文件夹
1、前言... 4
2、方案... 4
2.1、环境及软件... 4
2.2、IP规划... 4
2.3、架构图... 4
3、安装设置MYSQL半同步... 5
4、Keepalived实现MYSQL的高可用... 11
1、前言
近期研究了下高可用的东西,这里总结一下mysql主主双活的架构方案,总体上提高服务的高可用性。出现故障也不须要手动切换。提高总体的维护效率。
确定改造的话,仅仅须要让他们的程序中使用vip地址就能够,实现起来比較easy。
2、方案
本案例先使用两台linux做双机MASTER-MASTER高可用(兴许能够考虑增加仅仅读SLAVER,用于提高查询性能),採用MYSQL5.6.x的半同步实现数据复制和同步。使用keepalived来监控MYSQL和提供VIP及浮动。不论什么一台主机宕机都不会影响对外提供服务(vip能够浮动),保持服务的高可用。
注 keepalived之间使用vrrp组播方式通信使用的IP地址是192.168.150.201
2.1、环境及软件
操作系统:RedHat 5.5
Keepalived:keepalived-1.2.12.tar.gz
MYSQL:MySQL-server- mysql-5.0.77-4.el5_4.2.x86_64.rpm MySQL-client- mysql-5.0.77-4.el5_4.2.x86_64.rpm
2.2、IP规划
名称
IP
备注
VIP
192.168.150.201
由KEEPALIVED产生的虚拟IP,也是前端client使用的IP
MYSQL_DB_MASTER1
192.168.150.145
数据库1server
MYSQL_DB_MASTER2
192.168.150.146
数据库2server
2.3、架构图
3、安装设置MYSQL半同步
3.1、安装MYSQL
这里选择的是使用社区版的MYSQL,请在www.mysql.com上下载相应操作系统的安装文件,这里依据官方的建议使用RPM包方式安装。也能够进行yum安装。
安装
假设是默认Redhat5.5安装后,
# 使用yum 安装 yum install mysql
yum install mysql-server
……
默认安装是在/usr以下。所以my.cnf是在/usr/my.cnf,这里为了习惯使用方法,复制到/etc/文件夹下
# mv /usr/my.cnf /etc/
标准方式改动password
# mysqladmin -u root -p oldPassword password
# New password:
# Confirm new password:
或
# mysqladmin -u root -p password newPassword
SQL直接更新方式改动password:
初始化rootpassword常见问题及处理
error: 'Access denied for user 'root'@'localhost' (usingpassword: NO)'
# /etc/init.d/mysql stop
# mysqld_safe --user=mysql --skip-grant-tables--skip-networking &
# mysql -u root mysql
mysql> UPDATE user SETPassword=PASSWORD('newpassword') where USER='root';
mysql> FLUSH PRIVILEGES;
mysql> quit
# /etc/init.d/mysql restart
# mysql -uroot -p
Enter password: <输入新设的密码newpassword>
mysql>
设置网络訪问权限
MYSQL的用户訪问策略是须要指定用户从某IP能够訪问那些权限。
详细使用方法请參考grant命令。这里为了简单起见,直接设置root用户在整个局域内(192.168.150.*)都能够訪问数据库。
# mysql -uroot -p
******
mysql> GRANT ALL PRIVILEGES ON *.* TO root@'192.168.150.%' IDENTIFIEDBY 'password' WITH GRANT OPTION;
mysql>FLUSH PRIVILEGES;
这样设置后,就能够通过内网的其它机器訪问数据库了。
改动数据文件路径
mysql安装后。默认的数据存储路径在/var/lib/mysql下。
这里改动为/data/mysql
# service mysql stop
# mkdir /data/mysql
# cp -r /var/lib/mysql/* /data/mysql
# chown -R mysql:mysql /data/mysql
vi /etc/my.cnf
改动里面的參数
[mysqld]
datadir= /data/mysql
socket= /data/mysql/mysql.sock
# 顺便配置下MYSQL使用mysql用户启动
user=mysql
保存退出
为/data/mysql/mysql.sock建立指向/var/lib/mysql/mysql.sock的软连接
# ln -s /data/mysql/mysql.sock /var/lib/mysql/mysql.sock
启动MYSQL并測试
# service mysql start
# mysql -uroot -p
******
mysql>
最可能出现的问题(假设你是Redhat6.x或CentOS6.x):selinux 照成服务无法正常启动,报错:
Starting MySQL. ERROR! The server quit without updatingPID file (/data/mysql/xxxxx.pid).
