HAproxy+Keepalied做MySQL负载均衡

HAproxy介绍

Keepalive介绍

应用服务器对只读的应用程序连接虚拟IP地址，连接到haproxy，然后通过haproxy将TCP协议转移到下面的2个MySQL SLAVE数据库服务器中。Haproxy在此做4层的TCP交换服务。keepalived为了防止haproxy单点故障。但是我这里使用两个VIP（虚拟IP）就可以做到haproxy代理服务器的高可用负载均衡。对于这两个VIP而言，可以添加两条DNS记录进行轮训，也可以在程序中直接分开使用，当任何一个haproxy节点有问题时这两个VIP都能转到同一个节点上继续提供服务。

实验架构如下图：

一、MySQL配置

这里安装两个MySQL实例，由于我是测试环境，我这里就在同一台机器上创建两个数据库实例进行测试（3306和3307）。数据库安装这里就不介绍了，但为了测试需要在两个实例中分别创建一个haproxy账号。

10.99.73.10:（3306,3307）

mysql> grant all on *.* to 'haproxy'@'%' identified by '123456';
Query OK, 0 rows affected, 2 warnings (0.00 sec)

mysql> flush privileges;
Query OK, 0 rows affected (0.00 sec)

mysql> grant all on *.* to 'haproxy'@'%' identified by '123456';

Query OK, 0 rows affected, 2 warnings (0.00 sec)

mysql> flush privileges;

Query OK, 0 rows affected (0.00 sec)

二、haproxy+keepalived配置

在配置之前，先将HA中个节点进行时间同步，主机名之间相互解析，以及配置双机互信，关闭防火墙和selinux。

haproxy-01

[root@haproxy-01 ~]# cat /etc/redhat-release 
CentOS Linux release 7.2.1511 (Core) 

[root@haproxy-01 ~]# systemctl stop firewalld
[root@haproxy-01 ~]# setenforce 0

[root@haproxy-01 ~]# cat /etc/hosts
172.29.28.193 haproxy-01
172.29.28.195 haproxy-02

[root@haproxy-01 ~]# ssh-keygen -t rsa -P ''
[root@haproxy-01 ~]# ssh-copy-id haproxy-02

[root@haproxy-01 ~]# yum install ntp
[root@haproxy-01 ~]# ntpdate s2c.time.edu.cn

[root@haproxy-01 ~]# cat /etc/redhat-release

CentOS Linux release 7.2.1511 (Core)

[root@haproxy-01 ~]# systemctl stop firewalld

[root@haproxy-01 ~]# setenforce 0

[root@haproxy-01 ~]# cat /etc/hosts

172.29.28.193 haproxy-01

172.29.28.195 haproxy-02

[root@haproxy-01 ~]# ssh-keygen -t rsa -P ''

[root@haproxy-01 ~]# ssh-copy-id haproxy-02

[root@haproxy-01 ~]# yum install ntp

[root@haproxy-01 ~]# ntpdate s2c.time.edu.cn

haproxy-02

[root@haproxy-01 ~]# cat /etc/redhat-release 
CentOS Linux release 7.2.1511 (Core) 

[root@haproxy-02 ~]# systemctl stop firewalld
[root@haproxy-02 ~]# setenforce 0

[root@haproxy-02 ~]# cat /etc/hosts
172.29.28.193 haproxy-01
172.29.28.195 haproxy-02

[root@haproxy-02 ~]# ssh-keygen -t rsa -P ''
[root@haproxy-02 ~]# ssh-copy-id haproxy-01

[root@haproxy-02 ~]# yum install ntp
[root@haproxy-02 ~]# ntpdate s2c.time.edu.cn

[root@haproxy-01 ~]# cat /etc/redhat-release

CentOS Linux release 7.2.1511 (Core)

