最近用HAproxy+keepalived+mysql复制测试高可用性Linux系统集群。
   HAProxy是一款免费的提供高可用性、负载均衡以及基于TCP（第四层）和HTTP（第七层）应用的代理软件，借助HAProxy可以快速并且可靠的提供基于TCP和HTTP应用的代理解决方案。
   Keepalived主要作用是LoadBalance master和LoadBalance backup之间的健康检查，实现故障转换。
   Mysql Replication主要作用是提高mysql并处理数据的能力以及实现容灾备份的作用。

项目拓扑图：

Haproxy服务器及其IP地址规划：

项目实施：

1.Mysql主从复制配置。

1)在master服务器上创建mysql用户（授权复制账户）。
grant replication slave on *.* to'rep'@'192.168.1.244'identified by 'rep123';
2)编辑master服务器的mysql配置文件my.cnf。
server-id = 1 //指定服务器的ID
log-bin = mysql-bin //开启二进制日志
binlog-ignore = mysql //忽略mysql和information_schema 数据库
binlog-ignore = information_schema
binlog-do-db = blog    //同步数据库，默认同步所有数据库
3)查看master状态。
mysql> show master status;
+---------------+----------+--------------+--------------------------+
| File          | Position | Binlog_Do_DB | Binlog_Ignore_DB         |
+---------------+----------+--------------+--------------------------+
| binlog.022343 |   339244 | blog         | mysql,information_schema |
+---------------+----------+--------------+--------------------------+
1 row in set (0.00 sec)
4)在slave端创建数据库blog,导出master端的blog库，导入到此库，并修改mysql主配置文件my.cnf server-id = 2 重启mysql数据库。
mysql> change master to-> master_host='192.168.1.243',-> master_port=3306,-> master_user='rep',-> master_password='rep123',-> master_log_file='binlog.022343',-> master_log_pos=339244;
5)mysql主从同步测试，show slave status\G;能看到Slave_IO_Running和Slave_SQL_Running都为YES即可。
mysql> show slave status\G;
*************************** 1. row ***************************Slave_IO_State: Waiting for master to send eventMaster_Host: 192.168.1.243Master_User: repMaster_Port: 3306Connect_Retry: 60Master_Log_File: binlog.022343Read_Master_Log_Pos: 339110Relay_Log_File: relaylog.005481Relay_Log_Pos: 339244Relay_Master_Log_File: binlog.022343Slave_IO_Running: YesSlave_SQL_Running: YesReplicate_Do_DB:Replicate_Ignore_DB: mysql,test,information_schemaReplicate_Do_Table:Replicate_Ignore_Table:Replicate_Wild_Do_Table:
Replicate_Wild_Ignore_Table:Last_Errno: 0Last_Error:Skip_Counter: 0Exec_Master_Log_Pos: 339110Relay_Log_Space: 339244Until_Condition: NoneUntil_Log_File:Until_Log_Pos: 0Master_SSL_Allowed: NoMaster_SSL_CA_File:Master_SSL_CA_Path:Master_SSL_Cert:Master_SSL_Cipher:Master_SSL_Key:Seconds_Behind_Master: 0
1 row in set (0.00 sec)

2.Haproxy安装及其配置，master和backup安装配置都是完全一样的。

wget http://haproxy.1wt.eu/download/1.4/src/haproxy-1.4.24.tar.gz
tar -zxvf haproxy-1.4.24.tar.gz
cd haproxy-1.4.24
make TARGET=linux26 PREFIX=/usr/local/haproxy
make install PREFIX=/usr/local/haproxy
cd /usr/local/haproxy
mkdir conf logs   //在此目录下面建立conf，logs目录分别存放HAproxy的配置文件，PID文件和日志文件。
vim conf/haproxy.conf
globalmaxconn 50000chroot /usr/local/haproxyuid 99gid 99daemonquietnbproc  2pidfile /usr/local/haproxy/logs/haproxy.pid#debug
defaultslog     globalmode    httpoption  httplog   #每次请求完毕后主动关闭http通道option  dontlognull #不记录健康检查的日志信息option forwardforoption redispatchoption abortoncloseretries 3log 127.0.0.1 local3maxconn 20000contimeout      5000clitimeout      50000srvtimeout      50000
listen 192.168.1.236bind *:80mode httpstats uri /admin                    #后端服务器状态查看地址stats auth admin:admin             #状态查看页面登陆帐号密码balance source   #调度算法，source是和nginx的ip_hash同理，解决session问题option httpcloseoption forwardfor
server web1 192.168.1.248:80 weight 5 check inter 2000 rise 2 fall 5
server web2 192.168.1.249:8080 weight 5 check inter 2000 rise 2 fall 5
#启动Haproxy服务
/usr/local/haproxy/sbin/haproxy -f haproxy.conf
#haproxy启动脚本
#!/bin/bash
BASE_DIR="/usr/local/haproxy"
ARGV="$@"
start()
{
echo "START HAPoxy SERVERS"
$BASE_DIR/sbin/haproxy -f $BASE_DIR/conf/haproxy.conf
}
stop()
{
echo "STOP HAPoxy Listen"
kill -TTOU $(cat $BASE_DIR/logs/haproxy.pid)
echo "STOP HAPoxy process"
kill -USR1 $(cat $BASE_DIR/logs/haproxy.pid)
}
case $ARGV in
start)
start
ERROR=$?
;;
stop)
stop
ERROR=$?
;;
restart)
stop
start
ERROR=$?
;;
*)
echo "hactl.sh [start|restart|stop]"
esac
exit $ERROR

