HAProxy+KeepAlived实现web服务高可用、动静分离等

大致规划:

主机 IP 描述
VIP 192.168.0.222 对外提供高可用IP
haproxy+keepalived (node1) 192.168.0.111

haproxy为后端两台WEB服务的做动静分离;keepalived为haproxy做高可用。

haproxy+keepalived (node2) 192.168.0.112
WEB                (node3) 192.168.0.113 提供静态请求响应
Apache+PHP+MySQL   (node4) 192.168.0.114 提供动态请求响应

HAProxy+KeepAlived实现web服务高可用、动静分离等

一、KeepAlived的简介及原理

Keepalived的作用是检测web服务器的状态,如果有一台web服务器死机,或工作出现故障,Keepalived将检测到,并将有故障的web服务器从系统中剔除,当web服务器工作正常后Keepalived自动将web服务器加入到服务器群中,这些工作全部自动完成,不需要人工干涉,需要人工做的只是修复故障的web服务器。

Layer3,4&7工作在IP/TCP协议栈的IP层,TCP层,及应用层,原理分别如下:

  • Layer3:Keepalived使用Layer3的方式工作式时,Keepalived会定期向服务器群中的服务器发送一个ICMP的数据包(既我们平时用的Ping程序),如果发现某台服务的IP地址没有激活,Keepalived便报告这台服务器失效,并将它从服务器群中剔除,这种情况的典型例子是某台服务器被非法关机。Layer3的方式是以服务器的IP地址是否有效作为服务器工作正常与否的标准。在本文中将采用这种方式。

  • Layer4:如果您理解了Layer3的方式,Layer4就容易了。Layer4主要以TCP端口的状态来决定服务器工作正常与否。如web server的服务端口一般是80,如果Keepalived检测到80端口没有启动,则Keepalived将把这台服务器从服务器群中剔除。

  • Layer7:Layer7就是工作在具体的应用层了,比Layer3,Layer4要复杂一点,在网络上占用的带宽也要大一些。Keepalived将根据用户的设定检查服务器程序的运行是否正常,如果与用户的设定不相符,则Keepalived将把服务器从服务器群中剔除。

--引用百度百科

二、安装相关软件

WEB和LAMP搭建这里不做赘述;安装完成后可以测试是否正常访问。

HAProxy+KeepAlived实现web服务高可用、动静分离等

HAProxy+KeepAlived实现web服务高可用、动静分离等

测试两台机器正常访问。

为node1和node2各自安装keepalived和haproxy;为了简便;yum安装即可

1
2
3
4
5
6
7
8
9
10
[root@node1 ~]# rpm -q keepalived haproxy
keepalived-1.2.7-3.el6.x86_64
haproxy-1.4.24-2.el6.x86_64
                                                                                                                                                                                                                                       
-----------------------------------------------
                                                                                                                                                                                                                                     
[root@node2 ~]# rpm -q keepalived haproxy
keepalived-1.2.7-3.el6.x86_64
haproxy-1.4.24-2.el6.x86_64
[root@node2 ~]#

三、配置haproxy实现动静分离

这里在node1上演示;node2同样的配置:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
[root@node1 ~]# vim /etc/haproxy/haproxy.cfg
#这里对应global段和default段不做任何修改;但是日志需要在global段开启;
#其余的全部注释;任何重新添加以下内容;
#具体含义上一篇以做解释;这里就不做详细介绍
frontend  web
    bind *:80
    acl url_static       path_beg       -/static /images /javascript /stylesheets
    acl url_static       path_end       -i .jpg .gif .png .css .js .html .htm
    acl url_dynamic      path_end       -i .php
                                                                                                                                                                                          
    use_backend static          if url_static
    use_backend dynamic         if url_dynamic
    default_backend             static
backend static
     balance    roundrobin
     server  node3 192.168.0.113:80 check
backend dynamic
     balance    roundrobin
     server  node4 192.168.0.114:80 check
listen stats
     mode http
     bind *:1234
     stats enable
     stats refresh 3s
     stats hide-version
     stats uri  /admin?stats
     stats realm HAProxy\ Statistics
     stats auth admin:haproxy
     stats admin if TRUE
                                                                                                                                                                                           