验证问题方法:
# setenforce 0
# service mysql start
Starting MySQL. SUCCESS!
假设能正常启动,解决这个问题。就确定是selinux造成的问题了。
解决的方法:直接禁用
vim /etc/selinux/config
改动:SELINUX=disabled
保存后,重新启动操作系统生效
如 果出现故障,一般解决这个问题的方式是查看MYSQL的日志文件,一般提示会比較清楚,或者依据错误信息区GOOGLE都能够找到答案。日志文件:${datadir}/${hostname}.err, 本例中改动了数据文件夹。所以日志文件是:/data/mysql/acooly1.err
OK,本节安装MYSQL完毕,然后另外一台server的MYSQL安装全然同样(半同步的设置有小不同,那以下会说明)。请參照以上说明安装另外一台MYSQL服务(MYSQL_DB_MASTER2)。
3.2、设置MYSQL双向半同步
本方案中为了实现故障转移功能。在利用半同步复制能力的同一时候,对总体方案进行了调整例如以下:
1. MYSQL_DB_MASTER1 和MYSQL_DB_MASTER2互为主备。即MYSQL_DB_MASTER1时MYSQL_DB_MASTER2的 主。MYSQL_DB_MASTER2是MYSQL_DB_MASTER1的主,通过半同步实现双向的同步复制(注意:这里不会出现冲突,由于备机仅仅会同步不是本机server_id的bin-log日志。通过两台机器的server_id隔离须要同步的bin-log)
2. 通过Keepalived实现MYSQLserver的监控和VIP的浮动,保证同一时间仅仅有一台MYSQLserver可用。
3. 该方案中,能够兼容后期的扩容,增加多台SALVE与两台MASTER异步同步实现数据的仅仅读查询,实现读写分离,提高总体性能。
配置/etc/my.cnf
分别改动/etc/my.cnf配置,打开bin-log功能和设置server_id
vi /etc/my.cnf
/etc/my.conf代码
[mysqld]
# MYSQL_DB_MASTER1设置为1。MYSQL_DB_MASTER2设置为2
server_id=1
log_bin=mysql-bin
安 装官方文档说明,主(MYSQL_DB_MASTER1)须要设置rpl_semi_sync_master_enabled=1和rpl_semi_sync_master_timeout=1000。备(MYSQL_DB_MASTER2)须要设置 rpl_semi_sync_slave_enabled=1,这里由于是双向同步。我都没有设置,经測试,没有设置也没有关系(临时没有深究,兴许在研究下)。
配置MYSQL_DB_MASTER1与MYSQL_DB_MASTER2的主备同步
在MYSQL_DB_MASTER1上操作配置MYSQL_DB_MASTER1为MYSQL_DB_MASTER2的主。
Mysql_db_master1代码
# mysql-uroot -p
Enterpassword:
......