[root@haproxy-02 ~]# systemctl stop firewalld

[root@haproxy-02 ~]# setenforce 0

[root@haproxy-02 ~]# cat /etc/hosts

172.29.28.193 haproxy-01

172.29.28.195 haproxy-02

[root@haproxy-02 ~]# ssh-keygen -t rsa -P ''

[root@haproxy-02 ~]# ssh-copy-id haproxy-01

[root@haproxy-02 ~]# yum install ntp

[root@haproxy-02 ~]# ntpdate s2c.time.edu.cn

安装haproxy和keepalived

[root@haproxy-01 ~]# yum install keepalived haproxy -y
[root@haproxy-02 ~]# yum install keepalived haproxy -y
[root@haproxy-01 ~]# keepalived -v
Keepalived v1.2.13 (11/05,2016)
[root@haproxy-01 ~]# haproxy -v
HA-Proxy version 1.5.18 2016/05/10

[root@haproxy-01 ~]# yum install keepalived haproxy -y

[root@haproxy-02 ~]# yum install keepalived haproxy -y

[root@haproxy-01 ~]# keepalived -v

Keepalived v1.2.13 (11/05,2016)

[root@haproxy-01 ~]# haproxy -v

HA-Proxy version 1.5.18 2016/05/10

配置（haproxy-01，haproxy-02）

[root@haproxy-01 ~]# cat /etc/haproxy/haproxy.cfg
#---------------------------------------------------------------------
# global configure
#---------------------------------------------------------------------
global                                        #定义全局配置段;
    log         127.0.0.1 local2              #通过rsyslog将日志进行归档记录;
    pidfile     /var/run/haproxy.pid          #指定pid文件;
    maxconn     8000                          #最大并发连接数;
    user        haproxy                       #运行haproxy的用户;
    group       haproxy                       #运行haproxy的组;
    daemon                                    #以守护进程的形式运行,即后台运行;
 
#---------------------------------------------------------------------
# common defaults that all the 'listen' and 'backend' sections will
# use if not designated in their block
#---------------------------------------------------------------------
defaults                                      #默认配置端;
    mode                    http              #工作模式,源码包编译默认为tcp;
    log                     global            #记录全局日志;
    option                  httplog           #详细记录http日志;
    option                  dontlognull       #不记录健康检测的日志信息;
    option http-server-close                  #启用服务器端主动关闭功能;
    option forwardfor       except 127.0.0.0/8#传递client端IP至后端real server;
    option                  redispatch        #基于cookie做会话保持时,后端对应存放session的服务器出现故障时,会话会被重定向至别的服务器;
    retries                 3                 #请求重传次数;
    timeout http-request    10s               #断开客户端连接的时长;
    timeout queue           1m                #一个请求在队列里的超时时长;
    timeout connect         10s               #设定在haproxy转发至后端upstream server时等待的超时时长;
    timeout client          1m                #client的一次非活动状态的超时时长;
    timeout server          1m                #等待服务器端的非活动的超时时长;
    timeout http-keep-alive 10s               #持久连接超时时长;
    timeout check           10s               #检查请求连接的超时时长;
    maxconn                 3000              #最大连接数;
 
listen mysql_3306 *:3306
    mode   tcp
    option tcpka
    balance     roundrobin
    #将此server命名为mysql_01,每隔3000ms检测一个健康状态,如果检测3次都失败,将此server剔除,在离线的状态下,只要检测1次成功,就让其上线;
    server mysql_01 10.99.73.10:3306 check inter 3000 rise 1 maxconn 4000 fall 3
    server mysql_02 10.99.73.10:3307 check inter 3000 rise 1 maxconn 4000 fall 3

listen state                                  #使用单独输出,不需要frontedn调用：定义haproxy的状态统计页面;
    bind *:8080                               #监听的地址;
    mode http                                 #http 7层工作模式:对应用层数据做深入分析,因此支持7层的过滤、处理、转换等机制;
    stats enable                              #开启统计页面输出;
    stats hide-version                        #隐藏状态页面版本号;
    stats uri /haproxyadmin?stats             #指定状态页的访问路径;
    stats auth admin:admin                    #基于用户名,密码验证;
    stats admin if TRUE                       #验证通过时运行登录;
    acl num1 src 172.29.0.0/16                #定义源地址为172.29.0.0/16网段的acl规则,将其命名为num1;
    tcp-request content accept if num1        #如果满足此规则,则允许访问;
    tcp-request content reject                #拒绝其他所有的访问;

