之前已经详细介绍了，下面整体记录下MySQL5.7基于GTID主从复制+并行复制+增强半同步复制+读写分离环境的实现过程，以便加深对mysql新特性GTID复制的理解和实际业务场景中部署应用。
一、实验环境
[root@mysql-master ~]# cat /etc/redhat-release CentOS Linux release 7.5.1804 (Core) 为了方便实验，关闭所有节点的防火墙[root@mysql-master ~]# systemctl stop firewalld[root@mysql-master ~]# firewall-cmd --statenot running[root@mysql-master ~]# cat /etc/sysconfig/selinux |grep "SELINUX=disabled"SELINUX=disabled[root@mysql-master ~]# setenforce 0 setenforce: SELinux is disabled[root@mysql-master ~]# getenforce Disabled
二、安装Mysql5.7
在三个mysql节点机上使用yum方式安装Mysql5.7，参考：https://www.cnblogs.com/kevingrace/p/8340690.html 安装MySQL yum资源库[root@mysql-master ~]# yum localinstall https://dev.mysql.com/get/mysql57-community-release-el7-8.noarch.rpm 安装MySQL 5.7[root@mysql-master ~]# yum install -y mysql-community-server 启动MySQL服务器和MySQL的自动启动[root@mysql-master ~]# systemctl start mysqld.service[root@mysql-master ~]# systemctl enable mysqld.service 设置登录密码由于MySQL从5.7开始不允许首次安装后使用空密码进行登录！为了加强安全性，系统会随机生成一个密码以供管理员首次登录使用，这个密码记录在/var/log/mysqld.log文件中，使用下面的命令可以查看此密码：[root@mysql-master ~]# cat /var/log/mysqld.log|grep 'A temporary password'2019-01-11T05:53:17.824073Z 1 [Note] A temporary password is generated for root@localhost: TaN.k:*Qw2xs 使用上面查看的密码TaN.k:*Qw2xs 登录mysql，并重置密码为123456[root@mysql-master ~]# mysql -p #输入默认的密码：TaN.k:*Qw2xs.............mysql> set global validate_password_policy=0;Query OK, 0 rows affected (0.00 sec) mysql> set global validate_password_length=1;Query OK, 0 rows affected (0.00 sec) mysql> set password=password("123456");Query OK, 0 rows affected, 1 warning (0.00 sec) mysql> flush privileges;Query OK, 0 rows affected (0.00 sec) 查看mysql版本[root@mysql-master ~]# mysql -p123456........mysql> select version();+-----------+| version() |+-----------+| 5.7.24 |+-----------+1 row in set (0.00 sec)=====================================================================温馨提示mysql5.7通过上面默认安装后，执行语句可能会报错：ERROR 1819 (HY000): Your password does not satisfy the current policy requirements这个报错与Mysql 密码安全策略validate_password_policy的值有关，validate_password_policy可以取0、1、2三个值：解决办法：set global validate_password_policy=0;set global validate_password_length=1; 三、配置基于GTID的主从复制
传统的基于binlog position复制的方式有个严重的缺点：如果slave连接master时指定的binlog文件错误或者position错误，会造成遗漏或者重复，
很多时候前后数据是有依赖性的，这样就会出错而导致数据不一致。从MYSQL5.6开始，mysql开始支持GTID复制。GTID的全称是global transaction id，表示的是全局事务ID。GTID的分配方式为uuid:trans_id，其中：uuid是每个mysql服务器都唯一的，记录在$datadir/auto.cnf中。如果复制结构中，任意两台服务器uuid重复的话(比如直接冷备份时，auto.conf中的内容是一致的)，在启动复制功能的时候会报错。这时可以删除auto.conf文件再重启mysqld。
基于GTID主从复制的优点大致有：
- 保证同一个事务在某slave上绝对只执行一次，没有执行过的gtid事务总是会被执行。- 不用像传统复制那样保证binlog的坐标准确，因为根本不需要binlog以及坐标。- 故障转移到新的master的时候很方便，简化了很多任务。- 很容易判断master和slave的数据是否一致。只要master上提交的事务在slave上也提交了，那么一定是一致的。- 当然，MySQL提供了选项可以控制跳过某些gtid事务，防止slave第一次启动复制时执行master上的所有事务而导致耗时过久。- 虽然对于row-based和statement-based的格式都能进行gtid复制，但建议采用row-based格式。基于GTID主从复制环境部署记录
1）mysql-master主数据库上的操作 在my.cnf文件中配置GTID主从复制[root@mysql-master ~]# cp /etc/my.cnf /etc/my.cnf.bak[root@mysql-master ~]# >/etc/my.cnf[root@mysql-master ~]# cat /etc/my.cnf[mysqld]datadir = /var/lib/mysqlsocket = /var/lib/mysql/mysql.sock symbolic-links = 0 log-error = /var/log/mysqld.logpid-file = /var/run/mysqld/mysqld.pid #GTID:server_id = 1gtid_mode = onenforce_gtid_consistency = on #binloglog_bin = mysql-binlog-slave-updates = 1binlog_format = rowsync-master-info = 1sync_binlog = 1 #relay logskip_slave_start = 1 配置后，记得重启mysql服务[root@mysql-master ~]# systemctl restart mysqld 登录mysql，并查看master状态， 发现多了一项"Executed_Gtid_Set "mysql> show master status;+-------------------+----------+--------------+------------------+-------------------+| File | Position | Binlog_Do_DB | Binlog_Ignore_DB | Executed_Gtid_Set |+-------------------+----------+--------------+------------------+-------------------+| mysql-bin.000001 | 154 | | | |+-------------------+----------+--------------+------------------+-------------------+1 row in set (0.00 sec) mysql> show global variables like '%uuid%';+---------------+--------------------------------------+| Variable_name | Value |+---------------+--------------------------------------+| server_uuid | 317e2aad-1565-11e9-9c2e-005056ac6820 |+---------------+--------------------------------------+1 row in set (0.00 sec) 查看确认gtid功能打开mysql> show global variables like '%gtid%';+----------------------------------+-------+| Variable_name | Value |+----------------------------------+-------+| binlog_gtid_simple_recovery | ON || enforce_gtid_consistency | ON || gtid_executed | || gtid_executed_compression_period | 1000 || gtid_mode | ON || gtid_owned | || gtid_purged | || session_track_gtids | OFF |+----------------------------------+-------+8 rows in set (0.00 sec) 查看确认binlog日志功能打开mysql> show variables like 'log_bin';+---------------+-------+| Variable_name | Value |+---------------+-------+| log_bin | ON |+---------------+-------+1 row in set (0.00 sec) 授权slave复制用户，并刷新权限mysql> grant replication slave,replication client on *.* to slave@'172.16.60.212' identified by "slave@123";Query OK, 0 rows affected, 1 warning (0.03 sec) mysql> flush privileges;Query OK, 0 rows affected (0.04 sec) mysql> show grants for slave@'172.16.60.212';+-------------------------------------------------------------------------------+| Grants for slave@172.16.60.212 |+-------------------------------------------------------------------------------+| GRANT REPLICATION SLAVE, REPLICATION CLIENT ON *.* TO 'slave'@'172.16.60.212' |+-------------------------------------------------------------------------------+1 row in set (0.00 sec) 再次查看master状态mysql> show master status;+-------------------+----------+--------------+------------------+------------------------------------------+| File | Position | Binlog_Do_DB | Binlog_Ignore_DB | Executed_Gtid_Set |+-------------------+----------+--------------+------------------+------------------------------------------+| mysql-bin.000001 | 622 | | | 317e2aad-1565-11e9-9c2e-005056ac6820:1-2 |+-------------------+----------+--------------+------------------+------------------------------------------+1 row in set (0.00 sec) 这里需要注意一下：启动配置之前，同样需要对从服务器进行初始化。对从服务器初始化的方法基本和基于日志点是相同的，只不过在启动了GTID模式后，在备份中所记录的就不是备份时的二进制日志文件名和偏移量了，而是记录的是备份时最后的GTID值。 需要先在主数据库机器上把目标库备份一下，假设这里目标库是kevin（为了测试效果，下面手动创建）mysql> show databases;+--------------------+| Database |+--------------------+| information_schema || mysql || performance_schema || sys |+--------------------+4 rows in set (0.00 sec) mysql> CREATE DATABASE kevin CHARACTER SET utf8 COLLATE utf8_general_ci;Query OK, 1 row affected (0.02 sec) mysql> use kevin;Database changedmysql> create table if not exists haha (id int(10) PRIMARY KEY AUTO_INCREMENT,name varchar(50) NOT NULL);Query OK, 0 rows affected (0.27 sec) mysql> insert into kevin.haha values(1,"congcong"),(2,"huihui"),(3,"grace"); Query OK, 3 rows affected (0.06 sec)Records: 3 Duplicates: 0 Warnings: 0 mysql> select * from kevin.haha;+----+----------+| id | name |+----+----------+| 1 | congcong || 2 | huihui || 3 | grace |+----+----------+3 rows in set (0.00 sec) 把kevin库备份出来[root@mysql-master ~]# mysqldump --single-transaction --master-data=2 --triggers --routines --databases kevin -uroot -p123456 > /root/kevin.sql 这里稍微注意下：mysql5.6使用mysqldump备份时，指定备份的具体库，使用--databasemysql5.7使用mysqldump备份时，指定备份的具体库，使用--databases [root@mysql-master ~]# ls /root/kevin.sql/root/kevin.sql[root@mysql-master ~]# cat /root/kevin.sql-- MySQL dump 10.13 Distrib 5.7.24, for Linux (x86_64)---- Host: localhost Database: kevin-- -------------------------------------------------------- Server version 5.7.24-log..........................