keepalived + haproxy 实现web 双主模型的高可用负载均衡-转

1260阅读 0评论2015-12-23 yong_kang
分类:架构设计与优化


原创作品,允许转载,转载时请务必以超链接形式标明文章 原始出处 、作者信息和本声明。否则将追究法律责任。http://xz159065974.blog.51cto.com/8618592/1405812
1、本文的架构图:

 

阐述各服务器用途:

   1、haproxy在本构架中实现的是:负载均衡

   2、keepalived实现对haproxy的高可用

   3、apache static 实现静态页面的访问

   4、aoache dynamic实现动态页面的访问,图中有两个是实现负载均衡的

 

配置各功能模块:

   一、配置haproxy和keepalived

 

验证:

        1、当一台keepalived宕机后,VIP会不会转移到另外一台服务器

        2、当一台haproxy服务出故障,VIP会不会转移到另外一台服务器

 

注意:

        那如果keepalived宕机了,haproxy服务还正常运行,我们要不要让另外一台服务器把VIP夺过去呢?

        理论上来讲:最好不要,但是我们的keepalived中的脚本监控着haproxy的进程,keepalived宕机之后,就无从得知haproxy的健康状态,也不能决定自己的优先权priority降不降低了。所以,理论上来讲最好不要,但是实际中光靠keepalived是做不到的。

 

 

配置:

 

1、给两台服务器分别安装上keepalived

 

1
[root@station139 ~]# yum -y install keepalived

 

 

2、配置keepalived

 

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
51
52
53
[root@node2 ~]# vim /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
   notification_email {
        root@localhost    配置服务状态变化发送邮件到哪个地址
   }
   notification_email_from kaadmin@localhost
   smtp_server 127.0.0.1    给哪个smtp服务器发邮件
   smtp_connect_timeout 30    联系上面smtp服务器30秒联系不上,就超时
   router_id LVS_DEVEL
}
vrrp_script chk_haproxy {     本脚本是用来检测该服务器上haproxy服务的健康状态的
    script "killall -0 haproxy"
    interval 1
    weight -2
}
vrrp_instance VI_1 {
    state MASTER   这太服务器为主的keepalived
    interface eth0    通过eth0网卡广播
    virtual_router_id 200  虚拟路由id要改,如果在一个局域网中有多个keepalived集群
    priority 100   优先级
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 11112222
    }
    track_script {
        chk_haproxy
    }
    virtual_ipaddress {
        192.168.1.200   本机的虚拟IP
   }
    notify_master "/etc/keepalived/notify.sh master" 各不用状态下运行的脚本
    notify_backup "/etc/keepalived/notify.sh backup"
    notify_fault "/etc/keepalived/notify.sh fault"
}
vrrp_instance VI_2 {   另外一台主keepalived的从
    state BACKUP
    interface eth0
    virtual_router_id 57
    priority 99   设置要比另外一台主keepalived的优先级低
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 1111
    }
    track_script {
        chk_mantaince_down
   }
    virtual_ipaddress {
        192.168.1.201
   }
}

3、写keepalived处在不同状态下所运行的脚本

 

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
#!/bin/bash
# Author: MageEdu
# description: An example of notify script
#
vip=192.168.1.200
contact='root@localhost'
notify() {
    mailsubject="`hostname` to be $1: $vip floating"
    mailbody="`date '+%F %H:%M:%S'`: vrrp transition, `hostname` changed to be $1"
    echo $mailbody | mail -s "$mailsubject" $contact
}
                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   
case "$1" in
    master)
        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
给脚本以执行权限:
chmod  +x  /etc/keepalived/notify.sh

 

4、配置haproxy

   因为要实现动静分离,那么我们在配置文件中,就要定义动态资源静态资源转移到不同的服务上去

 

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
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
[root@node2 ~]# yum -y install haproxy  安装haproxy
[root@node2 ~]# vim /etc/haproxy/haproxy.cfg
    #
    log         127.0.0.1 local2
    chroot      /var/lib/haproxy
    pidfile     /var/run/haproxy.pid
    maxconn     4000
    user        haproxy
    group       haproxy
    daemon
    # turn on stats unix socket
    stats socket /var/lib/haproxy/stats
#---------------------------------------------------------------------
# common defaults that all the 'listen' and 'backend' sections will
# use if not designated in their block
#---------------------------------------------------------------------
defaults
    mode                    http      指定haproxy工作模式为http
    log                     global
    option                  httplog
    option                  dontlognull
    option http-server-close       当客户端超时时,允许服务端断开连接
    option forwardfor       except 127.0.0.0/8  在http的响应头部加入forwardfor
    option                  redispatch    #在使用了基于cookie的会话保持的时候,通常加这么一项,一旦后端某一server宕机时,能够将其会话重新派发到其它的upstream servers
    retries                 3
    timeout http-request    10s
    timeout queue           1m
    timeout connect         10s
    timeout client          1m
    timeout server          1m
    timeout http-keep-alive 10s
    timeout check           10s
    maxconn                 3000
#---------------------------------------------------------------------
# main frontend which proxys to the backends
#---------------------------------------------------------------------
frontend  main *:80  前端代理
    acl url_static       path_beg       -i /static /images /javascript /stylesheets
    acl url_static       path_end       -i .jpg .gif .png .css .js
    acl url_dynamic      path_end       -i .php
    use_backend static          if url_static
    default_backend             dynamic
#---------------------------------------------------------------------
# static backend for serving up images, stylesheets and such
#---------------------------------------------------------------------
backend static   后端的静态请求响应
    balance     roundrobin
    server      static 192.168.1.100:80 inter 3000 rise 2 fall 3 check maxconn 5000
#---------------------------------------------------------------------
# round robin balancing between the various backends
#---------------------------------------------------------------------
backend dynamic   后端的动态请求响应
    balance     roundrobin
    server  dynamic1 192.168.1.101:80 inter 3000 rise 2 fall 3 check maxconn 5000
    server  dynamic2 192.168.1.102:80 inter 3000 rise 2 fall 3 check maxconn 5000
listen statistics
        mode http
        bind *:8080
~       stats enable
        stats auth admin:admin
        stats uri /admin?stats   指定URI的访问路径
        stats admin if TRUE
        stats hide-version
        stats refresh 5s
        acl allow src 192.168.0.0/24   定义访问控制列表
        tcp-request content accept if allow
        tcp-request content reject

