Keepalived+HAProxy实现MySQL高可用负载均衡的配置

 Keepalived

由于在生产环境使用了mysqlcluster,需要实现高可用负载均衡,这里提供了keepalived+haproxy来实现.

keepalived主要功能是实现真实机器的故障隔离及负载均衡器间的失败切换.可在第3,4,5层交换.它通过VRRPv2(Virtual Router Redundancy Protocol) stack实现的.

Layer3: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。

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。

Layer5: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。

Software Design

keepalived启动后会有单个进程

8352 ?    Ss   0:00 /usr/sbin/keepalived
8353 ?    S   0:00 \_ /usr/sbin/keepalived
8356 ?    S   0:01 \_ /usr/sbin/keepalived

父进程:内存管理,子进程管理等等

子进程:VRRP子进程

子进程:Healthchecking 子进程
实例

2台mysqlcluster 10.1.6.203 master  10.1.6.205 backup

vip 10.1.6.173

目的访问10.1.6.173 3366端口 分别轮询通过haproxy转发到10.1.6.203 3306 和10.1.6.205 3306

mysqlcluster搭建参照之前博客,这里在2台机上安装keepalived

root@10.1.6.203:~# apt-get install keepalived
root@10.1.6.203:~# cat /etc/keepalived/keepalived.conf
vrrp_script chk_haproxy {
    script "killall -0 haproxy"  # verify the pid existance
    interval 2          # check every 2 seconds
    weight -2          # add 2 points of prio if OK
}

vrrp_instance VI_1 {
    interface eth1        # interface to monitor
    state MASTER
    virtual_router_id 51     # Assign one ID for this route
    priority 101         # 101 on master, 100 on backup
    nopreempt
    debug

    virtual_ipaddress {
        10.1.6.173
    }

    track_script {    #注意大括号空格
        chk_haproxy
    }

    notify_master /etc/keepalived/scripts/start_haproxy.sh #表示当切换到master状态时,要执行的脚本
    notify_fault /etc/keepalived/scripts/stop_keepalived.sh #故障时执行的脚本
    notify_stop  /etc/keepalived/scripts/stop_haproxy.sh #keepalived停止运行前运行notify_stop指定的脚本 }

VRRPD配置包括三个类:

• VRRP同步组(synchroization group)
• VRRP实例(VRRP Instance)
• VRRP脚本

这里使用了 VRRP实例, VRRP脚本

注意配置选项:

stat:指定instance(Initial)的初始状态，就是说在配置好后，这台服务器的初始状态就是这里指定的，但这里指定的不算，还是得要通过竞选通过优先级来确定，里如果这里设置为master，但如若他的优先级不及另外一台，那么这台在发送通告时，会发送自己的优先级，另外一台发现优先级不如自己的高，那么他会就回抢占为master

• interface：实例绑定的网卡，因为在配置虚拟IP的时候必须是在已有的网卡上添加的
• priority 101：设置本节点的优先级，优先级高的为master
• debug：debug级别
• nopreempt:设置为不抢占

vrrp_script chk_haproxy {
    script "killall -0 haproxy"  # verify the pid existance
    interval 2          # check every 2 seconds 脚本执行间隔
    weight -2          # add 2 points of prio if OK 脚本结果导致的优先级变更：2表示优先级+2；-2则表示优先级-2
}

然后在实例(vrrp_instance)里面引用，有点类似脚本里面的函数引用一样：先定义，后引用函数名

    track_script {
        chk_haproxy
    }

注意：VRRP脚本(vrrp_script)和VRRP实例(vrrp_instance)属于同一个级别

root@10.1.6.203:scripts# cat start_haproxy.sh
#!/bin/bash

sleep 5
get=`ip addr |grep 10.1.6.173 |wc -l`
echo $get >> /etc/keepalived/scripts/start_ha.log