---- GTID state at the beginning of the backup-- SET @@GLOBAL.GTID_PURGED='317e2aad-1565-11e9-9c2e-005056ac6820:1-5'; 然后把备份的/root/kevin.sql文件拷贝到mysql-slave1从数据库服务器上[root@mysql-master ~]# rsync -e "ssh -p22" -avpgolr /root/kevin.sql root@172.16.60.212:/root/ ==============================================================2）mysql-slave1从数据库上的操作 在my.cnf文件中配置GTID主从复制与主服务器配置大概一致，除了server_id不一致外，从服务器还可以在配置文件里面添加："read_only＝on" ,使从服务器只能进行读取操作，此参数对超级用户无效，并且不会影响从服务器的复制；[root@mysql-slave1 ~]# cp /etc/my.cnf /etc/my.cnf.bak[root@mysql-slave1 ~]# >/etc/my.cnf[root@mysql-slave1 ~]# vim /etc/my.cnf[mysqld]datadir = /var/lib/mysqlsocket = /var/lib/mysql/mysql.sock symbolic-links = 0 log-error = /var/log/mysqld.logpid-file = /var/run/mysqld/mysqld.pid #GTID:server_id = 2gtid_mode = onenforce_gtid_consistency = on #binloglog_bin = mysql-binlog-slave-updates = 1binlog_format = rowsync-master-info = 1sync_binlog = 1 #relay logskip_slave_start = 1read_only = on 配置完成后，记得重启mysql服务[root@mysql-slave1 ~]# systemctl restart mysqld 接着将主数据库目标库的备份数据kevin.sql导入到从数据库里[root@mysql-slave1 ~]# ls /root/kevin.sql/root/kevin.sql[root@mysql-slave1 ~]# mysql -p123456.........mysql> show databases;+--------------------+| Database |+--------------------+| information_schema || mysql || performance_schema || sys |+--------------------+4 rows in set (0.00 sec) mysql> source /root/kevin.sql; mysql> select * from kevin.haha;+----+----------+| id | name |+----+----------+| 1 | congcong || 2 | huihui || 3 | grace |+----+----------+3 rows in set (0.00 sec) 在从数据库里，使用change master 配置主从复制mysql> stop slave;Query OK, 0 rows affected, 1 warning (0.00 sec) mysql> change master to master_host='172.16.60.211',master_user='slave',master_password='slave@123',master_auto_position=1; Query OK, 0 rows affected, 2 warnings (0.26 sec) mysql> start slave;Query OK, 0 rows affected (0.02 sec) mysql> show slave status \G;*************************** 1. row *************************** Slave_IO_State: Waiting for master to send event Master_Host: 172.16.60.211 Master_User: slave Master_Port: 3306 Connect_Retry: 60 Master_Log_File: mysql-bin.000001 Read_Master_Log_Pos: 1357 Relay_Log_File: mysql-slave1-relay-bin.000002 Relay_Log_Pos: 417 Relay_Master_Log_File: mysql-bin.000001 Slave_IO_Running: Yes Slave_SQL_Running: Yes................................ Executed_Gtid_Set: 317e2aad-1565-11e9-9c2e-005056ac6820:1-5 Auto_Position: 1 由上面可知，mysql-slave1节点已经和mysql-master节点配置了主从同步关系 3） mysql-master主数据库上进行状态查看和测试测试插入 mysql> show master status;+-------------------+----------+--------------+------------------+------------------------------------------+| File | Position | Binlog_Do_DB | Binlog_Ignore_DB | Executed_Gtid_Set |+-------------------+----------+--------------+------------------+------------------------------------------+| mysql-bin.000001 | 1357 | | | 317e2aad-1565-11e9-9c2e-005056ac6820:1-5 |+-------------------+----------+--------------+------------------+------------------------------------------+1 row in set (0.00 sec) mysql> show slave hosts;+-----------+------+------+-----------+--------------------------------------+| Server_id | Host | Port | Master_id | Slave_UUID |+-----------+------+------+-----------+--------------------------------------+| 2 | | 3306 | 1 | 2c1efc46-1565-11e9-ab8e-00505688047c |+-----------+------+------+-----------+--------------------------------------+1 row in set (0.00 sec) mysql> insert into kevin.haha values(4,"beijing"),(5,"hefei"),(10,"xihu");Query OK, 3 rows affected (0.06 sec)Records: 3 Duplicates: 0 Warnings: 0 mysql> delete from kevin.haha where id<4;Query OK, 3 rows affected (0.10 sec) mysql> select * from kevin.haha;+----+---------+| id | name |+----+---------+| 4 | beijing || 5 | hefei || 10 | xihu |+----+---------+3 rows in set (0.00 sec) 4）mysql-slave1从数据库上查看mysql> select * from kevin.haha;+----+---------+| id | name |+----+---------+| 4 | beijing || 5 | hefei || 10 | xihu |+----+---------+3 rows in set (0.00 sec) 发现mysql-slave1从数据库已经将新插入的数据同步过来了，由此，基于GTID的主从同步复制关系已经正常部署完成了！
四、并行复制 (解决主从复制延迟问题)
一般Mysql主从复制有三个线程参与，都是单线程：Binlog Dump(主) -> IO Thread (从) -> SQL Thread(从)。复制出现延迟一般出在两个地方:
- SQL线程忙不过来 (可能需要应用数据量较大，可能和从库本身的一些操作有锁和资源的冲突；主库可以并发写，SQL线程不可以；主要原因)- 网络抖动导致IO线程复制延迟(次要原因)。MySQL主从复制延迟的解决办法：MySQL从5.6开始有了SQL Thread多个的概念，可以并发还原数据，即并行复制技术。并行复制的机制，是MySQL的一个非常重要的特性，可以很好的解决MySQL主从延迟问题！
在MySQL 5.6中，设置参数slave_parallel_workers = 4(>1)，即可有4个SQL Thread（coordinator线程）来进行并行复制，其状态为：Waiting for an evant from Coordinator。但是其并行只是基于Schema的，也就是基于库的。如果数据库实例中存在多个Schema，这样设置对于Slave复制的速度可以有比较大的提升。通常情况下单库多表是更常见的一种情形，那基于库的并发就没有卵用。其核心思想是：不同schema下的表并发提交时的数据不会相互影响，即slave节点可以用对relay log中不同的schema各分配一个类似SQL功能的线程，来重放relay log中主库已经提交的事务，保持数据与主库一致。
MySQL 5.6版本支持所谓的并行复制，但是其并行只是基于schema的，也就是基于库的。如果用户的MySQL数据库实例中存在多个schema，对于从机复制的速度的确可以有比较大的帮助。但是基于schema的并行复制存在两个问题：
1) crash safe功能不好做，因为可能之后执行的事务由于并行复制的关系先完成执行，那么当发生crash的时候，这部分的处理逻辑是比较复杂的。2) 最为关键的问题是这样设计的并行复制效果并不高，如果用户实例仅有一个库，那么就无法实现并行回放，甚至性能会比原来的单线程更差。而 单库多表是比多库多表更为常见的一种情形 。注意：mysql 5.6的MTS是基于库级别的并行，当有多个数据库时，可以将slave_parallel_workers设置为数据库的数量，为了避免新建库后来回修改，也可以将该参数设置的大一些。设置为库级别的事务时，不允许这样做，会报错。
在MySQL 5.7中，引入了基于组提交的并行复制（官方称为Enhanced Multi-threaded Slaves，即MTS），设置参数slave_parallel_workers>0并且global.slave_parallel_type＝‘LOGICAL_CLOCK’，即可支持一个schema下，slave_parallel_workers个的worker线程并发执行relay log中主库提交的事务。其核心思想：一个组提交的事务都是可以并行回放（配合binary log group commit）；slave机器的relay log中 last_committed相同的事务（sequence_num不同）可以并发执行。其中，变量slave-parallel-type可以有两个值：１）DATABASE 默认值，基于库的并行复制方式；２）LOGICAL_CLOCK，基于组提交的并行复制方式；
MySQL 5.7开启Enhanced Multi-Threaded Slave很简单，只需要在Slave从数据库的my.cnf文件中如下配置即可:
# slave slave-parallel-type=LOGICAL_CLOCK slave-parallel-workers=8 #一般建议设置4-8，太多的线程会增加线程之间的同步开销 master_info_repository=TABLE relay_log_info_repository=TABLE relay_log_recovery=ON
MySQL 5.7是基于组提交的并行复制，并且是支持"真正"的并行复制功能，这其中最为主要的原因：就是slave服务器的回放与主机是一致的, 即master服务器上是怎么并行执行的slave上就怎样进行并行回放。不再有库的并行复制限制，对于二进制日志格式也无特殊的要求（基于库的并行复制也没有要求）。
MySQL5.7的并行复制，期望最大化还原主库的并行度，实现方式是在binlog event中增加必要的信息，以便slave节点根据这些信息实现并行复制。MySQL5.7的并行复制建立在group commit的基础上，所有在主库上能够完成prepared的语句表示没有数据冲突，就可以在slave节点并行复制。
所以在并行复制环境中，除了在Slace从数据库中配置之外，还需要在Master主数据库上的my.cnf文件中添加binlog_group_commit配置，否则从库无法做到基于事物的并行复制:
binlog_group_commit_sync_delay = 100 binlog_group_commit_sync_no_delay_count = 10
binlog_group_commit_sync_delay，这个参数控制着日志在刷盘前日志提交要等待的时间，默认是0也就是说提交后立即刷盘，但是并不代表是关闭了组提交，当设置为0以上的时候，就允许多个事物的日志同时间一起提交刷盘，也就是我们说的组提交。组提交是并行复制的基础，我们设置这个值的大于0就代表打开了组提交的延迟功能，而组提交是默认开启的。最大值只能设置为1000000微妙。
binlog_group_commit_sync_no_delay_count ，这个参数表示我们在binlog_group_commit_sync_delay等待时间内，如果事物数达到这个参数的设定值，就会触动一次组提交，如果这个值设为0的话就不会有任何的影响。如果到达时间但是事物数并没有达到的话，也是会进行一次组提交操作的。MySQL 5.7并行复制的思想简单易懂，一言以蔽之： 一个组提交的事务都是可以并行回放 ，因为这些事务都已进入到事务的prepare阶段，则说明事务之间没有任何冲突（否则就不可能提交）。为了兼容MySQL 5.6基于库的并行复制，5.7引入了新的变量slave-parallel-type，其可以配置的值有：
- DATABASE：默认值，基于库的并行复制方式- LOGICAL_CLOCK：基于组提交的并行复制方式支持并行复制的GTID
如何知道事务是否在一组中，又是一个问题，因为原版的MySQL并没有提供这样的信息。在MySQL 5.7版本中，其设计方式是将组提交的信息存放在GTID中。那么如果用户没有开启GTID功能，即将参数gtid_mode设置为OFF呢？故MySQL 5.7又引入了称之为Anonymous_Gtid的二进制日志event类型，如:mysql> SHOW BINLOG EVENTS in 'mysql-bin.000003';+------------------+-----+----------------+-----------+-------------+-----------------------------------------------+| Log_name | Pos | Event_type | Server_id | End_log_pos | Info |+------------------+-----+----------------+-----------+-------------+-----------------------------------------------+| mysql-bin.000003 | 4 | Format_desc | 88 | 123 | Server ver: 5.7.7-rc-debug-log, Binlog ver: 4 || mysql-bin.000003 | 123 | Previous_gtids | 88 | 194 | f11232f7-ff07-11e4-8fbb-00ff55e152c6:1-2 || mysql-bin.000003 | 194 | Anonymous_Gtid | 88 | 259 | SET @@SESSION.GTID_NEXT= 'ANONYMOUS' || mysql-bin.000003 | 259 | Query | 88 | 330 | BEGIN || mysql-bin.000003 | 330 | Table_map | 88 | 373 | table_id: 108 (aaa.t) || mysql-bin.000003 | 373 | Write_rows | 88 | 413 | table_id: 108 flags: STMT_END_F |......
