 #使用 uname -r 查看系统信息,如 x86_64 GNU/Linux,此时该参数就为 x86_64
PREFIX=/usr/local/haprpxy #haprpxy 安装路径
[root@ha-master haproxy-1.8.12]# ldd haproxy | grep ssl
libssl.so.10 => /usr/lib64/libssl.so.10 (0x00000031d0400000)
[root@ha-master haproxy-1.8.12]# make install PREFIX=/usr/local/haproxy
[root@ha-master haproxy-1.8.12]# mkdir -p /usr/local/haproxy/conf
[root@ha-master haproxy-1.8.12]# mkdir -p /etc/haproxy
[root@ha-master haproxy-1.8.12]# cp /usr/local/src/haproxy-1.8.12/examples/option-http_proxy.cfg /usr/local/haproxy/conf/haproxy.cfg
[root@ha-master haproxy-1.8.12]# ln -s /usr/local/haproxy/conf/haproxy.cfg /etc/haproxy/haproxy.cfg
[root@ha-master haproxy-1.8.12]# cp -r /usr/local/src/haproxy-1.8.12/examples/errorfiles /usr/local/haproxy/errorfiles
[root@ha-master haproxy-1.8.12]# ln -s /usr/local/haproxy/errorfiles /etc/haproxy/errorfiles
[root@ha-master haproxy-1.8.12]# mkdir -p /usr/local/haproxy/log
[root@ha-master haproxy-1.8.12]# touch /usr/local/haproxy/log/haproxy.log
[root@ha-master haproxy-1.8.12]# ln -s /usr/local/haproxy/log/haproxy.log /var/log/haproxy.log
[root@ha-master haproxy-1.8.12]# cp /usr/local/src/haproxy-1.8.12/examples/haproxy.init /etc/rc.d/init.d/haproxy
[root@ha-master haproxy-1.8.12]# chmod x /etc/rc.d/init.d/haproxy
[root@ha-master haproxy-1.8.12]# chkconfig haproxy on
[root@ha-master haproxy-1.8.12]# ln -s /usr/local/haproxy/sbin/haproxy /usr/sbin
2) haroxy.cfg文件进行负载配置
[root@ha-master haproxy-1.8.12]# cd /usr/local/haproxy/conf/
[root@ha-master conf]# cp haproxy.cfg haproxy.cfg.bak
[root@ha-master conf]# > haproxy.cfg
[root@ha-master conf]# vim haproxy.cfg
global
log 127.0.0.1 local3 info
maxconn 65535
chroot /usr/local/haproxy
uid 99
gid 99
daemon
defaults
log global
mode http
retries 3
option redispatch
stats uri /haproxy
stats refresh 30s
stats realm haproxy-status
stats auth admin:dxInCtFianKtL]36
stats hide-version
maxconn 65535
timeout connect 5000
timeout client 50000
timeout server 50000
frontend http-in
mode http
maxconn 65535
bind :80
log global
option httplog
option httpclose
acl is_01 hdr_beg(host) www.kevin.com
use_backend web-server if is_01
backend web-server
mode http
balance roundrobin
cookie SERVERID insert indirect nocache
option httpclose
option forwardfor
server web01 172.16.60.204:80 weight 1 cookie 3 check inter 2000 rise 2 fall 5
server web02 172.16.60.205:80 weight 1 cookie 4 check inter 2000 rise 2 fall 5
3) 配置HAProxy日志
[root@ha-master conf]# vim /etc/rsyslog.conf
.......
$ModLoad imudp #取消注释 ,这一行不注释,日志就不会写
$UDPServerRun 514 #取消注释 ,这一行不注释,日志就不会写
.......
local3.* /var/log/haproxy.log #这一行必须要写,因为在haproxy.cfg里global全局定义好的日志级别
[root@ha-master conf]# vim /etc/sysconfig/rsyslog
SYSLOGD_OPTIONS="-r -m 0" #接收远程服务器日志
重启syslog日志服务
[root@ha-master conf]# service rsyslog restart
Shutting down system logger: [ OK ]
Starting system logger: [ OK ]
4) 设置haproxy负载均衡的最大并发连接数
查看内核
[root@ha-master conf]# sysctl -a | grep file
fs.file-nr = 992 0 386459
fs.file-max = 386459
查看应用层面的需求
[root@ha-master conf]# cat /usr/local/haproxy/conf/haproxy.cfg
global #全局参数设置
maxconn 65535 #设置最大连接数
更改系统层面
[root@ha-master conf]# vim /etc/security/limits.conf #最后一行增加
* - nofile 65535
5) 重启两台HA机器的haproxy
[root@ha-master conf]# /etc/init.d/haproxy start
Starting haproxy: [ OK ]
[root@ha-master conf]# ps -ef|grep haproxy
nobody 13080 1 0 16:43 ? 00:00:00 /usr/sbin/haproxy -D -f /etc/haproxy/haproxy.cfg -p /var/run/haproxy.pid
root 13083 11940 0 16:43 pts/0 00:00:00 grep haproxy
[root@ha-master conf]# lsof -i:80
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
haproxy 13080 nobody 4u IPv4 428975 0t0 TCP *:http (LISTEN)
将www.kevin.com域名解析到两个HA节点上, 即172.16.60.208 和 172.16.60.207上
接着访问http://www.kevin.com/, 则发现访问结果是"this is test page of realserver01:172.16.60.204" , 不断刷新, 访问结果也是这个.
只有当172.16.60.204这个节点的nginx挂了, 访问结果才变成"this is test page of realserver02:172.16.60.205", 即请求转发到正常的realserver节点上.
从haproxy.cfg文件中可以看出, 虽然配置了"balance roundrobin"这个选项, 即客户端每一次访问, 都跳转到后端不同的服务器上. 但是并没有生效!
因为又配置了"cookie SERVERID insert indirect nocache", 即保持客户端session会话同步的配置, 所以客户端请求会一直转发到同一个realserver节点上,直至
这个节点发生故障才会转发到另外正常的节点上.
把"cookie SERVERID insert indirect nocache" 这个配置去掉或注释掉, 再次访问http://www.kevin.com/, 就会发现每刷新一次, 请求都会转发到不同的realserver
节点上, 即"balance roundrobin" 配置生效!
访问http://www.kevin.com/haproxy, 输入haproxy.cfg文件中配置的用户名和密码admin:dxInCtFianKtL]36, 即可打开haproxy监控页面
从上图可以看出, 此时监控的后端两个realserver节点的服务都是OK的(配置文件中定义的web01和web02此时都是绿色状态)。 现在尝试关闭rs-205的nginx服务, 刷新http://www.kevin.com/haproxy监控页面, 发现web02变成红色,即此时该节点服务是故障状态!然后重启rs-205的nginx服务,再次刷出监控页面, 发现web02就又恢复到正常的绿色状态了!
4) 安装配置Heartbeat (两台HA节点机进行同样操作)