最近看了一下mysql5.7的MGR集群挺不错的，有单主和多主模式，于是乎搭建测试了一下效果还不错，我指的不错是搭建和维护方面都比较简单。网上绝大多数都是单主模式，当然我这里也是，为了加深印象，特意记录一下搭建过程，等以后再去尝试多主模式，相信大家现在数据库的瓶颈基本都是在写，读写分离虽然是一种可行的解决方案，但是如果数据量很大，写一样会有问题，虽然有些解决方案能部署多个主节点，能同时进行读写，但是脑裂又是一个严重的问题，所以这里MGR集群内置了自动化脑裂防护机制又得到了很多人的青睐，这里MGR简称MySQL Group Replication是MySQL官方于2016年12月推出的一个全新的高可用与高扩展的解决方案。注意本文这里不再阐述原理性的东西。
注意：我这里采用编译安装的方式，如果想简单直接yum安装mysql5.7也行，mysql编译安装需要的磁盘空间还是比较大的，一般在7G左右，所以要提前规划好，用三个节点比较接近生产环境，而且更直接清晰。
详细部署信息如下：
| 主机名 | IP地址 | 安装软件 | 用途 |
| apache | 192.168.2.25 | cmake、boost、mysql | 节点 |
| nginx | 192.168.2.26 | cmake、boost、mysql | 节点 |
| kibana | 192.168.2.30 | cmake、boost、mysql | 节点 |
1、三台机器准备工作
rpm -qa mysql mariadb
如果有则卸载即可！
写入hosts文件映射关系，集群用得到
192.168.2.25 apache
192.168.2.26 nginx
192.168.2.30 kibana
2、安装依赖包
yum install gcc gcc-c++ ncurses-devel -y
3、安装cmake，下载地址：https://cmake.org/download/
tar zxvf cmake-3.7.2.tar.gz
cd make-3.7.2
./configure
gmake && gmake install
4、安装boost，因为mysql5.7需要，注意这里下载版本是1_59_0和mysql版本是对应的，如果你的MySQL版本和我的不一样，不添加-DWITH_BOOST这个参数时它会报错告诉你需要下载boost的哪个版本。
tar zxvf boost_1_59_0.tar.gz
cp -r boost_1_59_0 /usr/local/boost
5、安装mysql5.7.17及初始化操作
groupadd mysql
useradd -M -s /sbin/nologin mysql -g mysql
tar zxvf mysql-5.7.17.tar.gz
cd mysql-5.7.17
cmake -DCMAKE_INSTALL_PREFIX=/data/mysql -DSYSCONFDIR=/etc -DDEFAULT_CHARSET=utf8 -DDEFAULT_COLLATION=utf8_general_ci -DWITH_EXTRA_CHARSETS=all -DWITH_BOOST=/usr/local/boost
make
make install
chown -R mysql.mysql /data/mysql
mv /etc/my.cnf /etc/my.cnf.default
cp /data/mysql/support-files/my-default.cnf /etc/my.cnf
/data/mysql/bin/mysqld –initialize –user=mysql –basedir=/data/mysql –datadir=/data/mysql/data //注意初始化会生成一个随机的密码，请牢记
echo “PATH=$PATH:/data/mysql/bin” >> /etc/profile
source /etc/profile
cp /data/mysql/support-files/mysql.server /etc/rc.d/init.d/mysqld
chmod +x /etc/rc.d/init.d/mysqld
以上步骤在三台机器上都需要执行
6、开始搭建MGR集群环境，修改第一个节点的my.cnf文件，内容如下：
# For advice on how to change settings please see
# http://dev.mysql.com/doc/refman/5.7/en/server-configuration-defaults.html
# *** DO NOT EDIT THIS FILE. It’s a template which will be copied to the
# *** default location during install, and will be replaced if you
# *** upgrade to a newer version of MySQL.
[mysqld]