这意味着在 MySQL 5.7版本中即使不开启GTID，每个事务开始前也是会存在一个Anonymous_Gtid ，而这GTID中就存在着组提交的信息。
LOGICAL_CLOCK
然而，通过上述的SHOW BINLOG EVENTS，并没有发现有关组提交的任何信息。但是通过mysqlbinlog工具，用户就能发现组提交的内部信息:如下查看一个binlog日志
root@localhost:~# mysqlbinlog mysql-bin.0000002 | grep last_committed#190112 17:45:16 server id 1 end_log_pos 219 CRC32 0xca2ee8c2 GTID last_committed=0 sequence_number=1 rbr_only=yes#190112 17:45:21 server id 1 end_log_pos 506 CRC32 0xe8047dd2 GTID last_committed=1 sequence_number=2 rbr_only=yes#190112 17:45:16 server id 1 end_log_pos 219 CRC32 0xca2ee8c2 GTID last_committed=2 sequence_number=3 rbr_only=yes#190112 17:45:21 server id 1 end_log_pos 506 CRC32 0xe8047dd2 GTID last_committed=3 sequence_number=4 rbr_only=yes
可以发现较之原来的二进制日志内容多了last_committed和sequence_number， last_committed表示事务提交的时候，上次事务提交的编号，如果事务具有相同的last_committed，表示这些事务都在一组内，可以进行并行的回放 (一般是当执行的sql语句并发数大的情况下会进行组提交) 。上面这个binlog日志里没有组提交信息(last_committed数值都不相等)， 下一个事物的last_committed永远都和上一个事物的sequence_number是相等的，这是因为事物是顺序提交的！这么理解起来并不奇怪。
下面看一下组提交模式的事物：
root@localhost:~# mysqlbinlog mysql-bin.0000006 | grep last_committed#150520 14:23:11 server id 88 end_log_pos 259 CRC32 0x4ead9ad6 GTID last_committed=0 sequence_number=1#150520 14:23:11 server id 88 end_log_pos 1483 CRC32 0xdf94bc85 GTID last_committed=0 sequence_number=2#150520 14:23:11 server id 88 end_log_pos 2708 CRC32 0x0914697b GTID last_committed=0 sequence_number=3#150520 14:23:11 server id 88 end_log_pos 3934 CRC32 0xd9cb4a43 GTID last_committed=0 sequence_number=4#150520 14:23:11 server id 88 end_log_pos 5159 CRC32 0x06a6f531 GTID last_committed=0 sequence_number=5#150520 14:23:11 server id 88 end_log_pos 6386 CRC32 0xd6cae930 GTID last_committed=0 sequence_number=6#150520 14:23:11 server id 88 end_log_pos 7610 CRC32 0xa1ea531c GTID last_committed=6 sequence_number=7#150520 14:23:11 server id 88 end_log_pos 8834 CRC32 0x96864e6b GTID last_committed=6 sequence_number=8#150520 14:23:11 server id 88 end_log_pos 10057 CRC32 0x2de1ae55 GTID last_committed=6 sequence_number=9#150520 14:23:11 server id 88 end_log_pos 11280 CRC32 0x5eb13091 GTID last_committed=6 sequence_number=10#150520 14:23:11 server id 88 end_log_pos 12504 CRC32 0x16721011 GTID last_committed=6 sequence_number=11#150520 14:23:11 server id 88 end_log_pos 13727 CRC32 0xe2210ab6 GTID last_committed=6 sequence_number=12#150520 14:23:11 server id 88 end_log_pos 14952 CRC32 0xf41181d3 GTID last_committed=12 sequence_number=13...
例如上述last_committed为0的事务有6个，这意味什么呢？意味着这6个事物是作为一个组提交的，6个事物在perpare截断获取相同的last_committed而且相互不影响，最终是会作为一个组进行提交?，这就是所谓的组提交！上面表示组提交时提交了6个事务，而这6个事务在从机是可以进行并行回放的。
总之：MySQL 5.7推出的Enhanced Multi-Threaded Slave解决了困扰MySQL长达数十年的复制延迟问题，再次提醒一些无知的PostgreSQL用户，不要再停留在之前对于MySQL的印象，物理复制也不一定肯定比逻辑复制有优势，而MySQL 5.7的MTS已经完全可以解决延迟问题。总之， 5.7版本后，主从复制延迟问题将不存在！
基于GTID的并行复制环境部署记录
1）在mysql-slave1从数据库的my.cnf 文件中添加下面MTS并行复制的配置[root@mysql-slave1 ~]# vim /etc/my.cnf[mysqld]datadir = /var/lib/mysqlsocket = /var/lib/mysql/mysql.sock symbolic-links = 0 log-error = /var/log/mysqld.logpid-file = /var/run/mysqld/mysqld.pid #GTID:server_id = 2gtid_mode = onenforce_gtid_consistency = on #binloglog_bin = mysql-binlog-slave-updates = 1binlog_format = rowsync-master-info = 1sync_binlog = 1 #relay logskip_slave_start = 1read_only = on #MTSslave-parallel-type = LOGICAL_CLOCK #开启逻辑时钟的复制slave-parallel-workers = 4 #这里设置线程数为4 (最大线程数不能超过16，即最大线程为16)master_info_repository = TABLErelay_log_info_repository = TABLErelay_log_recovery = on 配置完成后，记得重启mysql服务[root@mysql-slave1 ~]# systemctl restart mysqld 2）在mysql-master主数据库的my.cnf 文件中添加最后两行配置[root@mysql-master ~]# vim /etc/my.cnf[mysqld]datadir = /var/lib/mysqlsocket = /var/lib/mysql/mysql.sock symbolic-links = 0 log-error = /var/log/mysqld.logpid-file = /var/run/mysqld/mysqld.pid #GTID:server_id = 1gtid_mode = onenforce_gtid_consistency = on #binloglog_bin = mysql-binlog-slave-updates = 1binlog_format = rowsync-master-info = 1sync_binlog = 1 #relay logskip_slave_start = 1 #不配置binlog_group_commit从库无法做到基于事物的并行复制binlog_group_commit_sync_delay = 100 binlog_group_commit_sync_no_delay_count = 10 #为了数据安全再配置sync_binlog=1innodb_flush_log_at_trx_commit =1 #这个参数控制binlog写入 磁盘的方式。设置为1时，表示每次commit;都写入磁盘。这个刷新的是redo log 即ib_logfile0，而不是binlog 配置完成后，记得重启mysql服务[root@mysql-master ~]# systemctl restart mysqld 登录mysql，查看binlog_group_commit mysql> show variables like 'binlog_group_commit_%';+-----------------------------------------+-------+| Variable_name | Value |+-----------------------------------------+-------+| binlog_group_commit_sync_delay | 100 || binlog_group_commit_sync_no_delay_count | 10 |+-----------------------------------------+-------+2 rows in set (0.00 sec) 设置binlog_group_commit的上面两个参数，否则从库无法做到基于事物的并行复制! 这两个参数共同决定了是否触发组提交操作！第二个参数表示该事务组提交之前总共等待累积到多少个事务(如上要累计到10个事务)；第一个参数则表示该事务组总共等待多长时间后进行提交(如上要总共等待100毫秒的时间)，任何一个条件满足则进行后续操作。因为有这个等待，可以让更多事务的binlog通过一次写binlog文件磁盘来完成提交，从而获得更高的吞吐量。 3）登录mysql-slave1从数据库 上面在配置并行复制后，主从数据库的mysqld服务都重启了，此时需要重启从数据库上的slave，这样才能恢复正常的主从同步状态！记住：只要主数据库的mysqld服务重启，那么从数据库上就要重启slave，以恢复主从同步状态！！！ mysql> show slave status \G; ...................... Slave_IO_Running: No Slave_SQL_Running: No mysql> stop slave;Query OK, 0 rows affected, 1 warning (0.00 sec) mysql> start slave;Query OK, 0 rows affected (0.10 sec) mysql> show slave status \G;.................... Slave_IO_Running: Yes Slave_SQL_Running: Yes.......... Executed_Gtid_Set: 317e2aad-1565-11e9-9c2e-005056ac6820:1-11 Auto_Position: 1 这样，mysql-slave1从数据库就恢复了与mysql-master主数据库的同步关系了 查看优化项mysql> use mysql; #切入到mysql库Database changedmysql> select * from slave_worker_info; #查看slave_worker_info表+----+---------------------------------+---------------+-------------------+----------------+---------------------------------+--------------------------+----------------------------+---------------------------+------------------+-----------------------+------------------------------------------------------------------+--------------+| Id | Relay_log_name | Relay_log_pos | Master_log_name | Master_log_pos | Checkpoint_relay_log_name | Checkpoint_relay_log_pos | Checkpoint_master_log_name | Checkpoint_master_log_pos | Checkpoint_seqno | Checkpoint_group_size | Checkpoint_group_bitmap | Channel_name |+----+---------------------------------+---------------+-------------------+----------------+---------------------------------+--------------------------+----------------------------+---------------------------+------------------+-----------------------+------------------------------------------------------------------+--------------+| 1 | ./mysql-slave1-relay-bin.000005 | 656 | mysql-bin.000008 | 481 | ./mysql-slave1-relay-bin.000005 | 369 | mysql-bin.000008 | 194 | 0 | 64 | | || 2 | | 0 | | 0 | | 0 | | 0 | 0 | 64 | | || 3 | | 0 | | 0 | | 0 | | 0 | 0 | 64 | | || 4 | | 0 | | 0 | | 0 | | 0 | 0 | 64 | | |+----+---------------------------------+---------------+-------------------+----------------+---------------------------------+--------------------------+----------------------------+---------------------------+------------------+-----------------------+------------------------------------------------------------------+--------------+4 rows in set (0.00 sec) 以上可以查看到，mysql库下的slave_worker_info表下查看到4个线程 也可以使用下面命令查看线程数，这个跟在my.cnf文件里配置的是一样的！mysql> show variables like '%slave_para%';+------------------------+---------------+| Variable_name | Value |+------------------------+---------------+| slave_parallel_type | LOGICAL_CLOCK || slave_parallel_workers | 4 |+------------------------+---------------+2 rows in set (0.00 sec) 4）同步复制测试在mysql-master主数据库插入新数据mysql> insert into kevin.haha values(21,"caiyi"),(22,"manman"),(23,"titi");Query OK, 3 rows affected (0.09 sec)Records: 3 Duplicates: 0 Warnings: 0 mysql> select * from kevin.haha;+----+---------+| id | name |+----+---------+| 4 | beijing || 5 | hefei || 10 | xihu || 21 | caiyi || 22 | manman || 23 | titi |+----+---------+6 rows in set (0.00 sec) 然后在mysql-slave1从数据库查看，发现新数据已经同步过来了mysql> select * from kevin.haha;+----+---------+| id | name |+----+---------+| 4 | beijing || 5 | hefei || 10 | xihu || 21 | caiyi || 22 | manman || 23 | titi |+----+---------+6 rows in set (0.00 sec)最后在mysql-master主数据库上查看组提交信息[root@mysql-master ~]# cd /var/lib/mysql[root@mysql-master mysql]# ll mysql-bin*-rw-r----- 1 mysql mysql 751 Jan 12 18:08 mysql-bin.000001-rw-r----- 1 mysql mysql 1365 Jan 12 19:28 mysql-bin.000002-rw-r----- 1 mysql mysql 1326 Jan 12 19:42 mysql-bin.000003-rw-r----- 1 mysql mysql 57 Jan 12 19:28 mysql-bin.index[root@mysql-master mysql]# mysqlbinlog mysql-bin.000003 |grep last_committed#190112 19:37:06 server id 1 end_log_pos 259 CRC32 0x893c0ae8 GTID last_committed=0 sequence_number=1 rbr_only=yes#190112 19:37:12 server id 1 end_log_pos 542 CRC32 0xc36def61 GTID last_committed=1 sequence_number=2 rbr_only=yes#190112 19:42:42 server id 1 end_log_pos 825 CRC32 0x361701a2 GTID last_committed=2 sequence_number=3 rbr_only=yes#190112 19:42:48 server id 1 end_log_pos 1108 CRC32 0x8ba858c7 GTID last_committed=3 sequence_number=4 rbr_only=yes如上，通过最新的binlog日志看到，last_committed数值都不一样，即没有事务是在同一个组内提交的，也就是说这个日志里没有组提交信息。这是因为没有满足binlog_grup_commit那两个参数的条件，所以没有进行组提交！
