采用MySQL-MMM做DB高可用时,遇到的一个小坑

一、服务器分布

wKioL1NeDCOBTrrmAAElNAi_h20772.jpg


二、MySQL-MMM 配置

(1)、公共配置【所有DB节点:Master1Master2Slave1Slave2   Monitor节点

# vim /etc/mysql-mmm/mmm_common.conf

active_master_role          writer
                                                                   
<host default>
    cluster_interface       eth1
                                                                      
    pid_path                /var/run/mmm_agentd.pid
    bin_path                /usr/lib/mysql-mmm/
                                                                      
    replication_user        repl
    replication_password    repl123456
                                                                      
    agent_user              mmm_agent
    agent_password          123456
</host>
                                                                   
<host db1>
    ip                      10.222.5.10
    mode                    master
    peer                    db2
</host>
                                                                   
<host db2>
    ip                      10.222.5.225
    mode                    master
    peer                    db1
</host>
                                                                   
<host db3>
    ip                      172.23.155.22
    mode                    slave
</host>
                                                                   
<host db4>
    ip                      172.24.18.68
    mode                    slave
</host>
                                                                   
<role writer>
    hosts                   db1,db2
    ips                     10.222.5.224
    mode                    exclusive
</role>

(2)Agent节点【所有DB节点Master1Master2Slave1Slave2

Master1节点:

# vim/etc/mysql-mmm/mmm_agent.conf

include mmm_common.conf
this db1

Master2节点:

# vim/etc/mysql-mmm/mmm_agent.conf

include mmm_common.conf
this db2

Slave1节点:

# vim/etc/mysql-mmm/mmm_agent.conf

include mmm_common.conf
this db3

Slave2节点:

# vim/etc/mysql-mmm/mmm_agent.conf

include mmm_common.conf
this db4

(3)Monitor节点

# vim /etc/mysql-mmm/mmm_mon.conf

include mmm_common.conf
                                                           
<monitor>
    ip                      10.137.23.103
    pid_path                /var/run/mmm_mond.pid
    bin_path                /usr/lib/mysql-mmm/
    status_path             /var/lib/misc/mmm_mond.status
    ping_ips                10.222.5.10, 10.222.5.225,172.23.155.22,172.24.18.68
    auto_set_online         10
</monitor>
                                                           
<host default>
    monitor_user            mmm_monitor
    monitor_password        123456
</host>
                                                            
debug 0


三、问题描述及解决方案

1、问题描述

当Master1发生故障,导致DB不可用时,VIP会自动漂移到Master2上,以实现高可用。但出现了一个问题,由于ARP老化时间过长,导致漂移过去的VIP不可用,也无法ping通。也就是说,MySQL-MMM没有考虑到ARP老化时间过长的情况而采取强刷ARP的方式。


2、解决方案

方法一(手动命令强刷)

获取网关地址:

wKioL1NeB2XituvIAABnQfVKpPM036.jpg

# arping -I eth1 -c 3 -s10.222.5.224 10.222.0.1

wKioL1NeB6WjHBIUAAD8Rd8qGwY810.jpg


方法二(代码级别修正)

# vim /usr/share/perl5/vendor_perl/MMM/Agent/Helpers/Actions.pm

wKioL1NeCh6QnjwfAAE3AeyOHDU937.jpg


方法三(SHELL脚本方式)

# vim /data/scripts/refresh_vip.sh

#!/bin/sh
PATH=/sbin:/bin:/usr/sbin:/usr/bin:/usr/local/bin:/usr/local/sbin
                                                     
MMM_COMM_CONF="/etc/mysql-mmm/mmm_common.conf"
ETH_NAME=`awk '/cluster_interface/{print $2}' ${MMM_COMM_CONF}`
VIP_ADDR=`grep -A 2 '<role writer>' ${MMM_COMM_CONF} | awk '/ips/{print $2}'`
GETWAY_ADDR=`/sbin/route | awk '/default/ {print $2}'`
                                                     
if [[ -n `/sbin/ip addr show ${ETH_NAME} | grep ${VIP_ADDR}` ]]; then
    /sbin/arping -I ${ETH_NAME} -c 3 -s ${VIP_ADDR} ${GETWAY_ADDR} >/dev/null 2>&1
fi

需要添加的crontab信息【每10秒刷新一次】:

* * * * * sleep 10; /data/scripts/refresh_vip.sh >/dev/null 2>&1

* * * * * sleep 20; /data/scripts/refresh_vip.sh >/dev/null 2>&1

* * * * * sleep 30; /data/scripts/refresh_vip.sh >/dev/null 2>&1

* * * * * sleep 40; /data/scripts/refresh_vip.sh >/dev/null 2>&1

* * * * * sleep 50; /data/scripts/refresh_vip.sh >/dev/null 2>&1



你可能感兴趣的:(mysql-mmm)