[root@haproxy-01 ~]# cat /etc/haproxy/haproxy.cfg

#---------------------------------------------------------------------

# global configure

#---------------------------------------------------------------------

global #定义全局配置段;

log 127.0.0.1 local2 #通过rsyslog将日志进行归档记录;

pidfile /var/run/haproxy.pid #指定pid文件;

maxconn 8000 #最大并发连接数;

user haproxy #运行haproxy的用户;

group haproxy #运行haproxy的组;

daemon #以守护进程的形式运行,即后台运行;

#---------------------------------------------------------------------

# common defaults that all the 'listen' and 'backend' sections will

# use if not designated in their block

#---------------------------------------------------------------------

defaults #默认配置端;

mode http #工作模式,源码包编译默认为tcp;

log global #记录全局日志;

option httplog #详细记录http日志;

option dontlognull #不记录健康检测的日志信息;

option http-server-close #启用服务器端主动关闭功能;

option forwardfor except 127.0.0.0/8#传递client端IP至后端real server;

option redispatch #基于cookie做会话保持时,后端对应存放session的服务器出现故障时,会话会被重定向至别的服务器;

retries 3 #请求重传次数;

timeout http-request 10s #断开客户端连接的时长;

timeout queue 1m #一个请求在队列里的超时时长;

timeout connect 10s #设定在haproxy转发至后端upstream server时等待的超时时长;

timeout client 1m #client的一次非活动状态的超时时长;

timeout server 1m #等待服务器端的非活动的超时时长;

timeout http-keep-alive 10s #持久连接超时时长;

timeout check 10s #检查请求连接的超时时长;

maxconn 3000 #最大连接数;

listen mysql_3306 *:3306

mode tcp

option tcpka

balance roundrobin

#将此server命名为mysql_01,每隔3000ms检测一个健康状态,如果检测3次都失败,将此server剔除,在离线的状态下,只要检测1次成功,就让其上线;

server mysql_01 10.99.73.10:3306 check inter 3000 rise 1 maxconn 4000 fall 3

server mysql_02 10.99.73.10:3307 check inter 3000 rise 1 maxconn 4000 fall 3

listen state #使用单独输出,不需要frontedn调用：定义haproxy的状态统计页面;

bind *:8080 #监听的地址;

mode http #http 7层工作模式:对应用层数据做深入分析,因此支持7层的过滤、处理、转换等机制;

stats enable #开启统计页面输出;

stats hide-version #隐藏状态页面版本号;

stats uri /haproxyadmin?stats #指定状态页的访问路径;

stats auth admin:admin #基于用户名,密码验证;

stats admin if TRUE #验证通过时运行登录;

acl num1 src 172.29.0.0/16 #定义源地址为172.29.0.0/16网段的acl规则,将其命名为num1;

tcp-request content accept if num1 #如果满足此规则,则允许访问;

tcp-request content reject #拒绝其他所有的访问;

对于global端的log配置，需要在/etc/rsyslog.conf配置文件中，添加’local2.* /var/log/haproxy.log’,并且启用$ModLoad imudp，$UDPServerRun 514，$ModLoad imtcp，$InputTCPServerRun 514此四项功能，最后重启rsyslog进程。