五、半同步复制
半同步复制: 默认情况下MySQL的复制是异步的，master将新生成的binlog发送给各slave后，无需等待slave的ack回复(slave将接收到的binlog写进relay log后才会回复ack)，直接就认为这次DDL/DML成功了, 半同步复制(semi-synchronous replication)是指master将新生成的binlog发送给各slave时, 只需等待一个(默认)slave返回的ack信息就返回成功。
MySQL 5.7对半同步复制作了大改进，新增了一个master线程。在MySQL 5.7以前，master上的binlog dump线程负责两件事：dump日志给slave的io_thread；接收来自slave的ack消息。它们是串行方式工作的。在MySQL 5.7中，新增了一个专门负责接受ack消息的线程ack collector thread。这样master上有两个线程独立工作，可以同时发送binlog到slave和接收slave的ack。还新增了几个变量，其中最重要的是 rpl_semi_sync_master_wait_point ，它使得MySQL半同步复制有两种工作模型。
半同步复制的两种类型
从MySQL 5.7.2开始，MySQL支持两种类型的半同步复制。这两种类型由变量 rpl_semi_sync_master_wait_point (MySQL 5.7.2之前没有该变量)控制，它有两种值：AFTER_SYNC和AFTER_COMMIT。在MySQL 5.7.2之后，默认值为AFTER_SYNC，在此版本之前，等价的类型为AFTER_COMMIT。这个变量控制的是master何时提交、何时接收ack以及何时回复成功信息给客户端的时间点。AFTER_SYNC模式：master将新事务写进binlog(buffer)后发送给slave，再sync到自己的binlog file(disk), 之后才允许接收slave的ack回复，接收到ack之后才会提交事务，并返回成功信息给客户端。AFTER_COMMIT模式：master将新事务写进binlog(buffer)后发送给slave，再sync到自己的binlog file(disk)，然后直接提交事务。之后才允许接收slave的ack回复，然后再返回成功信息给客户端。如下作图，理解起来就容易了。(前提: 已经设置了sync_binlog=1，否则binlog刷盘时间由操作系统决定)
根据上面的图解，接下来分析下这两种模式的优缺点。
AFTER_SYNC
- 对于所有客户端来说，它们看到的数据是一样的，因为它们看到的数据都是在接收到slave的ack后提交后的数据。- 这种模式下，如果master突然故障，不会丢失数据，因为所有成功的事务都已经写进slave的relay log中了，slave的数据是最新的。AFTER_COMMIT
- 不同客户端看到的数据可能是不一样的。对于发起事务请求的那个客户端，它只有在master提交事务且收到slave的ack后才能看到提交的数据。但对于那些非本次事务的请求客户端，它们在master提交后就能看到提交后的数据，这时候master可能还没收到slave的ack。- 如果master收到ack回复前，slave和master都故障了，那么将丢失这个事务中的数据。在MySQL 5.7.2之前，等价的模式是 AFTER_COMMIT ，在此版本之后，默认的模式为 AFTER_SYNC ，该模式能最大程度地保证数据安全性，且性能上并不比 AFTER_COMMIT 差。
基于GTID的半同步复制环境部署记录
Mysql 半同步复制配置可以参考: https://www.cnblogs.com/kevingrace/p/10228694.html开启半同步复制的方法有两种: mysql命令行启动; my.cnf文件里添加启动配置;推荐在my.cnf文件里添加半同步启动配置方式1) 配置mysql-master主数据库上 my.cnf文件，添加启动半同步复制的配置[root@mysql-master ~]# vim /etc/my.cnf[mysqld]datadir = /var/lib/mysqlsocket = /var/lib/mysql/mysql.sock symbolic-links = 0 log-error = /var/log/mysqld.logpid-file = /var/run/mysqld/mysqld.pid #GTID:server_id = 1gtid_mode = onenforce_gtid_consistency = on #binloglog_bin = mysql-binlog-slave-updates = 1binlog_format = rowsync-master-info = 1sync_binlog = 1 #relay logskip_slave_start = 1 #不配置binlog_group_commit从库无法做到基于事物的并行复制binlog_group_commit_sync_delay = 100 binlog_group_commit_sync_no_delay_count = 10 #为了数据安全再配置sync_binlog=1innodb_flush_log_at_trx_commit =1#这个参数控制binlog写入 磁盘的方式。设置为1时，表示每次commit;都写入磁盘。这个刷新的是redo log 即ib_logfile0，而不是binlog#开启半同步复制 (超时时间为1s)plugin-load=rpl_semi_sync_master=semisync_master.sorpl_semi_sync_master_enabled = 1rpl_semi_sync_master_timeout = 1000 配置完成后，记得重启mysql服务[root@mysql-master ~]# systemctl restart mysqld2) 配置mysql-slave从数据库上 my.cnf文件，添加启动半同步复制的配置[root@mysql-slave1 ~]# vim /etc/my.cnf[mysqld]datadir = /var/lib/mysqlsocket = /var/lib/mysql/mysql.sock symbolic-links = 0 log-error = /var/log/mysqld.logpid-file = /var/run/mysqld/mysqld.pid #GTID:server_id = 2gtid_mode = onenforce_gtid_consistency = on #binloglog_bin = mysql-binlog-slave-updates = 1binlog_format = rowsync-master-info = 1sync_binlog = 1 #relay logskip_slave_start = 1read_only = on #MTSslave-parallel-type = LOGICAL_CLOCK #开启逻辑时钟的复制slave-parallel-workers = 4 #最大线程16master_info_repository = TABLErelay_log_info_repository = TABLErelay_log_recovery = on# 开启半同步复制plugin-load=rpl_semi_sync_slave=semisync_slave.sorpl_semi_sync_slave_enabled=1配置完成后，记得重启mysql服务[root@mysql-slave1 ~]# systemctl restart mysqld3）在mysql-master主数据库上加载 (前提是/usr/lib64/mysql/plugin/semisync_master.so 文件存在。 一般mysql安装后就默认产生)mysql> INSTALL PLUGIN rpl_semi_sync_master SONAME 'semisync_master.so';Query OK, 0 rows affected (0.25 sec)查看插件是否加载成功mysql> SELECT PLUGIN_NAME, PLUGIN_STATUS FROM INFORMATION_SCHEMA.PLUGINS WHERE PLUGIN_NAME LIKE '%semi%';+----------------------+---------------+| PLUGIN_NAME | PLUGIN_STATUS |+----------------------+---------------+| rpl_semi_sync_master | ACTIVE |+----------------------+---------------+1 row in set (0.00 sec)查看半同步是否在运行mysql> show status like 'Rpl_semi_sync_master_status';+-----------------------------+-------+| Variable_name | Value |+-----------------------------+-------+| Rpl_semi_sync_master_status | ON |+-----------------------------+-------+1 row in set (0.00 sec)4）在mysql-slave1从数据库上加载 (前提是/usr/lib64/mysql/plugin/semisync_slave.so 文件存在。 一般mysql安装后就默认产生)mysql> INSTALL PLUGIN rpl_semi_sync_slave SONAME 'semisync_slave.so';Query OK, 0 rows affected (0.03 sec)查看插件是否加载成功mysql> SELECT PLUGIN_NAME, PLUGIN_STATUS FROM INFORMATION_SCHEMA.PLUGINS WHERE PLUGIN_NAME LIKE '%semi%';+---------------------+---------------+| PLUGIN_NAME | PLUGIN_STATUS |+---------------------+---------------+| rpl_semi_sync_slave | ACTIVE |+---------------------+---------------+1 row in set (0.00 sec)这个时候，查看mysql-slave1的半同步是否运行mysql> show status like 'Rpl_semi_sync_slave_status';+----------------------------+-------+| Variable_name | Value |+----------------------------+-------+| Rpl_semi_sync_slave_status | OFF |+----------------------------+-------+1 row in set (0.01 sec)发现还是OFF，明明已经在my.cnf文件里开启了！这是因为此时还没有生效，必须从数据库上的IO线程才能生产！！！mysql> STOP SLAVE IO_THREAD;Query OK, 0 rows affected, 1 warning (0.00 sec)mysql> START SLAVE IO_THREAD; Query OK, 0 rows affected (0.00 sec)然后再查看mysql-slave1的半同步状态，发现就已经开启了！mysql> show status like 'Rpl_semi_sync_slave_status';+----------------------------+-------+| Variable_name | Value |+----------------------------+-------+| Rpl_semi_sync_slave_status | ON |+----------------------------+-------+1 row in set (0.00 sec)查看此时的主从同步状态mysql> show slave status \G; ...................... Slave_IO_Running: Yes Slave_SQL_Running: No发现主从同步异常，这个时候再重启下slave即可！mysql> stop slave;Query OK, 0 rows affected (0.06 sec)mysql> start slave;Query OK, 0 rows affected (0.10 sec)mysql> show slave status \G;........................ Slave_IO_Running: Yes Slave_SQL_Running: Yes5) 在mysql-master主数据库上查看mysql> show status like '%Rpl_semi%';+--------------------------------------------+-------+| Variable_name | Value |+--------------------------------------------+-------+| Rpl_semi_sync_master_clients | 1 || Rpl_semi_sync_master_net_avg_wait_time | 0 || Rpl_semi_sync_master_net_wait_time | 0 || Rpl_semi_sync_master_net_waits | 0 || Rpl_semi_sync_master_no_times | 0 || Rpl_semi_sync_master_no_tx | 0 || Rpl_semi_sync_master_status | ON || Rpl_semi_sync_master_timefunc_failures | 0 || Rpl_semi_sync_master_tx_avg_wait_time | 0 || Rpl_semi_sync_master_tx_wait_time | 0 || Rpl_semi_sync_master_tx_waits | 0 || Rpl_semi_sync_master_wait_pos_backtraverse | 0 || Rpl_semi_sync_master_wait_sessions | 0 || Rpl_semi_sync_master_yes_tx | 0 |+--------------------------------------------+-------+14 rows in set (0.00 sec)从上面信息，发现Rpl_semi_sync_master_clients的数值为1，说明此时mysql-master主数据库已经有一个半同步复制的从机，即mysql-slave1节点。Rpl_semi_sync_master_yes_tx的数值为0， 说明此时还没有半同步复制的sql语句被执行。接着在mysql-master主数据库上插入和更新数据，测试半同步复制mysql> select * from kevin.haha;+----+---------+| id | name |+----+---------+| 4 | beijing || 5 | hefei || 10 | xihu || 21 | caiyi || 22 | manman || 23 | titi |+----+---------+6 rows in set (0.01 sec)mysql> insert into kevin.haha values(30,"shikui"),(31,"shibo"),(32,"shijuan");Query OK, 3 rows affected (0.15 sec)Records: 3 Duplicates: 0 Warnings: 0mysql> update kevin.haha set id=8 where name="beijing";Query OK, 1 row affected (0.22 sec)Rows matched: 1 Changed: 1 Warnings: 0接着再去mysql-slave1从数据库上查看，发现新数据已经同步过来了mysql> select * from kevin.haha;+----+---------+| id | name |+----+---------+| 5 | hefei || 8 | beijing || 10 | xihu || 21 | caiyi || 22 | manman || 23 | titi || 30 | shikui || 31 | shibo || 32 | shijuan |+----+---------+9 rows in set (0.00 sec)接着再去mysql-master主数据库上查看，如下：mysql> show status like '%Rpl_semi%';+--------------------------------------------+--------+| Variable_name | Value |+--------------------------------------------+--------+| Rpl_semi_sync_master_clients | 1 || Rpl_semi_sync_master_net_avg_wait_time | 0 || Rpl_semi_sync_master_net_wait_time | 0 || Rpl_semi_sync_master_net_waits | 2 || Rpl_semi_sync_master_no_times | 0 || Rpl_semi_sync_master_no_tx | 0 || Rpl_semi_sync_master_status | ON || Rpl_semi_sync_master_timefunc_failures | 0 || Rpl_semi_sync_master_tx_avg_wait_time | 133362 || Rpl_semi_sync_master_tx_wait_time | 266725 || Rpl_semi_sync_master_tx_waits | 2 || Rpl_semi_sync_master_wait_pos_backtraverse | 0 || Rpl_semi_sync_master_wait_sessions | 0 || Rpl_semi_sync_master_yes_tx | 2 |+--------------------------------------------+--------+14 rows in set (0.01 sec)从上面信息可发现Rpl_semi_sync_master_yes_tx的数值为2，即发生了两条半同步复制的sql语句，就是上面执行的那两条(insert 和update) sql语句。