-------------------------------------------------------------------
#复制一份到node2即可;各自启动haproxy  
[root@node1 ~]# scp /etc/haproxy/haproxy.cfg node2:/etc/haproxy/
haproxy.cfg                                                          100% 3896     3.8KB/s   00:00  
[root@node1 ~]#

查看stats信息:

HAProxy+KeepAlived实现web服务高可用、动静分离等

如后端有多台机器;则可以实现负载均衡;这里未做实例。

四、配置keepalived实现haproxy的高可用

同样在node1上配置;但是node2需要更改两个地方:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
[root@node1 ~]# vim /etc/keepalived/keepalived.conf
! Configuration File for keepalived
                                                                                                                                               
global_defs {                 #全局配置
   notification_email {
        root@localhost        #接受邮件方
   }
   notification_email_from soul@localhost  #发件人
   smtp_server 127.0.0.1                   #邮件服务器
   smtp_connect_timeout 30                 #超时时长
   router_id LVS_DEVEL                     #ID;随意即可
}
                                                                                                                                               
vrrp_script chk_down {                     #添加监控规则
    script "[[ -f /root/down ]] && exit 1 || exit 0"
#上面说明如果在/roo/下有down文件;就失败;转移到从上;否则正常
    interval 1             #多久监控一次
    weight 2               #权重
}
                                                                                                                                               
vrrp_script chk_haproxy {  #监控haproxy服务的
    script "pidof haproxy &> /dev/null && exit 0 || exit 1"
    interval 1
    weight 2
}
                                                                                                                                               
vrrp_instance VI_1 {        #添加一个实例
    state MASTER            #定义主从    #注意node2从的上为BACKUP
    interface eth0          #网络接口
    virtual_router_id 222   #虚拟路由ID;根据该ID生成虚拟MAC;保证其唯一性
    priority 100            #优先级别;BACKUP要低于MASTER
    advert_int 1            #心跳广播间隔
    authentication {        #认证
        auth_type PASS      #明文认证
        auth_pass 1111      #password
    }
    virtual_ipaddress {     #虚拟IP
        192.168.0.222
    }
    track_script {          #脚本追踪;生效上述定义脚本
        chk_down
        chk_haproxy
    }
    notify_master "/etc/keepalived/notify.sh master"   #邮件通知
    notify_backup "/etc/keepalived/notify.sh backup"
    notify_fault "/etc/keepalived/notify.sh fault"
}
#更改以上信息;其余配置文件中的信息全部注释即可
---------------------------------------------------------
#复制一份到node2;根据上述描述进行更改

提供haproxy脚本:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
[root@node1 ~]# vim /etc/keepalived/notify.sh
#!/bin/bash
# Author: MageEdu <linuxedu@foxmail.com>
# description: An example of notify script
#
vip=192.168.0.222
contact='root@localhost'
notify() {         #定义发送邮件格式等信息
    mailsubject="`hostname` to be $1: $vip floating"
    mailbody="`date '+%F %H:%M:%S'`: vrrp transition, `hostname` changed to be $1"
    echo $mailbody | mail -"$mailsubject" $contact
}
                                                                                                                                   
case "$1" in
    master)             #根据参数进行执行对应命令
        notify master  
        /etc/rc.d/init.d/haproxy start
        exit 0
    ;;
    backup)
        notify backup
        /etc/rc.d/init.d/haproxy stop
        exit 0
    ;;
    fault)
        notify fault
        /etc/rc.d/init.d/haproxy stop
        exit 0
    ;;
    *)
        echo 'Usage: `basename $0` {master|backup|fault}'
        exit 1
    ;;
esac
                                                                                                                                   
