作者：china-lg
http://www.china-lg.com
A(M)[192.168.33.11\192.168.43.11]->B(Backup)[192.168.33.13\192.168.43.13]->C(M/S)[192.168.33.15\192.168.43.15]->s1、s2....
公用IP：192.168.33.100
本例可实现以下功能：
一、实现mysql replication
A(M)-->C(M/S)-->s1、s2、s3....
性能：降低A服务器的负载
扩展性：可扩展到20台slave服务器。
二、实现实时备份、安全可靠功能
利用drbd(号称网络RAID)将A服务器与C服务器的数据进行实时备份
如果仅A服务器down掉了，通过heartbeatB服务器则会自动切换，变成A服务器的角色.
如果仅C服务器down掉了，则需要在B服务器上，进行手动切换，变成C服务器的角色.
如果A和C服务器都down掉了，则需要改进本例，方可解决问题.(如：再增加一台backup机器)
本例需要的软件包如下：
drbd-0.7.23-1.c4.x86_64.rpm
heartbeat-2.0.7-1.c4.x86_64.rpm
heartbeat-pils-2.0.7-1.c4.x86_64.rpm
heartbeat-stonith-2.0.7-1.c4.x86_64.rpm
kernel-module-drbd-2.6.9-42.0.10.ELsmp-0.7.23-1.el4.centos.x86_64.rpm
mysql-5.0.33.tar.gz
★1.分区：
在A、C机器分出/opt分区
在B机器上分出/opt1、/opt2分区
且大小相同/opt=/opt1=/op2
★2.安装mysql并进行mysql replication设定.
[mdbrw01 ~]#tar -zxf mysql-5.0.33.tar.gz
[mdbrw01 ~]#cd mysql-5.0.33
[mdbrw01 ~]#./configure --prefix=/usr/local/mysql --with-mysqld-ldflags=-all-static --with-mysqld-user=mysql --with-charset=cp932 --with-pthread CFLAGS=-O3 CXXFLAGS=-O3 CXX=gcc
[mdbrw01 ~]#make && make install
[mdbrw01 ~]#/usr/local/mysql/bin/mysql -A -e "grant REPLICATION SLAVE on *.* to [email=slaver@]slaver@"%[/email]" Identified by"slave;FLUSH PRIVILEGES;"
[mdbrw01 ~]# /usr/local/mysql/bin/mysqladmin shutdown
[mdbrw01 ~]# mv /usr/local/mysql /opt
[mdbrw01 ~]#ln -s /opt/mysql /usr/local/mysql
replication设定比较简单，在此就不说明了.
★3.分别在A、B、C机器上安装以下软件
drbd-0.7.23-1.c4.x86_64.rpm
kernel-module-drbd-2.6.9-42.0.10.ELsmp-0.7.23-1.el4.centos.x86_64.rpm
heartbeat-pils-2.0.7-1.c4.x86_64.rpm
heartbeat-stonith-2.0.7-1.c4.x86_64.rpm
heartbeat-2.0.7-1.c4.x86_64.rpm
(请按照顺序来进行安装)
==========配置drbd===============
★4.在A服务器上
[mdbrw01 ~]#vi /etc/drbd.conf
resource r0 {
protocol C;
startup {
degr-wfc-timeout 120;
}
disk {
on-io-error detach;
}
net {
}
syncer {
rate 10M;
group 1;
al-extents 257;
}
on mdbrw01{
device /dev/drbd0;
disk /dev/sda5;
address 192.168.43.11:7788;
meta-disk internal;
}
on mdbbk01
device /dev/drbd0;
disk /dev/sda5;
address 192.168.43.13:7788;
meta-disk internal;
}
}
启动Drbd并设置为Primary
[mdbrw01 ~]# /etc/init.d/drbd start
[mdbrw01 ~]#drbdadm -- --do-what-I-say primary all
建立专用块及文件系统
[mdbrw01 ~]#mknod /dev/drbd0 b 147 0
[mdbrw01 ~]#mkfs /dev/drbd0
在B服务器上
[mdbrwbak ~]#vi /etc/drbd.conf
resource r0 {
protocol C;
startup {
degr-wfc-timeout 120;
}
disk {
on-io-error detach;
}
net {
}
syncer {
group 1;
al-extents 257;
}
on mdbrw01{
device /dev/drbd0;
disk /dev/sda5;
address 192.168.43.11:7788;
meta-disk internal;
}
on mdbbk01{
device /dev/drbd0;
disk /dev/sda5;
address 192.168.43.13:7788;