以上都是在mysql-master主数据库和mysql-slave1从数据库之间实现的基于GTID的主从复制、并行复制、半同步复制，即"一主一从"架构。 现在再把mysql-slave2的从节点添加进去，调整为"一主两从"的同步架构。添加操作记录如下:
1）在mysql-master主数据库上操作先添加同步权限mysql> grant replication slave,replication client on *.* to slave@'172.16.60.213' identified by "slave@123";Query OK, 0 rows affected, 1 warning (0.14 sec)mysql> flush privileges;Query OK, 0 rows affected (0.14 sec)备份主库[root@mysql-master ~]# mysqldump -u root -p'123456' --lock-all-tables --master-data=2 --flush-logs --all-databases --triggers --routines --events > 213_slave.sql将备份文件拷贝到mysql-slave2从数据库上[root@mysql-master ~]# rsync -e "ssh -p22" -avpgolr /root/213_slave.sql root@172.16.60.213:/root/记录当前的gtidmysql> show global variables like 'gtid_%';+----------------------------------+-------------------------------------------+| Variable_name | Value |+----------------------------------+-------------------------------------------+| gtid_executed | 317e2aad-1565-11e9-9c2e-005056ac6820:1-34 || gtid_executed_compression_period | 1000 || gtid_mode | ON || gtid_owned | || gtid_purged | 317e2aad-1565-11e9-9c2e-005056ac6820:1-16 |+----------------------------------+-------------------------------------------+5 rows in set (0.00 sec)2）在mysql-slave2从数据库上操作首先在my.cnf文件添加相关同步配置 （跟mysql-slave1从节点的my.cnf配置相比，除了将server_id修改为3之外，其他配置内容都一样！）[root@mysql-slave2 ~]# cp /etc/my.cnf /etc/my.cnf.bak[root@mysql-slave2 ~]# >/etc/my.cnf[root@mysql-slave2 ~]# vim /etc/my.cnf[mysqld]datadir = /var/lib/mysqlsocket = /var/lib/mysql/mysql.sock symbolic-links = 0 log-error = /var/log/mysqld.logpid-file = /var/run/mysqld/mysqld.pid #GTID:server_id = 3gtid_mode = onenforce_gtid_consistency = on #binloglog_bin = mysql-binlog-slave-updates = 1binlog_format = rowsync-master-info = 1sync_binlog = 1 #relay logskip_slave_start = 1read_only = on #MTSslave-parallel-type = LOGICAL_CLOCK #开启逻辑时钟的复制slave-parallel-workers = 4 #最大线程16master_info_repository = TABLErelay_log_info_repository = TABLErelay_log_recovery = on # 开启半同步复制plugin-load=rpl_semi_sync_slave=semisync_slave.sorpl_semi_sync_slave_enabled=1配置完成后，不要忘记重启mysql服务[root@mysql-slave2 ~]# systemctl restart mysqld恢复备份库[root@mysql-slave2 ~]# ll /root/213_slave.sql -rw-r--r-- 1 root root 805694 Jan 13 00:26 /root/213_slave.sqlmysql> show databases;+--------------------+| Database |+--------------------+| information_schema || mysql || performance_schema || sys |+--------------------+4 rows in set (0.00 sec)mysql> source /root/213_slave.sql;mysql> select * from kevin.haha;+----+---------+| id | name |+----+---------+| 5 | hefei || 8 | beijing || 10 | xihu || 21 | caiyi || 22 | manman || 23 | titi || 30 | shikui || 31 | shibo || 32 | shijuan |+----+---------+9 rows in set (0.00 sec)检查一下mysql-slave2从数据库上当前的gtid mysql> show global variables like 'gtid_%';+----------------------------------+-------------------------------------------+| Variable_name | Value |+----------------------------------+-------------------------------------------+| gtid_executed | 317e2aad-1565-11e9-9c2e-005056ac6820:1-34 || gtid_executed_compression_period | 1000 || gtid_mode | ON || gtid_owned | || gtid_purged | 317e2aad-1565-11e9-9c2e-005056ac6820:1-34 |+----------------------------------+-------------------------------------------+5 rows in set (0.01 sec)由于是从master-master主库备份恢复过来的，所以此时mysql-slave2从数据库的gtid和主库的gtid是一样的！接着进行主从同步mysql> stop slave;Query OK, 0 rows affected, 1 warning (0.00 sec)mysql> change master to master_host='172.16.60.211',master_user='slave',master_password='slave@123',master_auto_position=1; Query OK, 0 rows affected, 2 warnings (0.18 sec)mysql> start slave;Query OK, 0 rows affected (0.17 sec)=================================================================如果遇到报错：ERROR 1872 (HY000): Slave failed to initialize relay log info structure from the repository解决办法：stop slave;reset slave;start slave;=================================================================查看，发现主从同步正常mysql> show slave status \G;......... Slave_IO_Running: Yes Slave_SQL_Running: Yes并行复制查看mysql> show variables like '%slave_para%';+------------------------+---------------+| Variable_name | Value |+------------------------+---------------+| slave_parallel_type | LOGICAL_CLOCK || slave_parallel_workers | 4 |+------------------------+---------------+2 rows in set (0.01 sec)mysql> select * from mysql.slave_worker_info; +----+----------------+---------------+-----------------+----------------+---------------------------+--------------------------+----------------------------+---------------------------+------------------+-----------------------+------------------------------------------------------------------+--------------+| Id | Relay_log_name | Relay_log_pos | Master_log_name | Master_log_pos | Checkpoint_relay_log_name | Checkpoint_relay_log_pos | Checkpoint_master_log_name | Checkpoint_master_log_pos | Checkpoint_seqno | Checkpoint_group_size | Checkpoint_group_bitmap | Channel_name |+----+----------------+---------------+-----------------+----------------+---------------------------+--------------------------+----------------------------+---------------------------+------------------+-----------------------+------------------------------------------------------------------+--------------+| 1 | | 0 | | 0 | | 0 | | 0 | 0 | 64 | | || 2 | | 0 | | 0 | | 0 | | 0 | 0 | 64 | | || 3 | | 0 | | 0 | | 0 | | 0 | 0 | 64 | | || 4 | | 0 | | 0 | | 0 | | 0 | 0 | 64 | | |+----+----------------+---------------+-----------------+----------------+---------------------------+--------------------------+----------------------------+---------------------------+------------------+-----------------------+------------------------------------------------------------------+--------------+4 rows in set (0.00 sec)接着在mysql-slave2从数据库上开启半同步mysql> INSTALL PLUGIN rpl_semi_sync_slave SONAME 'semisync_slave.so';Query OK, 0 rows affected (0.03 sec)mysql> SELECT PLUGIN_NAME, PLUGIN_STATUS FROM INFORMATION_SCHEMA.PLUGINS WHERE PLUGIN_NAME LIKE '%semi%';+---------------------+---------------+| PLUGIN_NAME | PLUGIN_STATUS |+---------------------+---------------+| rpl_semi_sync_slave | ACTIVE |+---------------------+---------------+1 row in set (0.00 sec)mysql> STOP SLAVE IO_THREAD;Query OK, 0 rows affected (0.02 sec)mysql> START SLAVE IO_THREAD; Query OK, 0 rows affected (0.00 sec)mysql> show status like 'Rpl_semi_sync_slave_status'; +----------------------------+-------+| Variable_name | Value |+----------------------------+-------+| Rpl_semi_sync_slave_status | ON |+----------------------------+-------+1 row in set (0.00 sec)3） 现在回到mysql-master主数据库查看mysql> show slave hosts;+-----------+------+------+-----------+--------------------------------------+| Server_id | Host | Port | Master_id | Slave_UUID |+-----------+------+------+-----------+--------------------------------------+| 3 | | 3306 | 1 | 2658b203-1565-11e9-9f8b-005056880888 || 2 | | 3306 | 1 | 2c1efc46-1565-11e9-ab8e-00505688047c |+-----------+------+------+-----------+--------------------------------------+2 rows in set (0.00 sec)如上信息可知mysql-master主数据库现在有两个从数据库，分别为mysql-slave1 和 mysql-slave2mysql> show status like '%Rpl_semi%'; +--------------------------------------------+--------+| Variable_name | Value |+--------------------------------------------+--------+| Rpl_semi_sync_master_clients | 2 || Rpl_semi_sync_master_net_avg_wait_time | 0 || Rpl_semi_sync_master_net_wait_time | 0 || Rpl_semi_sync_master_net_waits | 4 || Rpl_semi_sync_master_no_times | 0 || Rpl_semi_sync_master_no_tx | 0 || Rpl_semi_sync_master_status | ON || Rpl_semi_sync_master_timefunc_failures | 0 || Rpl_semi_sync_master_tx_avg_wait_time | 98077 || Rpl_semi_sync_master_tx_wait_time | 392310 || Rpl_semi_sync_master_tx_waits | 4 || Rpl_semi_sync_master_wait_pos_backtraverse | 0 || Rpl_semi_sync_master_wait_sessions | 0 || Rpl_semi_sync_master_yes_tx | 4 |+--------------------------------------------+--------+14 rows in set (0.00 sec)如上信息可知，mysql-master主数据库现在有两个半同步复制的从库，即mysql-slave1 和mysql-slave24）现在测试下同步效果在mysql-master主数据库上更新数据mysql> delete from kevin.haha where id>10;Query OK, 6 rows affected (0.20 sec)mysql> insert into kevin.haha values(11,"changping"),(12,"wangjing");Query OK, 2 rows affected (0.38 sec)Records: 2 Duplicates: 0 Warnings: 0在mysql-slave1从数据库查看，发现已经同步过来了mysql> select * from kevin.haha;+----+-----------+| id | name |+----+-----------+| 5 | hefei || 8 | beijing || 10 | xihu || 11 | changping || 12 | wangjing |+----+-----------+5 rows in set (0.00 sec)在mysql-slave2从数据库查看，发现已经同步过来了mysql> select * from kevin.haha;+----+-----------+| id | name |+----+-----------+| 5 | hefei || 8 | beijing || 10 | xihu || 11 | changping || 12 | wangjing |+----+-----------+5 rows in set (0.00 sec)