3.Haproxy开启系统日志支持。

vim /etc/syslog.conf
#添加：
local3.*        /var/log/haproxy.log
local0.*        /var/log/haproxy.log
vim /etc/sysconfig/syslog
#修改：
SYSLOGD_OPTIONS="-r -m 0"
#重新启动syslog服务
/etc/init.d/syslog restart

4.Keepalived的安装配置。

wget http://www.keepalived.org/software/keepalived-1.2.5.tar.gz
ln -s /usr/src/kernels/2.6.18-164.el5-x86_64/ /usr/src/linux
tar -zxvf keepalived-1.2.5.tar.gz
cd keepalived-1.2.5
./configure --prefix=/usr/local/keepalived
make && make install
cp keepalived/etc/init.d/keepalived.rh.init /etc/init.d/keepalived
chmod +x /etc/init.d/keepalived
cp keepalived/etc/init.d/keepalived.sysconfig /etc/sysconfig/keepalived
chkconfig --add keepalived
chkconfig --level 35 keepalived on
cp /usr/local/keepalived/sbin/keepalived /bin/
mkdir -p /etc/keepalived
cp /usr/local/keepalived/etc/keepalived/keepalived.conf /etc/keepalived/
#Haproxy master配置文件
vim /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {notification_email {shifeng_zhang88@163.com}notification_email_from shifeng_zhang88@163.comsmtp_server smtp.163.comsmtp_connect_timeout 30router_id LVS_Master
}
vrrp_script chk_http_port {script  "/etc/keepalived/check_haproxy.sh"interval        5       #脚本执行间隔weight         -5       #执行脚本后优先级变更：5表示优先级+5；-5则表示优先级-5
}
vrrp_instance VI_A {state MASTERinterface eth0virtual_router_id 50priority 100advert_int 1authentication {auth_type PASSauth_pass sfzhang1109}
track_script {chk_http_port}virtual_ipaddress {192.168.1.236    #haproxy虚拟IP}
}
#Haproxy backup配置文件只需改变state和priority的值
state BACKUP
priority 50
#启动keepalived服务
/etc/init.d/keepalived start

5.编辑check_haproxy.sh脚本，需要将haproxy启动脚本放到/etc/init.d/里面。

vim /etc/keepalived/check_haproxy.sh
#!/bin/bash
A=`ps -C haproxy --no-header |wc -l`
if [ $A -eq 0 ];then
/etc/init.d/haproxy restart
echo "Start haproxy" &> /dev/null
sleep 3
if [ `ps -C haproxy --no-header |wc -l` -eq 0 ];then
/etc/init.d/keepalived stop
echo "Stop keepalived" &> /dev/null
fi
fi
#chomd +x /etc/keepalived/check_haproxy.sh

6.Haproxy+Keepalived高可用性测试。

测试方法：当后端的WEB服务器down机时，haproxy会自动检测到并把请求发送到正常的服务器上面,通过haproxy监控页面（http://192.168.1.235/admin）可以查看。

说明：（希望大家提出宝贵的意见）

1）拓扑图只是项目的一个雏形，后期还要添加监控服务器和邮件等服务器。

2）Mysql主库负责用户数据的写入，slave负责用户数据的读取和数据的备份等，除了要监控mysql主从外还要监控Seconds_Behind_Master选项，以免产生主从数据不同步或者延时。

      3）当WEB的负载很大的时候，可以在haproxy添加多台物理机即可，并考虑把项目的图片分离出来做成单独的图片服务器。