#完成后复制一份到node2即可
#启动keepalived即可

五、测试haproxy的高可用

1
2
3
4
5
6
7
8
9
10
11
12
13
14
[root@node1 ~]# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UNKNOWN qlen 1000
    link/ether 00:0c:29:5e:1e:4f brd ff:ff:ff:ff:ff:ff
    inet 192.168.0.111/16 brd 192.168.255.255 scope global eth0
    inet 192.168.0.222/32 scope global eth0
    inet6 fe80::20c:29ff:fe5e:1e4f/64 scope link
       valid_lft forever preferred_lft forever
                                                                                                                               
#查看虚拟IP目前在node1上;

HAProxy+KeepAlived实现web服务高可用、动静分离等

测试访问正常;下面down掉node1看下;

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
#由于配置文件定义了一个down的脚本;直接建立文件即可
[root@node1 ~]# touch down
[root@node1 ~]# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UNKNOWN qlen 1000
    link/ether 00:0c:29:5e:1e:4f brd ff:ff:ff:ff:ff:ff
    inet 192.168.0.111/16 brd 192.168.255.255 scope global eth0
    inet6 fe80::20c:29ff:fe5e:1e4f/64 scope link
       valid_lft forever preferred_lft forever
You have new mail in /var/spool/mail/root    #这里也提示有邮件信息
                                                                                                            
------------------------------------------------------------------
                                                                                                            
[root@node2 ~]# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
    link/ether 00:0c:29:df:70:b6 brd ff:ff:ff:ff:ff:ff
    inet 192.168.0.112/16 brd 192.168.255.255 scope global eth0
    inet 192.168.0.222/32 scope global eth0
                                                                                                            
#测试已转移到node2上

同时测试页面也是正常的。

测试haproxy脚本是否能够执行

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
[root@node1 ~]# service haproxy stop
Stopping haproxy:                                          [  OK  ]
[root@node1 ~]# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UNKNOWN qlen 1000
    link/ether 00:0c:29:5e:1e:4f brd ff:ff:ff:ff:ff:ff
    inet 192.168.0.111/16 brd 192.168.255.255 scope global eth0
    inet6 fe80::20c:29ff:fe5e:1e4f/64 scope link
                                                                                   
----------------------------------------------------------------------
                                                                                   
[root@node2 ~]# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
    link/ether 00:0c:29:df:70:b6 brd ff:ff:ff:ff:ff:ff
    inet 192.168.0.112/16 brd 192.168.255.255 scope global eth0
    inet 192.168.0.222/32 scope global eth0
                                                                                   
---------------------------------------------------------------------
                                                                                   
[root@node1 ~]# tail -5 /var/log/messages
May  2 18:25:10 node1 Keepalived_vrrp[5331]: VRRP_Script(chk_haproxy) failed
May  2 18:25:10 node1 Keepalived_vrrp[5331]: VRRP_Instance(VI_1) Received higher prio advert
May  2 18:25:10 node1 Keepalived_vrrp[5331]: VRRP_Instance(VI_1) Entering BACKUP STATE
May  2 18:25:10 node1 Keepalived_vrrp[5331]: VRRP_Instance(VI_1) removing protocol VIPs.
May  2 18:25:10 node1 Keepalived_healthcheckers[5330]: Netlink reflector reports IP 192.168.0.222 removed
                                                                                   
#可以查看系统日志也记录详细信息
#测试启动haproxy后;虚拟IP会自动转回。

到此;HAProxy+KeepAlived配置以全部完成;相关功能也都已测试成功。



本文转自Mr_陈 51CTO博客,原文链接:http://blog.51cto.com/chenpipi/1405554,如需转载请自行联系原作者
上一篇:谷歌公司利用机器学习技术帮助市场营销人员改善广告追踪能力与效果


下一篇:线程中断方法interrupt() 与 cancel()