 

 

5、配置另外一台haproxy服务器

   因为两台服务器的配置大体相同,我们就直接讲以上配置好的复制文件和脚本文件都传到这台haproxy服务器上,做下修就可以了

 

 

1
2
3
4
5
6
7
8
9
[root@node2 ~]# scp /etc/keepalived/keepalived.conf root@192.168.1.121:/etc/keepalived/
root@192.168.1.121's password:
keepalived.conf                                                                               100% 4546     4.4KB/s   00:00
[root@node2 ~]# scp /etc/keepalived/notify.sh root@192.168.1.121:/etc/keepalived/
root@192.168.1.121's password:
notify.sh                                                                                     100%  751     0.7KB/s   00:00
[root@node2 ~]# scp /etc/haproxy/haproxy.cfg root@192.168.1.121:/etc/haproxy/
root@192.168.1.121's password:
haproxy.cfg                                                                                   100% 3529     3.5KB/s   00:00

 

    传输完成,接着来配置 /etc/keepalived/keepalived.conf  因为两个节点上的/etc/haproxy/haproxy.cfg相同不用更改

 

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
    interface eth0
! Configuration File for keepalived
global_defs {
   notification_email {
        root@localhost
   }
   notification_email_from kaadmin@localhost
   smtp_server 127.0.0.1
   smtp_connect_timeout 30
   router_id LVS_DEVEL
}
vrrp_script chk_haproxy {
    script "killall -0 haproxy"
    interval 1
    weight -2
}
vrrp_instance VI_1 {
    state BACKUP      这台把master改成 backup
    interface eth0
    virtual_router_id 200
    priority 99       优先级调的比上一个低
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 11112222
    }
    track_script {
        chk_haproxy
    }
    virtual_ipaddress {
        192.168.1.200
   }
}
vrrp_instance VI_2 {
    state MASTER      本台的这个要调成MASTER,上个是backup
    interface eth0
    virtual_router_id 57
    priority 100    这个优先级也要高于上个
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 1111
    }
    virtual_ipaddress {
        192.168.1.201
   }
    notify_master "/etc/keepalived/notify.sh master"
    notify_backup "/etc/keepalived/notify.sh backup"
    notify_fault "/etc/keepalived/notify.sh fault"
}

注意:

notify_master "/etc/keepalived/notify.sh master"
notify_backup "/etc/keepalived/notify.sh backup"

notify_fault "/etc/keepalived/notify.sh fault"

 

3个状态分别要执行的脚本,只能放在 MASTER中,原因是:因为是互为主从,每个主的都会有个另外一个主的从,如果

把这 “3个状态执行脚本” 写入到从的区域中,那么另外一个主的从状态就会执行这个脚本,因为就会停掉所要高可用的

程序,这就造成了,两个VIP全部转移到其中一个服务器上去。

 

我们来验证下,如果keepalived和haproxy分别宕机,vip会不会转移:

   在两个节点上都启动 keepalived和haproxy服务

 

1
2
3
4
[root@node2 ~]# service haproxy start
Starting haproxy:                                          [  OK  ]
[root@node2 ~]# service keepalived start
Starting keepalived:                                       [  OK  ]

 

 

以下为正常情况:

 

keepalived 1:

 

 

keepalived 2:

 

 

我们来模拟让第一个haproxy停止掉,再看下,VIP会不会全到 keepalived 2上面去:

 

1
2
[root@node2 ~]# service haproxy stop
Stopping haproxy:                                          [  OK  ]

 

查看keepalived 1 和 keepalived 2

 

 

 

   看,都过来了。。。。

 

 


 

验证负载均衡很动静分离

   我们给3个web服务不同的网页

   1、给apache static一个静态页面,来验证如果请求的不是以 .php结尾的网页文件都定向到这太服务器上来

   2、给apache dynamic 1 、2 分别两个 index.php ,实现对动态网页的负载均衡

 

我们给apache static 一个符合-i .jpg .gif .png .css .js的网页,就给个图片网页吧

 

apache static

 

1
scp 1.png root@192.168.1.100:/var/www/html

 

apache dynamic 1

 

1
2
3
4
5
vim /var/www/html/index.php
192.168.1.101
    phpinfo();
?>

 

 apache dynamic 2 

 

1
2
3
4
5
vim /var/www/html/index.php
192.168.1.102
    phpinfo();
?>

 

1、我们来请求 1.png 结尾的静态文件

 

 

2、我们来请求 .php结尾的页面

 

 

   如此看来,已经对以.php的动态页面做了负载均衡了

 

   我们再通过 192.168.1.201 这个虚拟 ip 访问试试:

 

 

由此看来,也实现了双主模型了,两个haproxy同时可以服务了。。。


 

3、我们来看看状态页面

 

 

 

 

 

本文出自 “linux运维” 博客,请务必保留此出处http://xz159065974.blog.51cto.com/8618592/1405812



上一篇:haproxy 配置 说明
下一篇:手动安装mysql,使用mysqld_safe启动mysql服务