到此，已经完成了Mysql5.7"一主两从"架构上基于GTID主从复制+并行复制+增强半同步复制环境，下面再说下读写分离环境：
六、读写分离 （MySQL ProxySQL）
ProxySQL是用C++语言开发的，也是percona推的一款中间件，虽然也是一个轻量级产品，但性能很好(据测试，能处理千亿级的数据)，功能也足够，能满足中间件所需的绝大多数功能，包括：
- 最基本的读写分离，且起来也十分方便。- 可定制基于用户、基于schema、基于语句的规则对SQL语句进行路由。换句话说，规则很灵活。基于schema和与语句级的规则，可以实现简单的sharding。- 可缓存查询结果。虽然ProxySQL的缓存策略比较简陋，但实现了基本的缓存功能，绝大多数时候也够用了。此外，作者已经打算实现更丰富的缓存策略。- 监控后端节点。ProxySQL可以监控后端节点的多个指标，包括：ProxySQL和后端的心跳信息，后端节点的read-only/read-write，slave和master的数据同步延迟性(replication lag)。ProxySQL是一个能实实在在用在生产环境的MySQL中间件，可以实现读写分离，支持 Query 路由功能，支持动态指定某个 SQL 进行 cache，支持动态加载配置、故障切换和一些 SQL的过滤功能。还有一些同类产品比如 DBproxy、MyCAT、OneProxy 等。但经过反复对比和测试之后，还是觉得ProxySQL是一款性能不谙，靠谱稳定的MySQL 中间件产品 ！
ProxySQL读写分离环境部署记录
下面操作是在mysql-proxy代理层节点上完成的（需要注意： 两个从库都要开启 read_only=on）：
1）先在本机安装mysql客户端，用于在本机连接到ProxySQL的管理接口[root@mysql-proxy ~]# vim /etc/yum.repos.d/mariadb.repo[mariadb] name = MariaDBbaseurl = http://yum.mariadb.org/10.3.5/centos6-amd64gpgkey=https://yum.mariadb.org/RPM-GPG-KEY-MariaDB gpgcheck=1安装mysql-clinet客户端[root@mysql-proxy ~]# yum install -y MariaDB-client =================================================如果遇到报错：Error: MariaDB-compat conflicts with 1:mariadb-libs-5.5.60-1.el7_5.x86_64 You could try using --skip-broken to work around the problem You could try running: rpm -Va --nofiles --nodigest解决办法：[root@mysql-proxy ~]# rpm -qa|grep mariadb*mariadb-libs-5.5.56-2.el7.x86_64[root@mysql-proxy ~]# rpm -e mariadb-libs-5.5.56-2.el7.x86_64 --nodeps[root@mysql-proxy ~]# yum install -y MariaDB-client
2）安装ProxySQL （中文安装手册: ）
[root@mysql-proxy ~]# vim /etc/yum.repos.d/proxysql.repo [proxysql_repo]name= ProxySQL YUM repositorybaseurl=http://repo.proxysql.com/ProxySQL/proxysql-1.4.x/centos/\$releasevergpgcheck=1gpgkey=http://repo.proxysql.com/ProxySQL/repo_pub_key执行安装[root@mysql-proxy ~]# yum -y install proxysql[root@mysql-proxy ~]# proxysql --versionProxySQL version 1.4.13-15-g69d4207, codename Truls启动ProxySQL[root@mysql-proxy ~]# chkconfig proxysql on[root@mysql-proxy ~]# systemctl start proxysql [root@mysql-proxy ~]# systemctl status proxysql 启动后会监听两个端口，默认为6032和6033。6032端口是ProxySQL的管理端口，6033是ProxySQL对外提供服务的端口 (即连接到转发后端的真正数据库的转发端口)。[root@mysql-proxy ~]# netstat -tunlpActive Internet connections (only servers)Proto Recv-Q Send-Q Local Address Foreign Address State PID/Program name tcp 0 0 0.0.0.0:6032 0.0.0.0:* LISTEN 23940/proxysql tcp 0 0 0.0.0.0:6033 0.0.0.0:* LISTEN 23940/proxysql
3）向ProxySQL中添加MySQL节点
ProxySQL也是有管理接口和客户端接口，通过配置文件/etc/proxysql.cnf可以看到管理和客户端接口的信息.[root@mysql-proxy ~]# cat /etc/proxysql.cnf .........admin_variables={ admin_credentials="admin:admin"# mysql_ifaces="127.0.0.1:6032;/tmp/proxysql_admin.sock" mysql_ifaces="0.0.0.0:6032"# refresh_interval=2000# debug=true}mysql_variables={ threads=4 max_connections=2048 default_query_delay=0 default_query_timeout=36000000 have_compress=true poll_timeout=2000# interfaces="0.0.0.0:6033;/tmp/proxysql.sock" interfaces="0.0.0.0:6033" default_schema="information_schema" stacksize=1048576 server_version="5.5.30" connect_timeout_server=3000..........通过上面，可以看到管理接口的端口是6032，账号密码是admin(可以动态修改),允许客户端连接， 客户端接口的端口是6033，账号密码通过管理接口去设置。在mysql-proxy本机使用mysql客户端连接到ProxySQL的管理接口(admin interface), 该接口的默认管理员用户和密码都是admin.下面是通过管理端口6032去连接的(下面连接命令中后面的--prompt 'admin'字段可以不加，也是可以登录进去的)[root@mysql-proxy ~]# mysql -uadmin -padmin -P6032 -h127.0.0.1 --prompt 'admin> 'Welcome to the MariaDB monitor. Commands end with ; or \g.Your MySQL connection id is 8Server version: 5.5.30 (ProxySQL Admin Module)Copyright (c) 2000, 2018, Oracle, MariaDB Corporation Ab and others.Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.admin> show databases;+-----+---------------+-------------------------------------+| seq | name | file |+-----+---------------+-------------------------------------+| 0 | main | || 2 | disk | /var/lib/proxysql/proxysql.db || 3 | stats | || 4 | monitor | || 5 | stats_history | /var/lib/proxysql/proxysql_stats.db |+-----+---------------+-------------------------------------+5 rows in set (0.001 sec)ProxySQL提供了几个库，每个库都有各自的意义；- main 内存配置数据库，表里存放后端db实例、用户验证、路由规则等信息。表名以 runtime_开头的表示proxysql当前运行的配置内容， 不能通过dml语句修改，只能修改对应的不以 runtime_ 开头的（在内存）里的表，然后 LOAD 使其生效， SAVE 使其存到硬盘以供下次重启加载。- disk 是持久化到硬盘的配置，sqlite数据文件。- stats 是proxysql运行抓取的统计信息，包括到后端各命令的执行次数、流量、processlist、查询种类汇总/执行时间等等。- monitor 库存储 monitor 模块收集的信息，主要是对后端db的健康/延迟检查。主要注意main和monitor数据库中的表admin> show tables from main;+--------------------------------------------+| tables |+--------------------------------------------+| global_variables || mysql_collations || mysql_group_replication_hostgroups || mysql_query_rules || mysql_query_rules_fast_routing || mysql_replication_hostgroups || mysql_servers || mysql_users || proxysql_servers || runtime_checksums_values || runtime_global_variables || runtime_mysql_group_replication_hostgroups || runtime_mysql_query_rules || runtime_mysql_query_rules_fast_routing || runtime_mysql_replication_hostgroups || runtime_mysql_servers || runtime_mysql_users || runtime_proxysql_servers || runtime_scheduler || scheduler |+--------------------------------------------+20 rows in set (0.001 sec)几个重要字段含义：global_variables 设置变量，包括监听的端口、管理账号等。mysql_collations 相关字符集和校验规则。mysql_query_rules 定义查询路由规则。admin> show tables from monitor;+------------------------------------+| tables |+------------------------------------+| mysql_server_connect_log || mysql_server_group_replication_log || mysql_server_ping_log || mysql_server_read_only_log || mysql_server_replication_lag_log |+------------------------------------+5 rows in set (0.000 sec)runtime_开头的是运行时的配置，这些是不能修改的。要修改ProxySQL的配置，需要修改了非runtime_表，修改后必须执行LOAD ... TO RUNTIME才能加载到RUNTIME生效，执行save ... to disk才能将配置持久化保存到磁盘。使用insert语句添加主机到mysql_servers表中，其中：hostgroup_id 10 表示写组，20表示读组。admin> insert into mysql_servers(hostgroup_id,hostname,port) values(10,'172.16.60.211',3306); Query OK, 1 row affected (0.001 sec)admin> insert into mysql_servers(hostgroup_id,hostname,port) values(10,'172.16.60.212',3306);Query OK, 1 row affected (0.000 sec)admin> insert into mysql_servers(hostgroup_id,hostname,port) values(10,'172.16.60.213',3306);Query OK, 1 row affected (0.000 sec)查看这3个节点是否插入成功，以及它们的状态。admin> select * from mysql_servers\G;*************************** 1. row *************************** hostgroup_id: 10 hostname: 172.16.60.211 port: 3306 status: ONLINE weight: 1 compression: 0 max_connections: 1000max_replication_lag: 0 use_ssl: 0 max_latency_ms: 0 comment: *************************** 2. row *************************** hostgroup_id: 10 hostname: 172.16.60.212 port: 3306 status: ONLINE weight: 1 compression: 0 max_connections: 1000max_replication_lag: 0 use_ssl: 0 max_latency_ms: 0 comment: *************************** 3. row *************************** hostgroup_id: 10 hostname: 172.16.60.213 port: 3306 status: ONLINE weight: 1 compression: 0 max_connections: 1000max_replication_lag: 0 use_ssl: 0 max_latency_ms: 0 comment: 3 rows in set (0.000 sec)ERROR: No query specifiedadmin> 如上修改后，加载到RUNTIME，并保存到diskadmin> load mysql servers to runtime;Query OK, 0 rows affected (0.005 sec)admin> save mysql servers to disk; Query OK, 0 rows affected (0.468 sec) 4）监控后端MySQL节点