# Provides UDP syslog reception
$ModLoad imudp
$UDPServerRun 514

# Provides TCP syslog reception
$ModLoad imtcp
$InputTCPServerRun 514

local2.*                                                /var/log/haproxy.log

# Provides UDP syslog reception

$ModLoad imudp

$UDPServerRun 514

# Provides TCP syslog reception

$ModLoad imtcp

$InputTCPServerRun 514

local2.* /var/log/haproxy.log

[root@haproxy-01 ~]# systemctl restart  rsyslog.service

1	[root@haproxy-01 ~]# systemctl restart rsyslog.service

启动两个节点

[root@haproxy-01 ~]# systemctl restart haproxy

1	[root@haproxy-01 ~]# systemctl restart haproxy

配置keepalived：双主模型

haproxy-01

[root@haproxy-01 ~]# cat /etc/keepalived/keepalived.conf
! Configuration File for keepalived

global_defs {
   notification_email {                               #定义收件人邮箱;
      18612100045@163.com 
   }
   notification_email_from root                       #定义发件人邮箱;
   smtp_server localhost                              #定义邮件服务器地址;
   smtp_connect_timeout 30                            #定有邮件服务器连接超时时长为30秒;
   router_id LVS_DEVEL                                #运行keepalive的机器的标识;
}

vrrp_instance VI_1 {                                  #定义VRRP实例,实例名自定义;
    state MASTER                                      #指定当前节点的角色,master为主,backup为从;
    interface eno16777736                             #直接HA监测的接口;
    virtual_router_id 51                              #虚拟路由标识,在同一VRRP实例中,主备服务器ID必须一样;
    priority 100                                      #定义节点优先级,数字越大越优先,主服务器优先级高于从服务器;
    advert_int 1                                      #设置主备之间永不检查时间间隔,单位为秒;
    authentication {                                  #设置主从之间验证类型和密码;
        auth_type PASS
        auth_pass 123456

    }
    virtual_ipaddress {
        172.29.28.243                                 #定义虚拟ip地址;
    }
}    

vrrp_instance VI_2 {
    state BACKUP
    interface eno16777736 
    virtual_router_id 52
    priority 50 
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 123456

    }
    virtual_ipaddress {
        172.29.28.254
    }
}

[root@haproxy-01 ~]# cat /etc/keepalived/keepalived.conf

! Configuration File for keepalived

global_defs {

notification_email { #定义收件人邮箱;

18612100045@163.com

}

notification_email_from root #定义发件人邮箱;

smtp_server localhost #定义邮件服务器地址;

smtp_connect_timeout 30 #定有邮件服务器连接超时时长为30秒;

router_id LVS_DEVEL #运行keepalive的机器的标识;

}

vrrp_instance VI_1 { #定义VRRP实例,实例名自定义;

state MASTER #指定当前节点的角色,master为主,backup为从;

interface eno16777736 #直接HA监测的接口;

virtual_router_id 51 #虚拟路由标识,在同一VRRP实例中,主备服务器ID必须一样;

priority 100 #定义节点优先级,数字越大越优先,主服务器优先级高于从服务器;

advert_int 1 #设置主备之间永不检查时间间隔,单位为秒;

authentication { #设置主从之间验证类型和密码;

auth_type PASS

auth_pass 123456

}

virtual_ipaddress {

172.29.28.243 #定义虚拟ip地址;

}

vrrp_instance VI_2 {

state BACKUP

interface eno16777736

virtual_router_id 52

priority 50

advert_int 1

authentication {

auth_type PASS

auth_pass 123456

}

virtual_ipaddress {

172.29.28.254

}

haproxy-02

[root@haproxy-02 ~]# cat /etc/keepalived/keepalived.conf
! Configuration File for keepalived

global_defs {
   notification_email {
      18612100045@163.com 
   }
   notification_email_from root
   smtp_server localhost
   smtp_connect_timeout 30
   router_id LVS_DEVEL 
}

vrrp_instance VI_1 {
    state BACKUP
    interface eno16777736
    virtual_router_id 51
    priority 50
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 123456

