MySQL Fabric实验（三）HA与Sharding

实验步骤：

1. 安装虚拟机

        使用VirtualBox安装四个CentOS release 6.4虚拟机，安装Python 2.6或以上版本，关闭iptables和selinux。虚拟机和网卡说明如下表所示。

主机名	内部网络IP	说明
fab_connector	192.168.56.101	安装Fabric和MySQL，建立一个MySQL实例，使用缺省的3306端口，存储MySQL实例的状态和路由信息
fab1	192.168.56.102	安装MySQL，建立三个MySQL数据库实例，端口分别是3326、3327、3328
fab2	192.168.56.103	安装MySQL，建立三个MySQL数据库实例，端口分别是3326、3327、3328
fab3	192.168.56.104	安装MySQL，建立三个MySQL数据库实例，端口分别是3326、3327、3328

 

虚拟机名称	网卡	连接方式	说明
fab_connector	网卡1	网络地址转换（NAT）	用于虚拟机访问宿主机和外网
网卡2	桥接网卡（192.168.1.119）	用于宿主机访问虚拟机
网卡3	内部网络	用于Fabric组内互联
fab1	网卡1	网络地址转换（NAT）	用于虚拟机访问宿主机和外网
网卡2	内部网络	用于Fabric组内互联
fab2	网卡1	网络地址转换（NAT）	用于虚拟机访问宿主机和外网
网卡2	内部网络	用于Fabric组内互联
fab3	网卡1	网络地址转换（NAT）	用于虚拟机访问宿主机和外网
网卡2	内部网络	用于Fabric组内互联

        因为只是出于实验的目的，所以使用root用户安装MySQL和Fabric，下载的软件包分别是mysql-5.7.10-linux-glibc2.5-x86_64.tar和mysql-utilities-1.5.6.tar.gz。实验环境如下图所示。



2. 在全部四个虚拟机上安装MySQL

cd /root
tar xvf mysql-5.7.10-linux-glibc2.5-x86_64.tar
tar zxvf mysql-5.7.10-linux-glibc2.5-x86_64.tar.gz
ln -s mysql-5.6.13-linux-glibc2.5-x86_64 mysql
groupadd mysql
useradd -r -g mysql mysql
chown -R mysql .

3. 在fab_connector上安装Fabric

cd /root
tar zxvf mysql-utilities-1.5.6.tar.gz
cd mysql-utilities-1.5.6
sudo python setup.py install

4. 准备配置文件

（1）在fab_connector上准备如下3个配置文件

[root@fab_connector ~]# cat /etc/my_fabric_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/mysql
port=3306
socket=/var/lib/mysql/mysql.sock

[root@fab_connector ~]# cat /etc/my_fabric.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/mysql
socket=/var/lib/mysql/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
port=3306
report-host=fab_connector
report-port=3306
server-id=1
log-bin=fab-bin.log

[root@fab_connector ~]# cat /etc/mysql/fabric.cfg
[DEFAULT]
prefix =
sysconfdir = /etc
logdir = /var/log

[statistics]
prune_time = 3600

[logging]
url = file:///var/log/fabric.log
level = INFO

[storage]
auth_plugin = mysql_native_password
database = fabric
user = fabric
address = localhost:3306
connection_delay = 1
connection_timeout = 6
password = secret
connection_attempts = 6

[failure_tracking]
notification_interval = 60
notification_clients = 50
detection_timeout = 1
detection_interval = 6
notifications = 300
detections = 3
failover_interval = 0
prune_time = 3600

[servers]
restore_user = fabric
unreachable_timeout = 5
backup_password = secret
backup_user = fabric
user = fabric
restore_password = secret
password = secret

[connector]
ttl = 1

[protocol.xmlrpc]
disable_authentication = no
ssl_cert =
realm = MySQL Fabric
ssl_key =
ssl_ca =
threads = 5
user = admin
address = 192.168.1.119:32274
password = secret

[executor]
executors = 5

[sharding]
prune_limit = 10000
mysqldump_program = /root/mysql/bin/mysqldump
mysqlclient_program = /root/mysql/bin/mysql

[protocol.mysql]
disable_authentication = no
ssl_cert =
ssl_key =
ssl_ca =
user = admin
address = 192.168.1.119:32275
password = secret