如上添加Mysql节点之后，还需要监控这些后端节点。对于后端是主从复制的环境来说，这是必须的，因为ProxySQL需要通过每个节点的read_only值来自动调整它们是属于读组还是写组。首先在后端master主数据节点上创建一个用于监控的用户名(只需在master上创建即可，因为会复制到slave上)，这个用户名只需具有USAGE权限即可。如果还需要监控复制结构中slave是否严重延迟于master(这个俗语叫做"拖后腿"，术语叫做"replication lag")，则还需具备replication client权限。即在mysql-master主数据库节点行执行：mysql> create user monitor@'172.16.60.%' identified by 'P@ssword1!'; Query OK, 0 rows affected (0.08 sec)mysql> grant replication client on *.* to monitor@'172.16.60.%';Query OK, 0 rows affected (0.07 sec)mysql> flush privileges;Query OK, 0 rows affected (0.07 sec)然后回到mysql-proxy代理层节点上配置监控admin> set mysql-monitor_username='monitor';Query OK, 1 row affected (0.001 sec)admin> set mysql-monitor_password='P@ssword1!';Query OK, 1 row affected (0.000 sec)修改后，加载到RUNTIME，并保存到diskadmin> load mysql variables to runtime;Query OK, 0 rows affected (0.001 sec)admin> save mysql variables to disk;Query OK, 96 rows affected (0.180 sec)验证监控结果：ProxySQL监控模块的指标都保存在monitor库的log表中。以下是连接是否正常的监控(对connect指标的监控)：(在前面可能会有很多connect_error，这是因为没有配置监控信息时的错误，配置后如果connect_error的结果为NULL则表示正常)admin> select * from mysql_server_connect_log;+---------------+------+------------------+-------------------------+------------------------------------------------------------------------+| hostname | port | time_start_us | connect_success_time_us | connect_error |+---------------+------+------------------+-------------------------+------------------------------------------------------------------------+| 172.16.60.213 | 3306 | 1547436747566683 | 0 | Access denied for user 'monitor'@'docker-node2' (using password: YES) || 172.16.60.211 | 3306 | 1547436748167350 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.212 | 3306 | 1547436748767837 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.212 | 3306 | 1547436807566915 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.211 | 3306 | 1547436808112262 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.213 | 3306 | 1547436808657603 | 0 | Access denied for user 'monitor'@'docker-node2' (using password: YES) || 172.16.60.213 | 3306 | 1547436867566829 | 0 | Access denied for user 'monitor'@'docker-node2' (using password: YES) || 172.16.60.211 | 3306 | 1547436868151063 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.212 | 3306 | 1547436868735236 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.212 | 3306 | 1547436927567060 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.211 | 3306 | 1547436928247532 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.213 | 3306 | 1547436928927772 | 0 | Access denied for user 'monitor'@'docker-node2' (using password: YES) || 172.16.60.211 | 3306 | 1547436987567058 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.212 | 3306 | 1547436988049160 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.213 | 3306 | 1547436988531003 | 0 | Access denied for user 'monitor'@'docker-node2' (using password: YES) || 172.16.60.212 | 3306 | 1547437047567194 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.213 | 3306 | 1547437048132480 | 0 | Access denied for user 'monitor'@'docker-node2' (using password: YES) || 172.16.60.211 | 3306 | 1547437048697765 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.211 | 3306 | 1547437107567258 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.212 | 3306 | 1547437108310411 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.213 | 3306 | 1547437109053480 | 0 | Access denied for user 'monitor'@'docker-node2' (using password: YES) || 172.16.60.213 | 3306 | 1547437167567303 | 0 | Access denied for user 'monitor'@'docker-node2' (using password: YES) || 172.16.60.211 | 3306 | 1547437168158076 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.212 | 3306 | 1547437168748763 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.213 | 3306 | 1547437227567525 | 0 | Access denied for user 'monitor'@'docker-node2' (using password: YES) || 172.16.60.211 | 3306 | 1547437228288921 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.212 | 3306 | 1547437229010153 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.211 | 3306 | 1547437287567412 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.212 | 3306 | 1547437288343609 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.213 | 3306 | 1547437289119851 | 0 | Access denied for user 'monitor'@'docker-node2' (using password: YES) || 172.16.60.213 | 3306 | 1547437295397678 | 1388 | NULL || 172.16.60.211 | 3306 | 1547437295826682 | 1377 | NULL || 172.16.60.212 | 3306 | 1547437296255659 | 8072 | NULL |+---------------+------+------------------+-------------------------+------------------------------------------------------------------------+33 rows in set (0.001 sec)以下是对心跳信息的监控(对ping指标的监控)admin> select * from mysql_server_ping_log;+---------------+------+------------------+----------------------+------------------------------------------------------------------------+| hostname | port | time_start_us | ping_success_time_us | ping_error |+---------------+------+------------------+----------------------+------------------------------------------------------------------------+| 172.16.60.213 | 3306 | 1547436798804839 | 0 | Access denied for user 'monitor'@'172.16.60.213' (using password: YES) || 172.16.60.211 | 3306 | 1547436798875555 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.212 | 3306 | 1547436798946212 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.213 | 3306 | 1547436808804803 | 0 | Access denied for user 'monitor'@'172.16.60.213' (using password: YES) || 172.16.60.212 | 3306 | 1547436808931415 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) || 172.16.60.211 | 3306 | 1547436809058082 | 0 | Access denied for user 'monitor'@'172.16.60.214' (using password: YES) |........................................| 172.16.60.213 | 3306 | 1547437295030407 | 335 | NULL || 172.16.60.211 | 3306 | 1547437295159540 | 218 | NULL || 172.16.60.212 | 3306 | 1547437295288638 | 172 | NULL || 172.16.60.211 | 3306 | 1547437365241682 | 337 | NULL || 172.16.60.211 | 3306 | 1547437375031403 | 435 | NULL |....................................| 172.16.60.213 | 3306 | 1547437395160777 | 469 | NULL || 172.16.60.211 | 3306 | 1547437395289919 | 377 | NULL |+---------------+------+------------------+----------------------+------------------------------------------------------------------------+183 rows in set (0.001 sec)但是，read_only和replication_lag的监控日志都为空。admin> select * from mysql_server_read_only_log;Empty set (0.000 sec)admin> select * from mysql_server_replication_lag_log;Empty set (0.000 sec)例如，指定写组的id为10，读组的id为20。admin> insert into mysql_replication_hostgroups values(10,20,1);Query OK, 1 row affected (0.000 sec)在该配置加载到RUNTIME生效之前，先查看下各mysql server所在的组。admin> select hostgroup_id,hostname,port,status,weight from mysql_servers;+--------------+---------------+------+--------+--------+| hostgroup_id | hostname | port | status | weight |+--------------+---------------+------+--------+--------+| 10 | 172.16.60.211 | 3306 | ONLINE | 1 || 10 | 172.16.60.212 | 3306 | ONLINE | 1 || 10 | 172.16.60.213 | 3306 | ONLINE | 1 |+--------------+---------------+------+--------+--------+3 rows in set (0.000 sec)3个节点都在hostgroup_id=10的组中。现在，将刚才mysql_replication_hostgroups表的修改加载到RUNTIME生效。admin> load mysql servers to runtime;Query OK, 0 rows affected (0.005 sec)admin> save mysql servers to disk;Query OK, 0 rows affected (0.406 sec)一加载，Monitor模块就会开始监控后端的read_only值，当监控到read_only值后，就会按照read_only的值将某些节点自动移动到读/写组。例如，此处所有节点都在id=10的写组，slave1和slave2都是slave，它们的read_only=1，这两个节点将会移动到id=20的组。如果一开始这3节点都在id=20的读组，那么移动的将是Master节点，会移动到id=10的写组。现在看结果admin> select hostgroup_id,hostname,port,status,weight from mysql_servers;+--------------+---------------+------+--------+--------+| hostgroup_id | hostname | port | status | weight |+--------------+---------------+------+--------+--------+| 10 | 172.16.60.211 | 3306 | ONLINE | 1 || 20 | 172.16.60.212 | 3306 | ONLINE | 1 || 20 | 172.16.60.213 | 3306 | ONLINE | 1 |+--------------+---------------+------+--------+--------+3 rows in set (0.000 sec)admin> select * from mysql_server_read_only_log;+---------------+------+------------------+-----------------+-----------+-------+| hostname | port | time_start_us | success_time_us | read_only | error |+---------------+------+------------------+-----------------+-----------+-------+| 172.16.60.212 | 3306 | 1547438982938183 | 694 | 1 | NULL || 172.16.60.211 | 3306 | 1547438982950903 | 741 | 0 | NULL |............................| 172.16.60.211 | 3306 | 1547439054961006 | 517 | 0 | NULL || 172.16.60.213 | 3306 | 1547439054977658 | 479 | 1 | NULL |+---------------+------+------------------+-----------------+-----------+-------+147 rows in set (0.001 sec)