-- 建立复制用户replication,同意内网内其它机器(訪问本机全部数据库)
mysql> grant replication slave on *.* to'repdb01'@'192.168.150.%' identified by '123456';
QueryOK, 0 rows affected (0.00 sec)
-- 查看master的bin-log状态,用户设置SLAVE同步的起点
mysql>show master status;
+------------------+----------+--------------+------------------+-------------------+
| File | Position | Binlog_Do_DB |Binlog_Ignore_DB | Executed_Gtid_Set |
+------------------+----------+--------------+------------------+-------------------+
| mysql-bin.000013 | 120 | | | |
+------------------+----------+--------------+------------------+-------------------+
1 row in set (0.02 sec) </span>
这里注意记录下当前bin-log文件名:mysql-bin.000013和位置:120。备机设置的时候须要用到。
在MYSQL_DB_MASTER2上操作配置
Mysql_db_master2代码
--设置SLAVE的MASTER和開始同步的文件位置
mysql>change master tomaster_host='192.168.150.145',master_user='repdb01',master_password='123456',master_log_file='mysql-bin.000003',master_log_pos=120;
--启动同步
mysql>start slave;
mysql>show slave status\G;
***************************1. row ***************************
Slave_IO_State: Waiting formaster to send event
Master_Host:192.168.150.145
Master_User: repdb01
Master_Port: 3306
Connect_Retry: 10
Master_Log_File:mysql-bin.000013
Read_Master_Log_Pos: 120
Relay_Log_File: acooly2-relay-bin.000020
Relay_Log_Pos: 283
Relay_Master_Log_File:mysql-bin.000013
Slave_IO_Running: Yes
Slave_SQL_Running: Yes
……
1 row in set (0.02sec)</span>
Slave_IO_Running: Yes和Slave_SQL_Running:Yes表示SLAVE已经正常启动并监听MASTER的数据发送事件。处理同步。
配置MYSQL_DB_MASTER2与MYSQL_DB_MASTER1的主备同步
全部配置操作与上一节,仅仅是把主备关系交换。
OK
假设全部的配置成功,启动MASTER1和MASTER2两个数据库,分别从两端创建数据库。表和更新表数据,另外一边都会实时同步。
也能够在两台机器中分别配置/etc/my.cnf
210.146
[root@rac2 keepalive]# cat /etc/my.cnf
[mysqld]
datadir=/var/lib/mysql
socket=/var/lib/mysql/mysql.sock
user=mysql
# Default to using old password format for compatibilitywith mysql 3.x
# clients (those using the mysqlclient10 compatibilitypackage).
old_passwords=1
# Disabling symbolic-links is recommended to preventassorted security risks;
# to do so, uncomment this line:
# symbolic-links=0
old_passwords=1
lower_case_table_names=1
default-character-set=utf8
default-storage-engine=innodb
max_connect_errors = 100000
#innodb_buffer_pool_size= 8G
max_connections = 500
default-character-set=utf8
server-id=2
#log-bin=mysqlbin
innodb_flush_log_at_trx_commit=1
sync_binlog=1
init_connect='SET NAMES utf8'
log-bin=mysqlbin
master-host=192.168.150.145
master-user=repdb01
master-pass=123456
master-connect-retry=60
replicate-do-db=db01
replicate-do-db=spring
master-port=3306
slave-net-timeout=60
[mysqld_safe]
log-error=/var/log/mysqld.log
pid-file=/var/run/mysqld/mysqld.pid
[client]
default-character-set=utf8
210.145 上面的配置
[root@rac1 ~]# cat /etc/my.cnf
[mysqld]
datadir=/var/lib/mysql
socket=/var/lib/mysql/mysql.sock
user=mysql
# Default to using old password format for compatibilitywith mysql 3.x
# clients (those using the mysqlclient10 compatibilitypackage).
old_passwords=1
# Disabling symbolic-links is recommended to preventassorted security risks;
# to do so, uncomment this line:
# symbolic-links=0
old_passwords=1
lower_case_table_names=1
default-character-set=utf8
default-storage-engine=innodb
max_connect_errors = 100000
#innodb_buffer_pool_size= 8G
max_connections = 500
default-character-set=utf8
#binlog_format=mixed
server-id=1
master-host=192.168.150.146