    }
    virtual_ipaddress {
        172.29.28.253
    }
}    

vrrp_instance VI_2 {
    state MASTER 
    interface eno16777736 
    virtual_router_id 52
    priority 100 
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 123456

    }
    virtual_ipaddress {
        172.29.28.254
    }
}

[root@haproxy-02 ~]# cat /etc/keepalived/keepalived.conf

! Configuration File for keepalived

global_defs {

notification_email {

18612100045@163.com

}

notification_email_from root

smtp_server localhost

smtp_connect_timeout 30

router_id LVS_DEVEL

}

vrrp_instance VI_1 {

state BACKUP

interface eno16777736

virtual_router_id 51

priority 50

advert_int 1

authentication {

auth_type PASS

auth_pass 123456

}

virtual_ipaddress {

172.29.28.253

}

vrrp_instance VI_2 {

state MASTER

interface eno16777736

virtual_router_id 52

priority 100

advert_int 1

authentication {

auth_type PASS

auth_pass 123456

}

virtual_ipaddress {

172.29.28.254

}

PS：修改了优先级和角色。

重启keepalived

[root@haproxy-01 ~]# systemctl restart keepalived.service
[root@haproxy-02 ~]# systemctl restart keepalived.service

1 2	[root@haproxy-01 ~]# systemctl restart keepalived.service [root@haproxy-02 ~]# systemctl restart keepalived.service

查看启动情况

[root@haproxy-01 ~]# netstat -nplt
Active Internet connections (only servers)
Proto Recv-Q Send-Q Local Address           Foreign Address         State       PID/Program name     
tcp        0      0 0.0.0.0:3306            0.0.0.0:*               LISTEN      3351/haproxy

[root@haproxy-02 ~]# netstat -nplt
Active Internet connections (only servers)
Proto Recv-Q Send-Q Local Address           Foreign Address         State       PID/Program name   
tcp        0      0 0.0.0.0:3306            0.0.0.0:*               LISTEN      3355/haproxy

[root@haproxy-01 ~]# netstat -nplt

Active Internet connections (only servers)

Proto Recv-Q Send-Q Local Address Foreign Address State PID/Program name

tcp 0 0 0.0.0.0:3306 0.0.0.0:* LISTEN 3351/haproxy

[root@haproxy-02 ~]# netstat -nplt

Active Internet connections (only servers)

Proto Recv-Q Send-Q Local Address Foreign Address State PID/Program name

tcp 0 0 0.0.0.0:3306 0.0.0.0:* LISTEN 3355/haproxy

查看这两个节点获取IP的情况

[root@haproxy-01 ~]# ip addr show eno16777736
2: eno16777736: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
    link/ether 00:0c:29:03:cc:88 brd ff:ff:ff:ff:ff:ff
    inet 172.29.28.193/22 brd 172.29.31.255 scope global dynamic eno16777736
       valid_lft 85737sec preferred_lft 85737sec
    inet 172.29.28.253/32 scope global eno16777736
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fe03:cc88/64 scope link 
       valid_lft forever preferred_lft forever

[root@haproxy-01 ~]# ip addr show eno16777736

2: eno16777736: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000

link/ether 00:0c:29:03:cc:88 brd ff:ff:ff:ff:ff:ff

inet 172.29.28.193/22 brd 172.29.31.255 scope global dynamic eno16777736

valid_lft 85737sec preferred_lft 85737sec

inet 172.29.28.253/32 scope global eno16777736

valid_lft forever preferred_lft forever

inet6 fe80::20c:29ff:fe03:cc88/64 scope link

valid_lft forever preferred_lft forever

[root@haproxy-02 ~]# ip addr show eno16777736
2: eno16777736: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
    link/ether 00:0c:29:d7:d3:ae brd ff:ff:ff:ff:ff:ff
    inet 172.29.28.195/22 brd 172.29.31.255 scope global dynamic eno16777736
       valid_lft 85000sec preferred_lft 85000sec
    inet 172.29.28.254/32 scope global eno16777736
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fed7:d3ae/64 scope link 
       valid_lft forever preferred_lft forever