meta-disk internal;
}
}
resource r1 {
protocol C;
incon-degr-cmd "echo '!DRBD! pri on incon-degr' | wall ; sleep 60 ; halt -f";
startup {
wfc-timeout 0;
degr-wfc-timeout 120;
}
disk {
on-io-error detach;
}
net {
}
syncer {
rate 4M;
group 1;
}
on mdbrw02{
device /dev/drbd1;
disk /dev/sda5;
address 192.168.43.15:7789;
meta-disk internal;
}
on mdbbk01 {
device /dev/drbd1;
disk /dev/sda6;
address 192.168.43.13:7789;
meta-disk internal;
}
}
启动Drbd
[mdbrwbak ~]# /etc/init.d/drbd start
建立专用块及文件系统
[mdbrwbak ~]#mknod /dev/drbd0 b 147 0
[mdbrwbak ~]#mknod /dev/drbd1 b 147 1
在C服务器上
[mdbrw02 ~]#vi /etc/drbd.conf
resource r0 {
protocol C;
startup {
degr-wfc-timeout 120;
}
disk {
on-io-error detach;
}
net {
}
syncer {
rate 10M;
group 1;
al-extents 257;
}
on mdbrw02{
device /dev/drbd1;
disk /dev/sda5;
address 192.168.43.15:7789;
meta-disk internal;
}
on mdbbk01{
device /dev/drbd1;
disk /dev/sda6;
address 192.168.43.13:7789;
meta-disk internal;
}
}
启动Drbd并设置为Primary
[mdbrw02 ~]# /etc/init.d/drbd start
[mdbrw02 ~]#drbdadm -- --do-what-I-say primary all
建立专用块及文件系统
[mdbrw02 ~]#mknod /dev/drbd1 b 147 1
[mdbrw02 ~]#mkfs /dev/drbd1
========配置heartbeat==========
分别在A、B服务器上创建authkeys和haresources文件
# vi /etc/ha.d/authkeys
auth 1
1 crc
#2 sha1 HI!
#3 md5 Hello!
[mdbrw01 ~]#chmod 600 /etc/ha.d/authkeys
[mdbrw01 ~]#vi /etc/ha.d/haresources
mdbrw01 IPaddr::192.168.33.100/24/eth1 mysql_umount mysql
[mdbrw01 ~]# cat /etc/ha.d/resource.d/mysql_umount
#!/bin/sh
unset LC_ALL; export LC_ALL
unset LANGUAGE; export LANGUAGE
prefix=/usr
exec_prefix=/usr
. /etc/ha.d/shellfuncs
case "$1" in
'start')
/sbin/drbdadm -- --do-what-I-say primary r0
#/sbin/drbdadm -- --do-what-I-say primary all
/bin/mount /dev/drbd0 /opt
;;
'pre-start')
;;
'post-start')
;;
'stop')
/bin/umount /opt
/sbin/drbdadm secondary r0
#/sbin/drbdadm secondary all
;;
'pre-stop')
;;
'post-stop')
;;
*)
echo "Usage: $0 { start | pre-start | post-start | stop | pre-stop | post-stop }"
;;
esac
exit 0
[mdbrw01 ~]# cat /etc/ha.d/resource.d/mysql
内容略..
[mdbrw01 ~]# cat /etc/ha.d/ha.cf
logfile /var/log/ha-log
logfacility local0
keepalive 625ms
deadtime 5
warntime 1250ms
initdead 30
udpport 699
bcast eth1 # Linux
auto_failback off
node mdbrw01
node mdbbk01
ping 192.168.33.1
respawn hacluster /usr/lib64/heartbeat/ipfail
=============启动heartbeat==========
分别在A、B服务器上
[mdbrw01 ~]#etc/init.d/heartbeat start
注意事项：
1.drbd.conf文件中每一个资源的配置需要相同.
2.在/opt mount状态下不能启动drbd
3.在drbd启动状态下添加、修改、删除drbd.conf文件中的资源，可能会导致drbd stop失败
4.A、B服务器上的ha.cf需要保持一致
奥运快报： 
Drbd+heartbeat+mysql replication来构建mysql的高可用性