if [ $get -eq 1 ]
then
    echo "`date +%c` success to get vip" >> /etc/keepalived/scripts/start_ha.log
    /usr/local/sbin/haproxy -f /etc/haproxy/haproxy.cfg
else
    echo "`date +%c` can not get vip" >> /etc/keepalived/scripts/start_ha.log
fi
root@10.1.6.203:scripts# cat stop_keepalived.sh
#!/bin/bash

pid=`pidof keepalived`
if [ $pid == "" ]
then
  echo "`date +%c` no keepalived process id" >> /etc/keepalived/scripts/stop_keep.log
else
  echo "`date +%c` will stop keepalived " >> /etc/keepalived/scripts/stop_keep.log
  /etc/init.d/keepalived stop
fi

/etc/init.d/keepalived stop

root@10.1.6.203:scripts# cat stop_haproxy.sh
#!/bin/bash

pid=`pidof haproxy`
echo "`date +%c` stop haproxy" >> /etc/keepalived/scripts/stop_ha.log
kill -9 $pid

同理配置10.1.6.205

root@10.1.6.205:~# cat /etc/keepalived/keepalived.conf
vrrp_script chk_haproxy {
  script "killall -0 haproxy"  # verify the pid existance
  interval 2          # check every 2 seconds
  weight 2           # add 2 points of prio if OK
}

vrrp_instance VI_1 {
  interface eth1        # interface to monitor
  state BACKUP
  virtual_router_id 51     # Assign one ID for this route
  priority 100         # 101 on master, 100 on backup
  virtual_ipaddress {
    10.1.6.173
  }

  track_script {
    chk_haproxy
  }

notify_master /etc/keepalived/scripts/start_haproxy.sh
notify_fault /etc/keepalived/scripts/stop_keepalived.sh
notify_stop /etc/keepalived/scripts/stop_haproxy.sh

}

 HAProxy

下面再介绍下haproxy

HAProxy是一款基于TCP(第四层)和HTTP(第七层)应用的代理软件,它也可作为负载均衡器.可以支持数以万计的并发连接.同时可以保护服务器不暴露到网络上,通过端口映射.它还自带监控服务器状态的页面.

安装haproxy

wget -O/tmp/haproxy-1.4.22.tar.gz http://haproxy.1wt.eu/download/1.4/src/haproxy-1.4.22.tar.gz
tar xvfz /tmp/haproxy-1.4.22.tar.gz -C /tmp/
cd /tmp/haproxy-1.4.22
make TARGET=linux26
make install

haproxy需要对每一个mysqlcluster服务器进行健康检查

1.在2台主机分别配置haproxy.cfg

root@10.1.6.203:scripts# cat /etc/haproxy/haproxy.cfg
global
    maxconn 51200 #默认最大连接数
    #uid 99
    #gid 99
    daemon    #以后台形式运行haproxy
    #quiet
    nbproc 1   #进程数量(可以设置多个进程提高性能)
    pidfile /etc/haproxy/haproxy.pid #haproxy的pid存放路径,启动进程的用户必须有权限访问此文件 

defaults
    mode tcp      #所处理的类别 (#7层 http;4层tcp )
    option redispatch  #serverId对应的服务器挂掉后,强制定向到其他健康的服务器
    option abortonclose #当服务器负载很高的时候，自动结束掉当前队列处理比较久的连接
    timeout connect 5000s  #连接超时
    timeout client 50000s #客户端超时
    timeout server 50000s  #服务器超时
    log 127.0.0.1 local0  #错误日志记录
    balance roundrobin  #默认的负载均衡的方式,轮询方式 

listen proxy
    bind 10.1.6.173:3366  #监听端口
    mode tcp        #http的7层模式
    option httpchk    #心跳检测的文件
    server db1 10.1.6.203:3306 weight 1 check port 9222 inter 12000 rise 3 fall 3   #服务器定义,check inter 12000是检测心跳频率 rise 3是3次正确认为服务器可用， fall 3是3次失败认为服务器不可用，weight代表权重
    server db2 10.1.6.205:3306 weight 1 check port 9222 inter 12000 rise 3 fall 3