[root@haproxy-02 ~]# ip addr show eno16777736

2: eno16777736: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000

link/ether 00:0c:29:d7:d3:ae brd ff:ff:ff:ff:ff:ff

inet 172.29.28.195/22 brd 172.29.31.255 scope global dynamic eno16777736

valid_lft 85000sec preferred_lft 85000sec

inet 172.29.28.254/32 scope global eno16777736

valid_lft forever preferred_lft forever

inet6 fe80::20c:29ff:fed7:d3ae/64 scope link

valid_lft forever preferred_lft forever

通过这两个节点的的IP可以看出，haproxy-01节点担任172.29.28.253的主，haproxy-02担任172.29.28.254的主，达到了负载均衡的效果。

三、测试

3.1 测试负载均衡

先来测试双VIP负载均衡效果，随便找个机器使用mysql客户端分别连接VIP地址，haproxy使用的调度算法是轮训，所以应该是每次分配到一个不同的real server服务器，效果如下：

$ mysql -h 172.29.28.253 -P3306 -uhaproxy -p123456
mysql> show variables like 'port';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| port          | 3306  |
+---------------+-------+
1 row in set (0.00 sec)

$ mysql -h 172.29.28.253 -P3306 -uhaproxy -p123456
mysql> show variables like 'port';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| port          | 3307  |
+---------------+-------+
1 row in set (0.00 sec)

$ mysql -h 172.29.28.253 -P3306 -uhaproxy -p123456

mysql> show variables like 'port';

+---------------+-------+

| Variable_name | Value |

+---------------+-------+

| port | 3306 |

+---------------+-------+

1 row in set (0.00 sec)

$ mysql -h 172.29.28.253 -P3306 -uhaproxy -p123456

mysql> show variables like 'port';

+---------------+-------+

| Variable_name | Value |

+---------------+-------+

| port | 3307 |

+---------------+-------+

1 row in set (0.00 sec)

然后，我们也可以测试另一个VIP。

$ mysql -h 172.29.28.254 -P3306 -uhaproxy -p123456
mysql> show variables like 'port';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| port          | 3306  |
+---------------+-------+
1 row in set (0.00 sec)

$ mysql -h 172.29.28.254 -P3306 -uhaproxy -p123456

mysql> show variables like 'port';

+---------------+-------+

| Variable_name | Value |

+---------------+-------+

| port | 3306 |

+---------------+-------+

1 row in set (0.00 sec)

3.2 测试故障转移

我们关闭掉haproxy-01节点。

[root@haproxy-01 ~]# systemctl stop keepalived.service

1	[root@haproxy-01 ~]# systemctl stop keepalived.service

然后去看haproxy-02节点漂移过来的VIP。

[root@haproxy-02 ~]# ip addr show eno16777736
2: eno16777736: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
    link/ether 00:0c:29:d7:d3:ae brd ff:ff:ff:ff:ff:ff
    inet 172.29.28.195/22 brd 172.29.31.255 scope global dynamic eno16777736
       valid_lft 82015sec preferred_lft 82015sec
    inet 172.29.28.253/32 scope global eno16777736
       valid_lft forever preferred_lft forever
    inet 172.29.28.254/32 scope global eno16777736
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fed7:d3ae/64 scope link 
       valid_lft forever preferred_lft forever

