keepalived个人简单理解就是实现一个虚IP在keepalived主从服务器之间切换的功能，当主的keeplived挂掉，从机会无缝接管该虚IP。

keepalived它作为一个辅助实现高可用性工具，一般都会配合某个具体服务工作，例如mysql,drbd，haproxy等，本篇介绍的就是haproxy,在前期做好keepalived配置后，做一个切换脚本，当haproxy挂掉后，脚本会立即执行keepalived关闭操作，从而让从机接管,该虚IP其实会和本机的物理IP做绑定，访问虚IP其实就可以理解为访问本机物理IP，该虚IP会在脚本被触发后切换到从机，而后访问虚IP也就是在访问从机物理IP，从而实现haproxy的高可用性。

这个实验模拟的功能其实就是利用haproxy实现apache服务器间的负载均衡，缓解并发压力，并保证haproxy-master若挂掉,haproxy-backup能无缝接管，实现WEB站点负载均衡+高可用性。保证客户端无缝获取网站资源。

解决方案：

系统环境：centos6.3

apache: httpd-2.4.4

haproxy: haproxy-1.4.8

keepalived：keepalived-1.2.7

haproxy VIP:                  192.168.7.70

haproxy-master(haproxy1):     192.168.7.71    ha1.example.com

haproxy-backup(haproxy2):     192.168.7.72    ha2.example.com

apache1:                      192.168.7.73    www1.example.com

apache2:                      192.168.7.74    www2.example.com

(haproxy1)为仅haproxy1配置

(haproxy2)为仅haproxy2配置

(haproxy1，haproxy2)为haproxy1和haproxy2都得配置

部署环境：

1.关闭iptables和SELINUX

# service iptables stop

# setenforce 0

# vi /etc/sysconfig/selinux

---------------

SELINUX=disabled

---------------

2.apache安装

配置详见传送门：http://www.showerlee.com/archives/6

注：haproxy+keepalived其实搭配nginx效果也不错，本人线上因为长期使用apache，所以这篇博文就以apache介绍为主。

安装完毕只需配置客户端浏览器访问根页面显示本机IP地址。

一.haproxy安装配置：(haproxy1，haproxy2)

# wget http://haproxy.1wt.eu/download/1.4/src/haproxy-1.4.8.tar.gz

# tar zxvf haproxy-1.4.8.tar.gz

# cd haproxy-1.4.8

# uname -a           //查看linux内核版本

# make TARGET=linux26 PREFIX=/usr/local/haproxy

# make install PREFIX=/usr/local/haproxy

# useradd -s /sbin/nologin haproxy

# passwd haproxy

# chown -R haproxy.haproxy /usr/local/haproxy

配置：

# vi /usr/local/haproxy/haproxy.cfg

-----------------

global

log 127.0.0.1 local0

maxconn 5120  

chroot /usr/local/haproxy  

user haproxy  

group haproxy  

daemon  

quiet  

nbproc  1  

pidfile /usr/local/haproxy/haproxy.pid

#启动服务后后会滚动日志，生产环境建议注释掉

debug  

defaults

log 127.0.0.1 local3  

mode http  

option httplog

option httpclose

option  dontlognull

#option  forwardfor  

option  redispatch

retries 2

maxconn 2000

balance source  

contimeout      5000  

clitimeout      50000  

srvtimeout      50000  

listen web_proxy :80

     server www1  192.168.7.73:80  weight 5 check inter 2000 rise 2 fall 5

     server www2  192.168.7.74:80  weight 5 check inter 2000 rise 2 fall 5

listen stats :8888  

mode http  

#transparent  

stats uri / haproxy-stats  

stats realm Haproxy \ statistic

#认证  

stats auth haproxy:password  

-----------------

启动haproxy

# /usr/local/haproxy/sbin/haproxy -f /usr/local/haproxy/haproxy.cfg &

注：这里加上“&”是为了让haproxy服务后台运行，去掉“&”可实时查看其滚动日志

日志：

------------------------

Available polling systems :

    sepoll : pref=400,  test result OK

     epoll : pref=300,  test result OK

      poll : pref=200,  test result OK

    select : pref=150,  test result OK

Total: 4 (4 usable), will use sepoll.

Using sepoll() as the polling mechanism.