listen haproxy_stats
    mode http
    bind 10.1.6.173:8888
    option httplog
    stats refresh 5s
    stats uri /status #网站健康检测URL，用来检测HAProxy管理的网站是否可以用，正常返回200，不正常返回503
    stats realm Haproxy Manager
    stats auth admin:p@a1SZs24 #账号密码

root@10.1.6.205:~$ cat /etc/haproxy/haproxy.cfg
global
    maxconn 51200
    #uid 99
    #gid 99
    daemon
    #quiet
    nbproc 1
    pidfile /etc/haproxy/haproxy.pid

defaults
    mode tcp
    option redispatch
    option abortonclose
    timeout connect 5000s
    timeout client 50000s
    timeout server 50000s
    log 127.0.0.1 local0
    balance roundrobin 

listen proxy
    bind 10.1.6.173:3366
    mode tcp
    option httpchk
    server db1 10.1.6.203:3306 weight 1 check port 9222 inter 12000 rise 3 fall 3
    server db2 10.1.6.205:3306 weight 1 check port 9222 inter 12000 rise 3 fall 3

listen haproxy_stats
    mode http
    bind 10.1.6.173:8888
    option httplog
    stats refresh 5s
    stats uri /status
    stats realm Haproxy Manager
    stats auth admin:p@a1SZs24

2.安装xinetd

root@10.1.6.203:~# apt-get install xinetd

3.在每个节点添加xinetd服务脚本和mysqlchk端口号

root@10.1.6.203:~# vim /etc/xinetd.d/mysqlchk
# default: on
# description: mysqlchk
service mysqlchk         #需要在servive定义
{
    flags      = REUSE
    socket_type   = stream
    port      = 9222
    wait      = no
    user      = nobody
    server     = /opt/mysqlchk
    log_on_failure += USERID
    disable     = no
    per_source   = UNLIMITED
    bind      = 10.1.6.173
}

root@10.1.6.203:~# vim /etc/services
mysqlchk    9222/tcp            # mysqlchk

4.编写mysqlchk监控服务脚本

root@10.1.6.203:~# ls -l /opt/mysqlchk
-rwxr--r-- 1 nobody root 1994 2013-09-17 11:27 /opt/mysqlchk
root@10.1.6.203:~# cat /opt/mysqlchk
#!/bin/bash
#
# This script checks if a mysql server is healthy running on localhost. It will
# return:
# "HTTP/1.x 200 OK\r" (if mysql is running smoothly)
# - OR -
# "HTTP/1.x 500 Internal Server Error\r" (else)
#
# The purpose of this script is make haproxy capable of monitoring mysql properly
#

MYSQL_HOST="localhost"
MYSQL_SOCKET="/var/run/mysqld/mysqld.sock"
MYSQL_USERNAME="mysqlchkusr"   #该账户密码需要在mysql里添加
MYSQL_PASSWORD="secret"
MYSQL_OPTS="-N -q -A"
TMP_FILE="/dev/shm/mysqlchk.$$.out"
ERR_FILE="/dev/shm/mysqlchk.$$.err"
FORCE_FAIL="/dev/shm/proxyoff"
MYSQL_BIN="/opt/mysqlcluster/mysql-cluster-gpl-7.2.6-linux2.6-x86_64/bin/mysql"
CHECK_QUERY="select 1"

preflight_check()
{
  for I in "$TMP_FILE" "$ERR_FILE"; do
    if [ -f "$I" ]; then
      if [ ! -w $I ]; then
        echo -e "HTTP/1.1 503 Service Unavailable\r\n"
        echo -e "Content-Type: Content-Type: text/plain\r\n"
        echo -e "\r\n"
        echo -e "Cannot write to $I\r\n"
        echo -e "\r\n"
        exit 1
      fi
    fi
  done
}