master-user=repdb01
master-pass=123456
master-connect-retry=60
replicate-do-db=db01
replicate-do-db=spring
init_connect='SET NAMES utf8'
log-bin=mysqlbin
[mysqld_safe]
log-error=/var/log/mysqld.log
pid-file=/var/run/mysqld/mysqld.pid
[client]
default-character-set=utf8
两个配置完毕后重新启动数据库,配置完毕。
4、Keepalived实现MYSQL的高可用
前面已经完毕MYSQL的双向实时同步的配置。本节使用Keepalived实现MYSQL双机的高可用,故障转移。通过VIP实现同一时候仅仅有一台MYSQL提供服务。
4.1、配置KEEPALIVED
两台MYSQLserver上的配置基本同样(除了keepalived.conf的个别配置外),详细操作配置例如以下:
设置keepalived的主配置文件/etc/keepalived/keepalived.conf实现本机的MYSQL的监控和VIP浮动。
安装keepalived
tar –zxvf keepalived-1.1.20.tar.gz
cd keepalived-1.1.20
./configure --prefix=/usr/local/keepalived/
make
make install
ln -s /usr/local/keepalived/etc/keepalived /etc/
ln -s /usr/local/keepalived/etc/rc.d/init.d/keepalived/etc/rc.d/init.d/
ln -s /usr/local/keepalived/etc/sysconfig/keepalived/etc/sysconfig/
ln -s /usr/local/keepalived/bin/genhash /bin/
ln -s /usr/local/keepalived/sbin/keepalived /sbin/
configure时注意Use IPVS Framework、IPVS sync daemon support 、Use VRRP Framework要返回yes。否则无法关联ipvs功能
master1 上的配置:
145
vi /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
router_id Mysql-HA
}
# 定义MYSQL监控的脚本。每2秒监控一次,详细脚本请见以下
vrrp_scriptcheck_mysql {
script"/root/keepalive/keepalived_check_mysql.sh"
interval 2
}
vrrp_sync_groupVG1 {
group {
KEEPALIVED_MYSQL
}
}
vrrp_instanceKEEPALIVED_MYSQL {
# 注意:两台MYSQLserver都设置为BACKUP
stateBACKUP
interface eth0
virtual_router_id 20
# # MASTER1设置为100,则MASTER2设置为稍低,如:90
priority 100
advert_int 1
# 仅仅有MASTER1设置不抢夺控制权
nopreempt
authentication {
auth_type PASS
auth_pass abcd1234
}
track_script {
check_mysql
}
virtual_ipaddress {
192.168.150.201
}
}
146上面的配置
! Configuration File for keepalived
global_defs {
router_id Mysql-HA
}
#定义MYSQL监控的脚本,每2秒监控一次,详细脚本请见以下
vrrp_scriptcheck_mysql {
script"/root/keepalive/keepalived_check_mysql.sh"
interval 2
}
vrrp_sync_groupVG1 {
group {
KEEPALIVED_MYSQL
}
}
vrrp_instanceKEEPALIVED_MYSQL {
#注意:两台MYSQLserver都设置为BACKUP
stateBACKUP
interface eth0
virtual_router_id 20
#MASTER1设置为100 MASTER2设置为90
priority 90
advert_int 1
authentication {
auth_type PASS
auth_pass abcd1234
}
track_script {
check_mysql
}
virtual_ipaddress {
192.168.150.201
}
}
检查本机MYSQL服务是否正常,假设MYSQL服务不可用,则停止KEEPALIVED服务,由备机的KEEPALIVED接管并提供VIP(指向备机)
/root/keepalive/keepalived_check_mysql.sh代码
[root@rac2 keepalive]# cat /root/keepalive/keepalived_check_mysql.sh
#!/bin/bash
MYSQL=/usr/bin/mysql
MYSQL_HOST=localhost
#MYSQL_USER=root
#MYSQL_PASSWORD=root
LOG_FILE=/root/keepalive/check_mysql.log
CHECK_TIME=3
#mysqlis working MYSQL_OK is 1 , mysql down MYSQL_OK is 0
MYSQL_OK=1
functioncheck_mysql_helth (){
$MYSQL -h $MYSQL_HOST -e "showstatus;" >/dev/null 2>&1
if [ $? = 0 ] ;then
MYSQL_OK=1
else
MYSQL_OK=0
fi
return $MYSQL_OK
}
while [$CHECK_TIME -ne 0 ]
do
let "CHECK_TIME -= 1"
check_mysql_helth
if [ $MYSQL_OK = 1 ] ; then
CHECK_TIME=0
echo `date --date=today +"%Y-%m-%d%H:%M:%S"` - [INFO] - mysqlavailable: success[$MYSQL_OK] >> $LOG_FILE
exit 0
fi
if [ $MYSQL_OK -eq 0 ] && [$CHECK_TIME -eq 0 ]
then
/etc/init.d/keepalived stop
echo `date --date=today +"%Y-%m-%d%H:%M:%S"` - [INFO] - mysqlinvaild. keepalived stop. >> $LOG_FILE
exit 1
fi
sleep 1
done
4.2、验证故障转移
配置完毕后。分别启动两台server的mysql和keepalived服务
分别启动MYSQL_DB_MASTER1和MYSQL_DB_MASTER2
详细命令例如以下:
# service mysql start
Starting MySQL.......... SUCCESS!