00000000:web_proxy.accept(0004)=0007 from [192.168.7.129:5752]

00000000:web_proxy.clireq[0007:ffff]: GET / HTTP/1.1

00000000:web_proxy.clihdr[0007:ffff]: Accept: text/html, application/xhtml+xml, */*

00000000:web_proxy.clihdr[0007:ffff]: Accept-Language: zh-CN

00000000:web_proxy.clihdr[0007:ffff]: User-Agent: Mozilla/5.0 (compatible; MSIE

9.0; Windows NT 6.1; WOW64; Trident/5.0)

00000000:web_proxy.clihdr[0007:ffff]: Accept-Encoding: gzip, deflate

00000000:web_proxy.clihdr[0007:ffff]: Host: 192.168.7.71

00000000:web_proxy.clihdr[0007:ffff]: If-Modified-Since: Tue, 28 May 2013 18:22:10

GMT

00000000:web_proxy.clihdr[0007:ffff]: If-None-Match: "10-4ddcb57ecf1ee"

00000000:web_proxy.clihdr[0007:ffff]: Connection: Keep-Alive

00000000:web_proxy.srvrep[0007:0008]: HTTP/1.1 304 Not Modified

00000000:web_proxy.srvhdr[0007:0008]: Date: Tue, 28 May 2013 19:48:35 GMT

00000000:web_proxy.srvhdr[0007:0008]: Server: Apache/2.4.4 (Unix)

00000000:web_proxy.srvhdr[0007:0008]: Connection: close

00000000:web_proxy.srvhdr[0007:0008]: ETag: "10-4ddcb57ecf1ee"

00000000:web_proxy.srvcls[0007:0008]

00000000:web_proxy.clicls[0007:0008]

00000000:web_proxy.closed[0007:0008]

------------------------

查看是否启动

# ps -ef|grep haproxy

--------------------------

haproxy   2255     1  0 03:33 ?        00:00:00 /usr/local/haproxy/sbin/haproxy -f

/usr/local/haproxy/haproxy.cfg

root      2323  2291  0 03:56 pts/1    00:00:00 grep haproxy

--------------------------

重启haproxy

# pkill haproxy

# /usr/local/haproxy/sbin/haproxy -f /usr/local/haproxy/haproxy.cfg

其中：

haproxy代理：192.168.7.71:80

apache1：    192.168.7.73:80

apache2：    192.168.7.74:80

统计页面监听的端口：8888

访问页面：

http://192.168.7.71:8888/haproxy-stats

认证账号/密码：haproxy/password

总结：

    通过日志可以看出，客户端192.168.7.129连接haproxy192.168.7.71的80端口，且客户端无论如何刷新页面，haproxy都只会把访问请求跳转到apache2的192.168.7.74地址，这是因为balance source这个参数会保持会话ID,如果改成balance roundrobin模式，那么客户端会轮流连接两台web服务器，线上还是建议使用balance source,这样会保持某一客户端在长时间内保持他的会话，不会来回跳转。

    这里其实已经利用haproxy完成了对两台apache服务器的负载均衡功能，但如何保证负载均衡的高可用性，这里就得利用keepalived的热备功能，保证haproxy1如果挂掉，haproxy2能实时接管，实现网站前端负载均衡高可用，这也是我们目前比较流行的组合haproxy+keepalived。

二.keepalived安装配置：(haproxy1，haproxy2）

# wget http://www.keepalived.org/software/keepalived-1.2.7.tar.gz

# tar zxvf keepalived-1.2.7.tar.gz

# cd keepalived-1.2.7

# ./configure --prefix=/usr/local/keepalived --with-kernel-

dir=/usr/src/kernels/2.6.32-279.el6.x86_64

# make && make install

设置keepalived启动脚本

# cp /usr/local/keepalived/etc/rc.d/init.d/keepalived /etc/rc.d/init.d/

# cp /usr/local/keepalived/etc/sysconfig/keepalived /etc/sysconfig/

# mkdir /etc/keepalived

# cp /usr/local/keepalived/sbin/keepalived /usr/sbin/

# chkconfig keepalived on

(haproxy1）

# vi /etc/keepalived/keepalived.conf

----------------------

! Configuration File for keepalived

global_defs {

router_id LVS_DEVEL

}

vrrp_instance VI_1 {

state MASTER

interface eth0

virtual_router_id 51

priority 150

advert_int 1

authentication {

auth_type PASS

auth_pass 1111

}

virtual_ipaddress {

192.168.7.70

}

}

----------------------

(haproxy2）

# vi /etc/keepalived/keepalived.conf

----------------------

! Configuration File for keepalived

global_defs {

router_id LVS_DEVEL

}

vrrp_instance VI_1 {

state BACKUP

interface eth0

virtual_router_id 51

priority 120

advert_int 1

authentication {

auth_type PASS

auth_pass 1111

}

virtual_ipaddress {

192.168.7.70

}

}

----------------------

启动keepalived

# service keepalived start

# ps -ef |grep keepalived

----------------------

root      1726  1722  0 10:57 ?        00:00:00 sh /etc/keepalived/check_haproxy.sh