（2）在fab1上准备如下6个配置文件

[root@fab_global_group ~]# cat /etc/my_group_1_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_1
port=3326
socket=/var/lib/group_1/mysql.sock
plugin_dir=/root/mysql-5.7.10-linux-glibc2.5-x86_64/lib/plugin

[root@fab_global_group ~]# cat /etc/my_group_2_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_2
port=3327
socket=/var/lib/group_2/mysql.sock
plugin_dir=/root/mysql-5.7.10-linux-glibc2.5-x86_64/lib/plugin

[root@fab_global_group ~]# cat /etc/my_group_3_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_3
port=3328
socket=/var/lib/group_3/mysql.sock
plugin_dir=/root/mysql-5.7.10-linux-glibc2.5-x86_64/lib/plugin

[root@fab_global_group ~]# cat /etc/my_group_1.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_1
port=3326
socket=/var/lib/group_1/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
report-host=fab_global_group
report-port=3326
server-id=11
log-bin=fab1a-bin.log
rpl_semi_sync_slave_enabled=1
rpl_semi_sync_master_enabled = 1
rpl_semi_sync_master_timeout = 1000

[root@fab_global_group ~]# cat /etc/my_group_2.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_2
port=3327
socket=/var/lib/group_2/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
report-host=fab_global_group
report-port=3327
server-id=12
log-bin=fab1a-bin.log
rpl_semi_sync_slave_enabled=1
rpl_semi_sync_master_enabled = 1
rpl_semi_sync_master_timeout = 1000

[root@fab_global_group ~]# cat /etc/my_group_3.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_3
port=3328
socket=/var/lib/group_3/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
report-host=fab_global_group
report-port=3328
server-id=13
log-bin=fab1a-bin.log
rpl_semi_sync_slave_enabled=1
rpl_semi_sync_master_enabled = 1
rpl_semi_sync_master_timeout = 1000

（3）在fab2上准备如下6个配置文件

[root@fab_group1 ~]# cat /etc/my_group_1_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_1
port=3326
socket=/var/lib/group_1/mysql.sock
plugin_dir=/root/mysql-5.7.10-linux-glibc2.5-x86_64/lib/plugin

[root@fab_group1 ~]# cat /etc/my_group_2_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_2
port=3327
socket=/var/lib/group_2/mysql.sock
plugin_dir=/root/mysql-5.7.10-linux-glibc2.5-x86_64/lib/plugin

[root@fab_group1 ~]# cat /etc/my_group_3_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_3
port=3328
socket=/var/lib/group_3/mysql.sock
plugin_dir=/root/mysql-5.7.10-linux-glibc2.5-x86_64/lib/plugin

[root@fab_group1 ~]# cat /etc/my_group_1.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_1
port=3326
socket=/var/lib/group_1/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
report-host=fab_group1
report-port=3326
server-id=21
log-bin=fab1a-bin.log
rpl_semi_sync_slave_enabled=1
rpl_semi_sync_master_enabled = 1
rpl_semi_sync_master_timeout = 1000

[root@fab_group1 ~]# cat /etc/my_group_2.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_2
port=3327
socket=/var/lib/group_2/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
report-host=fab_group1
report-port=3327
server-id=22
log-bin=fab1a-bin.log
rpl_semi_sync_slave_enabled=1
rpl_semi_sync_master_enabled = 1
rpl_semi_sync_master_timeout = 1000

[root@fab_group1 ~]# cat /etc/my_group_3.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_3
port=3328
socket=/var/lib/group_3/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
report-host=fab_group1
report-port=3328
server-id=23
log-bin=fab1a-bin.log
rpl_semi_sync_slave_enabled=1
rpl_semi_sync_master_enabled = 1
rpl_semi_sync_master_timeout = 1000

（4）在fab3上准备如下6个配置文件

[root@fab_group2 ~]# cat /etc/my_group_1_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_1
port=3326
socket=/var/lib/group_1/mysql.sock
plugin_dir=/root/mysql-5.7.10-linux-glibc2.5-x86_64/lib/plugin

[root@fab_group2 ~]# cat /etc/my_group_2_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_2
port=3327
socket=/var/lib/group_2/mysql.sock
plugin_dir=/root/mysql-5.7.10-linux-glibc2.5-x86_64/lib/plugin

[root@fab_group2 ~]# cat /etc/my_group_3_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_3
port=3328
socket=/var/lib/group_3/mysql.sock
plugin_dir=/root/mysql-5.7.10-linux-glibc2.5-x86_64/lib/plugin