[root@haproxy-02 ~]# ip addr show eno16777736

2: eno16777736: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000

link/ether 00:0c:29:d7:d3:ae brd ff:ff:ff:ff:ff:ff

inet 172.29.28.195/22 brd 172.29.31.255 scope global dynamic eno16777736

valid_lft 82015sec preferred_lft 82015sec

inet 172.29.28.253/32 scope global eno16777736

valid_lft forever preferred_lft forever

inet 172.29.28.254/32 scope global eno16777736

valid_lft forever preferred_lft forever

inet6 fe80::20c:29ff:fed7:d3ae/64 scope link

valid_lft forever preferred_lft forever

现在我们可以看出，两个IP都漂移到了haproxy-02节点上了，再把haproxy-01打开后，172.29.28.100又会飘回去。

并且在这期间我一直ping主节点的VIP，在切换过程中很快，只丢了两个包。

64 bytes from 172.29.28.254: icmp_seq=12 ttl=64 time=0.398 ms
64 bytes from 172.29.28.254: icmp_seq=15 ttl=64 time=0.061 ms
..........
--- 172.29.28.254 ping statistics ---
44 packets transmitted, 42 received, 4% packet loss, time 43009ms

64 bytes from 172.29.28.254: icmp_seq=12 ttl=64 time=0.398 ms

64 bytes from 172.29.28.254: icmp_seq=15 ttl=64 time=0.061 ms

..........

--- 172.29.28.254 ping statistics ---

44 packets transmitted, 42 received, 4% packet loss, time 43009ms

四、状态页面

在haproxy配置文件中提供了haproxy web，用来查看haproxy状态的。

访问：http://172.29.28.193:8080/haproxyadmin?stats

这个界面提供了很多信息，具体可以查询相关资料。

五、测试HAproxy对后端健康状态检查

在haproxy配置时，有这么两条记录。

server mysql_01 10.99.73.10:3306 check inter 3000 rise 1 maxconn 4000 fall 3
server mysql_02 10.99.73.10:3307 check inter 3000 rise 1 maxconn 4000 fall 3

1 2	server mysql_01 10.99.73.10:3306 check inter 3000 rise 1 maxconn 4000 fall 3 server mysql_02 10.99.73.10:3307 check inter 3000 rise 1 maxconn 4000 fall 3

将此server命名为mysql_01，每隔3000ms检测一个健康状态，如果检测3次都失败，将此server剔除，在离线的状态下，只要检测1次成功，就让其上线。

下面就来测试这些规则，首先找一台mysql客户端一直跑这个脚本，主要测试后端负载均衡。

while true;do
    mysql -h 172.29.28.253 -P3306 -uhaproxy -p123456 -e "show variables like 'port';" -Bn | grep port
done

while true;do

mysql -h 172.29.28.253 -P3306 -uhaproxy -p123456 -e "show variables like 'port';" -Bn | grep port

done

然我们在后端（MySQL）主机上，添加一条iptables模拟3306端口故障。

# 添加拒绝访问3306端口的规则;
$ iptables -A INPUT -p tcp --dport 3306 -j REJECT

# 删除拒绝访问3306端口的规则;
$ iptables -D INPUT -p tcp --dport 3306 -j REJECT

# 添加拒绝访问3306端口的规则;

$ iptables -A INPUT -p tcp --dport 3306 -j REJECT

# 删除拒绝访问3306端口的规则;

$ iptables -D INPUT -p tcp --dport 3306 -j REJECT

查看状态页面，会发现haproxy已经剔除了3306端口，从状态页面可以看出Chk看到向后端检查了3次，Dwntme表示宕机的时间。

最后可以当一台后端MySQL故障后，看脚本执行情况。

六、为keepalived提供检测haproxy的脚本

通过前面的keepalived故障模拟可以看出，当keepalived挂掉之后，或者当主机宕机时，VIP是可以切换到备机工作。但是如果keepalived没有挂，并且主机也没有宕机，而是haproxy自身挂掉了呢？这个时候不会有任何切换动作，下面我们要做的就是为keepalived提供一个脚本用来检测haproxy是否正常工作，如果没有正常工作可以再次启动haproxy，或者关闭keepalived自身。这样一来，keepalived自身挂掉了，那么VIP就可以切换到备机上工作了。

如果您觉得本站对你有帮助，那么可以支付宝扫码捐助以帮助本站更好地发展，在此谢过。