return_ok()
{
  echo -e "HTTP/1.1 200 OK\r\n"
  echo -e "Content-Type: text/html\r\n"
  echo -e "Content-Length: 43\r\n"
  echo -e "\r\n"
  echo -e "<html><body>MySQL is running.</body></html>\r\n"
  echo -e "\r\n"
  rm $ERR_FILE $TMP_FILE
  exit 0
}
return_fail()
{
  echo -e "HTTP/1.1 503 Service Unavailable\r\n"
  echo -e "Content-Type: text/html\r\n"
  echo -e "Content-Length: 42\r\n"
  echo -e "\r\n"
  echo -e "<html><body>MySQL is *down*.</body></html>\r\n"
  sed -e 's/\n$/\r\n/' $ERR_FILE
  echo -e "\r\n"
  rm $ERR_FILE $TMP_FILE
  exit 1
}
preflight_check
if [ -f "$FORCE_FAIL" ]; then
    echo "$FORCE_FAIL found" > $ERR_FILE
    return_fail;
fi
$MYSQL_BIN $MYSQL_OPTS --host=$MYSQL_HOST --socket=$MYSQL_SOCKET --user=$MYSQL_USERNAME --password=$MYSQL_PASSWORD -e "$CHECK_QUERY" > $TMP_FILE 2> $ERR_FILE
if [ $? -ne 0 ]; then
    return_fail;
fi
return_ok;

测试

2个节点开启keepalived(主节点会获得vip,自动拉起haproxy),xinetd

root@10.1.6.203:~# ip add
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
2: eth0: <BROADCAST,MULTICAST> mtu 1500 qdisc pfifo_fast state DOWN qlen 1000
  link/ether 00:26:b9:36:0f:81 brd ff:ff:ff:ff:ff:ff
  inet 211.151.105.186/26 brd 211.151.105.191 scope global eth0
3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
  link/ether 00:26:b9:36:0f:83 brd ff:ff:ff:ff:ff:ff
  inet 10.1.6.203/24 brd 10.1.6.255 scope global eth1
  inet 10.1.6.173/32 scope global eth1
4: eth2: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN qlen 1000
  link/ether 00:26:b9:36:0f:85 brd ff:ff:ff:ff:ff:ff
5: eth3: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN qlen 1000
  link/ether 00:26:b9:36:0f:87 brd ff:ff:ff:ff:ff:ff
root@10.1.6.203:~# netstat -tunlp | grep ha
tcp    0   0 10.1.6.173:3366   0.0.0.0:*        LISTEN   1042/haproxy
tcp    0   0 10.1.6.203:8888   0.0.0.0:*        LISTEN   1042/haproxy
udp    0   0 0.0.0.0:56562      0.0.0.0:*              1042/haproxy
root@10.1.6.203:~# netstat -tunlp | grep xine
tcp    0   0 10.1.6.203:9222   0.0.0.0:*        LISTEN   30897/xinetd
root@10.1.6.203:~# ps -ef | grep haproxy
root   1042   1 0 Sep17 ?    00:00:00 /usr/local/sbin/haproxy -f /etc/haproxy/haproxy.cfg

测试:

通过vip10.1.6.173 3366访问cluster数据库(注意账户dave权限需要加3个ip10.1.6.203,10.1.6.205,10.1.6.173)

root@10.1.6.203:mgm# mysql -udave -p -h 10.1.6.173 -P 3366
Enter password:
Welcome to the MySQL monitor. Commands end with ; or \g.
Your MySQL connection id is 1344316
Server version: 5.5.22-ndb-7.2.6-gpl-log MySQL Cluster Community Server (GPL)

Type 'help;' or '\h' for help. Type '\c' to clear the buffer.

mysql> show databases;
+--------------------+
| Database      |
+--------------------+
| information_schema |
| dave       |
| test        |
+--------------------+
3 rows in set (0.01 sec)

mysql>

手动分别使keepalive,haproxy,数据库挂掉.vip10.1.6.173会自动漂到10.1.6.205从上，并不影响vip的访问

通过vip,haproxy查看各节点状态

http://10.1.6.173:8888/status