[root@fab_group2 ~]# cat /etc/my_group_1.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_1
port=3326
socket=/var/lib/group_1/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
report-host=fab_group2
report-port=3326
server-id=31
log-bin=fab1a-bin.log
rpl_semi_sync_slave_enabled=1
rpl_semi_sync_master_enabled = 1
rpl_semi_sync_master_timeout = 1000

[root@fab_group2 ~]# cat /etc/my_group_2.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_2
port=3327
socket=/var/lib/group_2/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
report-host=fab_group2
report-port=3327
server-id=32
log-bin=fab1a-bin.log
rpl_semi_sync_slave_enabled=1
rpl_semi_sync_master_enabled = 1
rpl_semi_sync_master_timeout = 1000

[root@fab_group2 ~]# cat /etc/my_group_3.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_3
port=3328
socket=/var/lib/group_3/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
report-host=fab_group2
report-port=3328
server-id=33
log-bin=fab1a-bin.log
rpl_semi_sync_slave_enabled=1
rpl_semi_sync_master_enabled = 1
rpl_semi_sync_master_timeout = 1000

5. 在fab_connector上启动MySQL

mysqld --defaults-file=/etc/my_fabric_init.cnf --initialize
# 记下初始化生成的临时密码
chown -R mysql /var/lib/mysql
mysqld --defaults-file=/etc/my_fabric_init.cnf --user=mysql &
mysql -u root -p -h127.0.0.1
# 修改初始密码，添加fabric用户
ALTER USER USER() IDENTIFIED BY 'new_password';
CREATE USER 'fabric'@'localhost' IDENTIFIED BY 'secret'
GRANT ALL ON fabric.* TO 'fabric'@'localhost'";
# 重启MySQL
mysqladmin -u root --protocol=tcp -h127.0.0.1 -p shutdown
mysqld --defaults-file=/etc/my_fabric.cnf --user=mysql &

6. 在fab1、fab2、fab3每个虚拟机上启动3个MySQL实例（下面的命令在三个虚拟机上都要执行）

mkdir /var/lib/group_1
mkdir /var/lib/group_2
mkdir /var/lib/group_3

mysqld --defaults-file=/etc/my_group_1_init.cnf --initialize
mysqld --defaults-file=/etc/my_group_2_init.cnf --initialize
mysqld --defaults-file=/etc/my_group_3_init.cnf --initialize
# 记下初始化生成的临时密码

chown -R mysql /var/lib/group_1
chown -R mysql /var/lib/group_2
chown -R mysql /var/lib/group_3

# 修改初始密码，添加fabric用户
mysqld --defaults-file=/etc/my_group_1_init.cnf --user=mysql &
mysql -u root -p -h127.0.0.1 -P3326

ALTER USER USER() IDENTIFIED BY 'new_password';
CREATE USER 'fabric'@'%' IDENTIFIED BY 'secret';
GRANT ALL ON *.* TO 'fabric'@'%';
INSTALL PLUGIN rpl_semi_sync_master SONAME 'semisync_master.so';
INSTALL PLUGIN rpl_semi_sync_slave SONAME 'semisync_slave.so';

mysqld --defaults-file=/etc/my_group_2_init.cnf --user=mysql &
mysql -u root -p -h127.0.0.1 -P3327

ALTER USER USER() IDENTIFIED BY 'new_password';
CREATE USER 'fabric'@'%' IDENTIFIED BY 'secret';
GRANT ALL ON *.* TO 'fabric'@'%';
INSTALL PLUGIN rpl_semi_sync_master SONAME 'semisync_master.so';
INSTALL PLUGIN rpl_semi_sync_slave SONAME 'semisync_slave.so';

mysqld --defaults-file=/etc/my_group_3_init.cnf --user=mysql &
mysql -u root -p -h127.0.0.1 -P3328

ALTER USER USER() IDENTIFIED BY 'new_password';
CREATE USER 'fabric'@'%' IDENTIFIED BY 'secret';
GRANT ALL ON *.* TO 'fabric'@'%';
INSTALL PLUGIN rpl_semi_sync_master SONAME 'semisync_master.so';
INSTALL PLUGIN rpl_semi_sync_slave SONAME 'semisync_slave.so';