# service keepalived start
Starting keepalived: [ OK ]
在MYSQL_DB_MASTER1查看VIP
Ip a
[root@rac1 keepalive]# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdiscnoqueue
link/loopback00:00:00:00:00:00 brd 00:00:00:00:00:00
inet127.0.0.1/8 scope host lo
inet6 ::1/128scope host
valid_lftforever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500qdisc pfifo_fast qlen 1000
link/etherbe:0c:3f:74:0f:b8 brd ff:ff:ff:ff:ff:ff
inet192.168.150.145/24 brd 192.168.150.255 scope global eth0
inet 192.168.150.201/32 scope global eth0
inet192.168.150.245/24 brd 192.168.150.255 scope global secondary eth0:2
inet6fe80::bc0c:3fff:fe74:fb8/64 scope link
valid_lftforever preferred_lft forever
3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500qdisc pfifo_fast qlen 1000
link/etherca:d9:c2:ab:89:c8 brd ff:ff:ff:ff:ff:ff
inet10.10.10.145/24 brd 10.10.10.255 scope global eth1
inet6fe80::c8d9:c2ff:feab:89c8/64 scope link
valid_lftforever preferred_lft forever
4: sit0: <NOARP> mtu 1480 qdisc noop
link/sit 0.0.0.0brd 0.0.0.0
210.201 在145 上面
从另外一台机器用vip地址登陆
mysql -h 192.168.150.201 -utest -p
查看时连接那一台机器
mysql> show variables like '%host%'
-> ;
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| hostname | rac1 |
+---------------+-------+
1 row in set (0.00 sec)
然后停止rac1上的mysql
依旧查看连接的状态
mysql> show variables like '%host%';
ERROR 2006 (HY000): MySQL server has gone away
No connection. Trying to reconnect...
Connection id: 4274
Current database: *** NONE ***
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| hostname | rac2 |
+---------------+-------+
1 row in set (3.09 sec)
mysql>
mysql> show variables like '%host%';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| hostname | rac2 |
+---------------+-------+
1 row in set (0.00 sec)
发现已经转移到mysql_db_master2上面
[root@rac1 keepalive]# service keepalived status
keepalived is stopped
查看keepalive的状态
以及mysql的检查日志
查看日志 [root@rac1 keepalive]# catcheck_mysql.log
2014-11-05 18:29:37 - [INFO] - mysql available:success[1]
2014-11-05 18:29:42 - [INFO] - mysql invaild.keepalived stop.
2014-11-05 18:29:43 - [INFO] - mysql invaild.keepalived stop.
发现服务已经停止
[root@rac2 keepalive]# ip a 在主机2上查看ip情况
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdiscnoqueue
link/loopback00:00:00:00:00:00 brd 00:00:00:00:00:00
inet127.0.0.1/8 scope host lo
inet6 ::1/128scope host
valid_lftforever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500qdisc pfifo_fast qlen 1000
link/ether42:1c:5e:e1:1c:3e brd ff:ff:ff:ff:ff:ff
inet192.168.150.146/24 brd 192.168.150.255 scope global eth0
inet 192.168.150.201/32 scope global eth0
能够看到VIP(210.201)已经切换到MYSQL_DB_MASTER2
从新启动MYSQL_DB_MASTER1的MYSQL服务和KEEPALIVED服务。可是这个时候是不会从新浮动到MASTER1的。由于我们配置 KEEPALIVED的时候都是配置的BACKUP。并且MASTER1配置了不抢占。
这与我们设计是相符的。(仅仅有等MASTER2出现问题的情况才从新有MASTER1接管)
这样的方式能够避免一台机器出现故障手动切换才干使服务恢复。提高系统的高可用性。
兴许多台双活实验进行中,敬请期待。