# Remove leading # and set to the amount of RAM for the most important data
# cache in MySQL. Start at 70% of total RAM for dedicated server, else 10%.
# innodb_buffer_pool_size = 128M
# Remove leading # to turn on a very important data integrity option: logging
# changes to the binary log between backups.
# log_bin
# These are commonly set, remove the # and set as required.
basedir = /data/mysql
datadir = /data/mysql/data
port = 3306
socket = /data/mysql/data/mysql.sock
log-error = /data/mysql/data/mysqld.log
pid-file = /data/mysql/data/mysqld.pid
# Remove leading # to set options mainly useful for reporting servers.
# The server defaults are faster for transactions and fast SELECTs.
# Adjust sizes as needed, experiment to find the optimal values.
# join_buffer_size = 128M
# sort_buffer_size = 2M
# read_rnd_buffer_size = 2M
sql_mode=NO_ENGINE_SUBSTITUTION,STRICT_TRANS_TABLES
# Group Replication
server_id = 1
gtid_mode = ON
enforce_gtid_consistency = ON
master_info_repository = TABLE
relay_log_info_repository = TABLE
binlog_checksum = NONE
log_slave_updates = ON
log_bin = binlog
binlog_format= ROW
transaction_write_set_extraction = XXHASH64
loose-group_replication_group_name = ‘ce9be252-2b71-11e6-b8f4-00212844f856’
loose-group_replication_start_on_boot = off
loose-group_replication_local_address = ‘192.168.2.25:33061’
loose-group_replication_group_seeds =’192.168.2.25:33061,192.168.2.26:33061,192.168.2.30:33061′
loose-group_replication_bootstrap_group = off
[client]
socket = /data/mysql/data/mysql.sock
启动mysql服务
/etc/init.d/mysqld start
set sql_log_bin=0;
create user rpl_user@’%’;
grant replication slave on *.* to rpl_user@’%’ identified by ‘rpl_pass’;
flush privileges;
set sql_log_bin=1;
change master to master_user=’rpl_user’,master_password=’rpl_pass’ for channel ‘group_replication_recovery’;
install PLUGIN group_replication SONAME ‘group_replication.so’;
set global group_replication_bootstrap_group=ON;
start group_replication;
set global group_replication_bootstrap_group=OFF;
select * from performance_schema.replication_group_members;
显示结果如下：
如果出现ONLINE，说明正常，这就是主节点，再搭建两个从节点。
7、第二个节点加入集群，复制刚刚的第一个节点的主配置文件my.cnf，只需要修改两个地方就行，已经用红色标注
# For advice on how to change settings please see
# http://dev.mysql.com/doc/refman/5.7/en/server-configuration-defaults.html
# *** DO NOT EDIT THIS FILE. It’s a template which will be copied to the
# *** default location during install, and will be replaced if you
# *** upgrade to a newer version of MySQL.
[mysqld]
# Remove leading # and set to the amount of RAM for the most important data
# cache in MySQL. Start at 70% of total RAM for dedicated server, else 10%.
# innodb_buffer_pool_size = 128M
# Remove leading # to turn on a very important data integrity option: logging
# changes to the binary log between backups.
# log_bin
# These are commonly set, remove the # and set as required.
basedir = /data/mysql
datadir = /data/mysql/data
port = 3306
socket = /data/mysql/data/mysql.sock
log-error = /data/mysql/data/mysqld.log
pid-file = /data/mysql/data/mysqld.pid
# Remove leading # to set options mainly useful for reporting servers.
# The server defaults are faster for transactions and fast SELECTs.
# Adjust sizes as needed, experiment to find the optimal values.
# join_buffer_size = 128M
# sort_buffer_size = 2M
# read_rnd_buffer_size = 2M
sql_mode=NO_ENGINE_SUBSTITUTION,STRICT_TRANS_TABLES
# Group Replication
server_id = 2
gtid_mode = ON
enforce_gtid_consistency = ON
master_info_repository = TABLE
relay_log_info_repository = TABLE
binlog_checksum = NONE
log_slave_updates = ON
log_bin = binlog
binlog_format= ROW
transaction_write_set_extraction = XXHASH64
loose-group_replication_group_name = ‘ce9be252-2b71-11e6-b8f4-00212844f856’
loose-group_replication_start_on_boot = off
loose-group_replication_local_address = ‘192.168.2.26:33061’
loose-group_replication_group_seeds =’192.168.2.25:33061,192.168.2.26:33061,192.168.2.30:33061′
loose-group_replication_bootstrap_group = off
[client]
socket = /data/mysql/data/mysql.sock
第二个节点执行如下命令：
set sql_log_bin=0;
create user rpl_user@’%’;
grant replication slave on *.* to rpl_user@’%’ identified by ‘rpl_pass’;
set sql_log_bin=1;
change master to master_user=’rpl_user’,master_password=’rpl_pass’ for channel ‘group_replication_recovery’;
install plugin group_replication SONAME ‘group_replication.so’;
set global group_replication_allow_local_disjoint_gtids_join=ON;
start group_replication;
显示结果如下：
同理第三个节点加入操作方法也和第二个节点一样。
截图如下：
查询哪个是主节点：
从上图来看很明显apache主机是主节点。
测试步骤：
1、在主库上创建一个库，然后创建表，在两个从库上查询数据是否同步？
2、两个从库只能执行查询操作？
2、手动关闭主库，确认两个从库其中一个是否会变成主库？而且是MEMBER_ID第一个字母按优先级排列的接管主库？
日常维护步骤：
1、如果从库某一节点关闭
start group_replication;
2、如果所有的库都关闭后，第一个库作为主库首先执行
set global group_replication_bootstrap_group=ON;
start group_replication;
剩下的库直接执行即可！
start group_replication;
3、如果主库故障，会自动从两个从库选出一个主库，主库启动后再次执行如下命令后会变成从库
start group_replication;