# 重启3个MySQL实例
mysqladmin -u root --protocol=tcp -h127.0.0.1 -P3326 -p shutdown
mysqladmin -u root --protocol=tcp -h127.0.0.1 -P3327 -p shutdown
mysqladmin -u root --protocol=tcp -h127.0.0.1 -P3328 -p shutdown
mysqld --defaults-file=/etc/my_group_1.cnf --user=mysql &
mysqld --defaults-file=/etc/my_group_2.cnf --user=mysql &
mysqld --defaults-file=/etc/my_group_3.cnf --user=mysql &

# 查看3个MySQL实例进程
ps -ef | grep mysql

7. 在fab_connector上启动Fabric

# 在状态存储（MySQL数据库实例）中建立Fabric库
mysqlfabric manage setup
# 以守护进程方式启动Fabric
mysqlfabric manage start --daemonize
# 检查fabric进程是否运行
mysqlfabric manage ping

8. 在fab_connector上配置HA

# 建立HA组
mysqlfabric group create global-group
mysqlfabric group create my_group1
mysqlfabric group create my_group2

# 在HA组中添加服务器
mysqlfabric group add global-group 192.168.56.102:3326
mysqlfabric group add global-group 192.168.56.103:3326
mysqlfabric group add global-group 192.168.56.104:3326

mysqlfabric group add my_group1 192.168.56.102:3327
mysqlfabric group add my_group1 192.168.56.103:3327
mysqlfabric group add my_group1 192.168.56.104:3327

mysqlfabric group add my_group2 192.168.56.102:3328
mysqlfabric group add my_group2 192.168.56.103:3328
mysqlfabric group add my_group2 192.168.56.104:3328

# 提升主服务器
mysqlfabric group promote global-group --slave_id=192.168.56.102:3326
mysqlfabric group promote my_group1 --slave_id=192.168.56.103:3327
mysqlfabric group promote my_group2 --slave_id=192.168.56.104:3328

# 至此HA特性配置完毕，使用下面的命令检查Fabric配置运行情况
# 查看HA组
mysqlfabric group lookup_groups

# 查看HA组中的服务器
mysqlfabric group lookup_servers global-group

mysqlfabric group lookup_servers my_group1

mysqlfabric group lookup_servers my_group2

# 查看组的健康状况
mysqlfabric group health global-group

mysqlfabric group health my_group1

mysqlfabric group health my_group2

9. 在fab_connector上配置Sharding

# 定义分片映射方法
mysqlfabric sharding create_definition HASH global-group
# 添加分片表及其字段
mysqlfabric sharding add_table 1 test.chat_message src_userid
# 添加分片
mysqlfabric sharding add_shard 1 "my_group1,my_group2" --state=enabled

# 至此Sharding特性配置完毕，使用下面的S命令检查Sharding配置情况
mysql -u root -p -h127.0.0.1 -e "select * from fabric.shard_maps"

mysql -u root -p -h127.0.0.1 -e "select * from fabric.shard_ranges"

mysql -u root -p -h127.0.0.1 -e "select * from fabric.shard_tables"

mysql -u root -p -h127.0.0.1 -e "select * from fabric.shards"

10. 在fab1上的global-group组上建立测试表

mysql -u root -p -h127.0.0.1 -P3326
create database test;
use test;
CREATE TABLE chat_message (
src_userid bigint(20) NOT NULL PRIMARY KEY
);

执行完后在fab1、fab2、fab3的9个实例中都已经建好了测试库表。

192.168.56.102上3326复制到192.168.56.103上3326（global-group HA组内复制）

192.168.56.102上3326复制到192.168.56.104上3326（global-group HA组内·复制）

192.168.56.102上3326复制到192.168.56.103:3327（global-group向my_group1复制）

192.168.56.102上3326复制到192.168.56.104:3328（global-group向my_group2复制）

192.168.56.103:3327复制到192.168.56.102上3327（my_group1 HA组内复制）

192.168.56.103:3327复制到192.168.56.104上3327（my_group1 HA组内复制）

192.168.56.104:3328复制到192.168.56.102上3328（my_group2 HA组内复制）

192.168.56.104:3328复制到192.168.56.103上3328（my_group2 HA组内复制）

11. 执行Java应用程序，向测试表中插入20条数据，Java程序代码如下。

import java.io.ByteArrayInputStream;
import java.io.InputStream;
import java.io.UnsupportedEncodingException;
import java.sql.Connection;
import java.sql.DriverManager;
import java.sql.PreparedStatement;
import java.sql.SQLException;