root      2678     1  0 11:15 ?        00:00:00 keepalived -D

root      2680  2678  0 11:15 ?        00:00:00 keepalived -D

root      2681  2678  0 11:15 ?        00:00:00 keepalived -D

root      2691  2178  0 11:15 pts/1    00:00:00 grep keepalived

-----------------------

创建haproxy+keepalived脚本：

实现当haproxy挂掉后，能再次启动haproxy,若无法再次启动则彻底关闭keepalived将VIP交给

从机处理。

(haproxy1，haproxy2）

# vi /etc/keepalived/check_haproxy.sh

---------------------

#!/bin/bash

while :

do

hapid=`ps -C haproxy --no-header |wc -l`

  if [ $hapid -eq 0 ];then

  /usr/local/haproxy/sbin/haproxy -f /usr/local/haproxy/haproxy.cfg

  sleep 5

    if [ $hapid -eq 0 ];then

    /etc/init.d/keepalived stop

    fi

  fi

  sleep 5

done

--------------------

改执行权限

# chmod 755 /etc/keepalived/check_haproxy.sh

强制后台执行（关闭客户端连接也会继续运行）

# nohup sh /etc/keepalived/check_haproxy.sh

注；该脚本参考酒哥的"构建高可用LINUX服务器"一书，需在keepalived开机正常启动后，方可在主从机执行。

本人实测，不能将该脚本做开机启动，会导致开机将keepalived服务关闭，如想实现开机自动执行该脚本，估计要做脚本语句优化，这里有兴趣的朋友可以帮忙修改一下。

三.测试：

1.在两台机器上分别执行ip add，目前显示VIP在haproxy1上与本机网卡绑定

(haproxy1）

# ip add

-----------------------

eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen

1000

   link/ether 00:0c:29:d2:c6:75 brd ff:ff:ff:ff:ff:ff

   inet 192.168.7.71/24 brd 192.168.7.255 scope global eth0

   inet 192.168.7.70/32 scope global eth0

   inet6 fe80::20c:29ff:fed2:c675/64 scope link

      valid_lft forever preferred_lft forever

-----------------------

(haproxy2）

# ip add

-----------------------

eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen

1000

   link/ether 00:0c:29:ff:21:19 brd ff:ff:ff:ff:ff:ff

   inet 192.168.7.72/24 brd 192.168.7.255 scope global eth0

   inet6 fe80::20c:29ff:feff:2119/64 scope link

      valid_lft forever preferred_lft forever

-----------------------

在浏览器访问该虚IP:

http://192.168.7.70

返回如图：

证明，keepalived让haproxy1接管负载均衡，将页面跳转到apache2上

2.停掉haproxy1上的haproxy服务，5秒后keepalived会自动将其再次启动

（haproxy1）

# pkill haproxy

等5秒

# ps -ef |grep haproxy

--------------

root      1712  1708  0 01:17 ?        00:00:00 sh /etc/keepalived/check_haproxy.sh

haproxy   2440     1  0 01:29 ?        00:00:00 /usr/local/haproxy/sbin/haproxy -f

/usr/local/haproxy/haproxy.cfg

root      2443  1807  0 01:29 pts/1    00:00:00 grep haproxy

--------------

3.停掉主的keepalived，备机马上接管服务

（haproxy1）

# service keepalived stop

# uname -a

---------------

Linux ha1.example.com 2.6.32-279.el6.x86_64 #1 SMP Fri Jun 22 12:19:21 UTC 2012

x86_64 x86_64 x86_64 GNU/Linux

---------------

（haproxy2）

# ip add

eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen

1000

   link/ether 00:0c:29:ff:21:19 brd ff:ff:ff:ff:ff:ff

   inet 192.168.7.72/24 brd 192.168.7.255 scope global eth0

   inet 192.168.7.70/32 scope global eth0

   inet6 fe80::20c:29ff:feff:2119/64 scope link

      valid_lft forever preferred_lft forever

# uname -a

---------------

Linux ha2.example.com 2.6.32-279.el6.x86_64 #1 SMP Fri Jun 22 12:19:21 UTC 2012

x86_64 x86_64 x86_64 GNU/Linux

---------------

现已跳转到haproxy2,在浏览器再次访问该虚IP:

http://192.168.7.70

返回如图：

---------大功告成----------

• 08-28CENTOS6.3下WEB远程IPMI管理 (阅读 (20,125) 抢沙发)
• 08-26CentOS 6.3下安装部署Oracle服务器 (阅读 (8,365) 抢沙发)