import org.apache.commons.lang3.RandomStringUtils;
import com.mysql.fabric.jdbc.FabricMySQLConnection;

public class Main {
private static String URL = "jdbc:mysql:fabric://192.168.1.119:32274/test?fabricShardTable=chat_message&fabricUsername=admin&fabri
cPassword=secret";
private static String USERNAME = "fabric";
private static String PWD = "secret";
private static int MAX = 20;
private static String SQL = "insert into chat_message(src_userid) values(?)";

public static void main(String[] args) throws ClassNotFoundException, SQLException, UnsupportedEncodingException {
long start = System.currentTimeMillis();
testInsert();
long end = System.currentTimeMillis();
System.out.println((end - start));
System.out.println(MAX / ((end - start) / 1000));
}

private static Connection getConnection() throws SQLException, ClassNotFoundException {
Class.forName("com.mysql.fabric.jdbc.FabricMySQLDriver");
Connection con = DriverManager.getConnection(URL,USERNAME,PWD);
return con;
}

private static void testInsert() throws ClassNotFoundException, SQLException {
int i = 0;
Connection con = getConnection();
con.setAutoCommit(false);
while (i < MAX) {
FabricMySQLConnection fcon=(FabricMySQLConnection) con;
fcon.setShardKey(String.valueOf(i));
PreparedStatement pt = con.prepareStatement(SQL);
pt.setLong(1, i);
pt.executeUpdate();
fcon.commit();
i++;
//fcon.close();
System.out.println(i);
}
con.close();
}
}

12. 分别查询9个实例，检查执行结果

在192.168.56.102、192.168.56.103、192.168.56.104上执行，结果如下图所示。

mysql -u root -p -h127.0.0.1 -P3326 -e "select * from test.chat_message"



在192.168.56.102、192.168.56.103、192.168.56.104上执行，结果如下图所示。

mysql -u root -p -h127.0.0.1 -P3327 -e "select * from test.chat_message"



在192.168.56.102、192.168.56.103、192.168.56.104上执行，结果如下图所示。

mysql -u root -p -h127.0.0.1 -P3328 -e "select * from test.chat_message"



        从结果可以看到，global-group组中没有数据，my_group1和my_group2组中按Fabric内置的HASH算法分别插入了9条和11条数据，并且复制到了HA组内其它服务器中。
13. 在fab2、fab3两个虚拟机上新各建1个MySQL实例，端口为3329，分别用于测试remove_shard、split_shard

（1）准备配置文件
# 在fab2上准备如下2个配置文件
# cat /etc/my_group_4_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_4
port=3329
socket=/var/lib/group_4/mysql.sock
plugin_dir=/root/mysql-5.7.10-linux-glibc2.5-x86_64/lib/plugin

# cat /etc/my_group_4.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_4
port=3329
socket=/var/lib/group_4/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
report-host=fab_group1
report-port=3329
server-id=24
log-bin=fab1a-bin.log
rpl_semi_sync_slave_enabled=1
rpl_semi_sync_master_enabled = 1
rpl_semi_sync_master_timeout = 1000

# 在fab3上准备如下2个配置文件
# cat /etc/my_group_4_init.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_4
port=3329
socket=/var/lib/group_4/mysql.sock
plugin_dir=/root/mysql-5.7.10-linux-glibc2.5-x86_64/lib/plugin

# cat /etc/my_group_4.cnf
[mysqld]
basedir=/root/mysql
datadir=/var/lib/group_4
port=3329
socket=/var/lib/group_4/mysql.sock
binlog-format=ROW
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
master-info-repository=TABLE
relay-log-info-repository=TABLE
sync-master-info=1
report-host=fab_group2
report-port=3329
server-id=34
log-bin=fab1a-bin.log
rpl_semi_sync_slave_enabled=1
rpl_semi_sync_master_enabled = 1
rpl_semi_sync_master_timeout = 1000 （2）启动实例（下面的命令在两个虚拟机上都要执行）

mkdir /var/lib/group_4
mysqld --defaults-file=/etc/my_group_4_init.cnf --initialize
# 记下初始化生成的临时密码
chown -R mysql /var/lib/group_4

# 修改初始密码，添加fabric用户
mysqld --defaults-file=/etc/my_group_4_init.cnf --user=mysql &
mysql -u root -p -h127.0.0.1 -P3329

ALTER USER USER() IDENTIFIED BY 'new_password';
CREATE USER 'fabric'@'%' IDENTIFIED BY 'secret';
GRANT ALL ON *.* TO 'fabric'@'%';
INSTALL PLUGIN rpl_semi_sync_master SONAME 'semisync_master.so';
INSTALL PLUGIN rpl_semi_sync_slave SONAME 'semisync_slave.so';

# 重启MySQL实例
mysqladmin -u root --protocol=tcp -h127.0.0.1 -P3329 -p shutdown
mysqld --defaults-file=/etc/my_group_4.cnf --user=mysql &

# 查看MySQL实例进程
ps -ef | grep mysql

14. 在fab_connector上配置HA

# 建立HA组
mysqlfabric group create my_group3
mysqlfabric group create my_group4

# 在HA组中添加服务器
mysqlfabric group add my_group3 192.168.56.103:3329
mysqlfabric group add my_group4 192.168.56.104:3329

# 提升主服务器
mysqlfabric group promote my_group3
mysqlfabric group promote my_group4

15. 测试move_shard

# 把分片1从my_group1移动到my_group3
mysqlfabric sharding move_shard 1 my_group3

执行前后的分片定义如下图所示。





        在192.168.56.103上执行，结果如下图所示。

        mysql -u root -p -h127.0.0.1 -P3327 -e "select * from test.chat_message"



        mysql -u root -p -h127.0.0.1 -P3329 -e "select * from test.chat_message"



        从结果可以看到，分片1中的数据复制到了my_group3中。move_shard缺省情况下实际上是执行了一次备份还原数据操作，而不是迁移数据，正如Fabric帮助中所描述的：

sharding move_shard shard_id group_id  [--update_only] [--synchronous]  

Move the shard represented by the shard_id to the destination group.

By default this operation takes a backup, restores it on the

destination     group and guarantees that source and destination

groups are synchronized     before pointing the shard to the new

group. If users just want to update     the state store and skip these

provisioning steps, the update_only     parameter must be set to true.

        现在my_group1中的数据还在，但其中已经没有分片了。此时插入新的数据，将会插入到my_group2、my_group3中。修改执行Java程序再插入20条新数据，然后查询数据。

        在192.168.56.103上执行，结果如下图所示。

        mysql -u root -p -h127.0.0.1 -P3328 -e "select * from test.chat_message"



        mysql -u root -p -h127.0.0.1 -P3329 -e "select * from test.chat_message"



        从图中可以看到，my_group3上的分片1中新增了16条数据，my_group2上的分片2中新增了4条数据。

16. 测试split_shard

# 把分片2分到my_group4组中
mysqlfabric sharding split_shard 2 my_group4

        执行后的分片定义如下图所示。



        从图中可以看出，分片2的定义没有了，被分成了分片3和4，分别存在于my_group2和my_group4组中。

        在192.168.56.104上执行，结果如下图所示。

        mysql -u root -p -h127.0.0.1 -P3328 -e "select * from test.chat_message"



        mysql -u root -p -h127.0.0.1 -P3329 -e "select * from test.chat_message"



        从图中可以看到，原来my_group2上分片2中的15条数据，现在被分布到my_group2和my_group4上，my_group2上有10条数据，my_group4上有5条数据。

当前限制

        当前版本的MySQL Fabric有如下限制：
分片对应用并不完全透明。应用不需要知道数据行存储在哪个服务器，也不需要关心数据存储的位置是否改变，但它在访问数据库时需要提供分片键。
所有的事务和查询被限制在分片表的单一分片和非分片表中。例如，多个分片的连接查询是不支持的。
连接器执行路由功能，这避免了基于代理的解决方案带来的额外延迟，但同时意味着需要Fabric连接器，当前支持PHP、Python和Java。
MySQL Fabric 本身并没有失败冗余，失败时需要重启。注意这并不表示是服务器池（HA或分片）的单点故障，因为当MySQL Fabric进程失效时，连接器可以使用本地缓存继续路由操作。

参考：

MySQL Fabric官方文档
http://mysqlhighavailability.com/mysql-fabric-adding-high-availability-and-scaling-to-mysql/