5）配置mysql_users
上面的所有配置都是关于后端MySQL节点的，现在可以配置关于SQL语句的，包括：发送SQL语句的用户、SQL语句的路由规则、SQL查询的缓存、SQL语句的重写等等。本小节是SQL请求所使用的用户配置，例如root用户。这要求我们需要先在后端MySQL节点添加好相关用户。这里以root和sqlsender两个用户名为例。首先，在mysql-master主数据库节点上执行：(只需master执行即可，会复制给两个slave)mysql> grant all on *.* to root@'172.16.60.%' identified by 'passwd';Query OK, 0 rows affected, 1 warning (0.06 sec)mysql> grant all on *.* to sqlsender@'172.16.60.%' identified by 'P@ssword1!';Query OK, 0 rows affected, 1 warning (0.15 sec)mysql> flush privileges;Query OK, 0 rows affected (0.06 sec)然后回到mysql-proxy代理层节点，配置mysql_users表，将刚才的两个用户添加到该表中。admin> insert into mysql_users(username,password,default_hostgroup) values('root','passwd',10);Query OK, 1 row affected (0.001 sec)admin> insert into mysql_users(username,password,default_hostgroup) values('sqlsender','P@ssword1!',10);Query OK, 1 row affected (0.000 sec)admin> load mysql users to runtime;Query OK, 0 rows affected (0.001 sec)admin> save mysql users to disk;Query OK, 0 rows affected (0.108 sec)mysql_users表有不少字段，最主要的三个字段为username、password和default_hostgroup：- username：前端连接ProxySQL，以及ProxySQL将SQL语句路由给MySQL所使用的用户名。- password：用户名对应的密码。可以是明文密码，也可以是hash密码。如果想使用hash密码，可以先在某个MySQL节点上执行 select password(PASSWORD)，然后将加密结果复制到该字段。- default_hostgroup：该用户名默认的路由目标。例如，指定root用户的该字段值为10时，则使用root用户发送的SQL语句默认 情况下将路由到hostgroup_id=10组中的某个节点。admin> select * from mysql_users\G*************************** 1. row *************************** username: root password: passwd active: 1 use_ssl: 0 default_hostgroup: 10 default_schema: NULL schema_locked: 0transaction_persistent: 1 fast_forward: 0 backend: 1 frontend: 1 max_connections: 10000*************************** 2. row *************************** username: sqlsender password: P@ssword1! active: 1 use_ssl: 0 default_hostgroup: 10 default_schema: NULL schema_locked: 0transaction_persistent: 1 fast_forward: 0 backend: 1 frontend: 1 max_connections: 100002 rows in set (0.000 sec)虽然这里没有详细介绍mysql_users表，但上面标注了"注意本行"的两个字段必须要引起注意。只有active=1的用户才是有效的用户。至于transaction_persistent字段，当它的值为1时，表示事务持久化：当某连接使用该用户开启了一个事务后，那么在事务提交/回滚之前，所有的语句都路由到同一个组中，避免语句分散到不同组。在以前的版本中，默认值为0，不知道从哪个版本开始，它的默认值为1。我们期望的值为1，所以在继续下面的步骤之前，先查看下这个值，如果为0，则执行下面的语句修改为1。admin> update mysql_users set transaction_persistent=1 where username='root';Query OK, 1 row affected (0.000 sec)admin> update mysql_users set transaction_persistent=1 where username='sqlsender';Query OK, 1 row affected (0.000 sec)admin> load mysql users to runtime;Query OK, 0 rows affected (0.001 sec)admin> save mysql users to disk;Query OK, 0 rows affected (0.098 sec)然后，另开一个终端，分别使用root用户和sqlsender用户测试下它们是否能路由到默认的hostgroup_id=10(它是一个写组)读、写数据。[root@mysql-proxy ~]# mysql -uroot -ppasswd -P6033 -h127.0.0.1 -e "select @@server_id" #这是通过转发端口6033连接的，连接的是转发到后端真正的数据库!+-------------+| @@server_id |+-------------+| 1 |+-------------+[root@mysql-proxy ~]# mysql -uroot -ppasswd -P6033 -h127.0.0.1 -e "create database proxy_test"[root@mysql-proxy ~]# mysql -uroot -ppasswd -P6033 -h127.0.0.1 -e "show databases;"+--------------------+| Database |+--------------------+| information_schema || kevin || mysql || performance_schema || proxy_test || sys |+--------------------+[root@mysql-proxy ~]# mysql -usqlsender -pP@ssword1! -P6033 -h127.0.0.1 -e 'use proxy_test;create table t(id int);' [root@mysql-proxy ~]# mysql -usqlsender -pP@ssword1! -P6033 -h127.0.0.1 -e 'show tables from proxy_test;'+----------------------+| Tables_in_proxy_test |+----------------------+| t |+----------------------+ 6）读写分离：配置路由规则
ProxySQL的路由规则非常灵活，可以基于用户、基于schema以及基于每个语句实现路由规则的定制。本案例作为一个入门配置，实现一个最简单的语句级路由规则，从而实现读写分离。必须注意，这只是实验，实际的路由规则绝不应该仅根据所谓的读、写操作进行分离，而是从各项指标中找出压力大、执行频繁的语句单独写规则、做缓存等等。和查询规则有关的表有两个:mysql_query_rules和mysql_query_rules_fast_routing，后者是前者的扩展表，1.4.7之后才支持该快速路由表。本案例只介绍第一个表。插入两个规则，目的是将select语句分离到hostgroup_id=20的读组，但由于select语句中有一个特殊语句SELECT...FOR UPDATE它会申请写锁，所以应该路由到hostgroup_id=10的写组。admin> insert into mysql_query_rules(rule_id,active,match_digest,destination_hostgroup,apply) VALUES (1,1,'^SELECT.*FOR UPDATE$',10,1), (2,1,'^SELECT',20,1);Query OK, 2 rows affected (0.001 sec) admin> load mysql query rules to runtime;Query OK, 0 rows affected (0.000 sec) admin> save mysql query rules to disk;Query OK, 0 rows affected (0.179 sec) 需要注意： select ... for update规则的rule_id必须要小于普通的select规则的rule_id，因为ProxySQL是根据rule_id的顺序进行规则匹配的。 再来测试下，读操作是否路由给了hostgroup_id=20的读组。[root@mysql-proxy ~]# mysql -uroot -ppasswd -P6033 -h127.0.0.1 -e 'select @@server_id'+-------------+| @@server_id |+-------------+| 3 |+-------------+ [root@mysql-proxy ~]# mysql -uroot -ppasswd -P6033 -h127.0.0.1 -e 'select @@server_id'+-------------+| @@server_id |+-------------+| 2 |+-------------+ 读操作已经路由给读组，再看看写操作。这里以事务持久化进行测试。[root@mysql-proxy ~]# mysql -uroot -ppasswd -P6033 -h127.0.0.1 -e 'start transaction;select @@server_id;commit;select @@server_id;'+-------------+| @@server_id |+-------------+| 1 |+-------------++-------------+| @@server_id |+-------------+| 3 |+-------------+ 显然，一切都按照预期进行。最后，如果想查看路由的信息，可查询stats库中的stats_mysql_query_digest表。以下是该表的一个输出格式示例(和本案例无关)。admin> SELECT hostgroup hg, sum_time, count_star, digest_text FROM stats_mysql_query_digest ORDER BY sum_time DESC;+----+----------+------------+----------------------------------+| hg | sum_time | count_star | digest_text |+----+----------+------------+----------------------------------+| 10 | 236771 | 1 | create table t(id int) || 10 | 53767 | 1 | create database proxy_test || 20 | 10344 | 1 | select @@server_id || 20 | 3350 | 3 | select @@server_id || 10 | 2469 | 2 | SELECT DATABASE() || 10 | 1678 | 2 | select @@server_id || 10 | 1193 | 1 | show databases || 10 | 1191 | 1 | start transaction || 10 | 815 | 1 | show tables from proxy_test || 10 | 289 | 1 | reate table t(id int) || 10 | 162 | 1 | commit || 10 | 0 | 6 | select @@version_comment limit ? || 10 | 0 | 4 | select @@version_comment limit ? || 10 | 0 | 1 | select @@version_comment limit ? |+----+----------+------------+----------------------------------+14 rows in set (0.002 sec)
7）最后测试通过proxysql插件进行读写分离
mysql-master主数据库查看mysql> select * from kevin.haha;+----+-----------+| id | name |+----+-----------+| 5 | hefei || 8 | beijing || 10 | xihu || 11 | changping || 12 | wangjing |+----+-----------+5 rows in set (0.00 sec) mysql-slave1和mysql-slave2两个从数据库查看mysql> select * from kevin.haha;+----+-----------+| id | name |+----+-----------+| 5 | hefei || 8 | beijing || 10 | xihu || 11 | changping || 12 | wangjing |+----+-----------+5 rows in set (0.00 sec) 在mysql-proxy代理层节点，通过proxysql进行数据写入，并查看[root@mysql-proxy ~]# mysql -uroot -ppasswd -P6033 -h127.0.0.1 -e 'insert into kevin.haha values(21,"zhongguo"),(22,"xianggang"),(23,"taiwan");'[root@mysql-proxy ~]# mysql -uroot -ppasswd -P6033 -h127.0.0.1 -e 'delete from kevin.haha where id=10' [root@mysql-proxy ~]# mysql -uroot -ppasswd -P6033 -h127.0.0.1 -e 'select * from kevin.haha;' +----+-----------+| id | name |+----+-----------+| 5 | hefei || 8 | beijing || 11 | changping || 12 | wangjing || 21 | zhongguo || 22 | xianggang || 23 | taiwan | 在mysql-master主数据库节点查看mysql> select * from kevin.haha;+----+-----------+| id | name |+----+-----------+| 5 | hefei || 8 | beijing || 11 | changping || 12 | wangjing || 21 | zhongguo || 22 | xianggang || 23 | taiwan |+----+-----------+7 rows in set (0.00 sec) 在mysql-slave1和mysql-slave2从数据库节点查看mysql> select * from kevin.haha;+----+-----------+| id | name |+----+-----------+| 5 | hefei || 8 | beijing || 11 | changping || 12 | wangjing || 21 | zhongguo || 22 | xianggang || 23 | taiwan |+----+-----------+7 rows in set (0.00 sec) 发现在客户端通过proxysql插件更新的数据，已经写到mysql-master主数据库上，并同步到mysql-slave1和mysql-slave2两个从数据库上了！==========================需要注意=======================上面测试的客户机是在mysql-proxy本机，所以用127.0.0.1.如果是在别的客户机上，那么连接命令中就用proxysql的地址，即"mysql -uroot -ppasswd -P6033 -h172.16.60.214"比如在mysql-slave2机器上连接proxysql插件：[root@mysql-slave2 ~]# mysql -uroot -ppasswd -P6033 -h172.16.60.214mysql: [Warning] Using a password on the command line interface can be insecure.Welcome to the MySQL monitor. Commands end with ; or \g.Your MySQL connection id is 47Server version: 5.5.30 (ProxySQL)Copyright (c) 2000, 2018, Oracle and/or its affiliates. All rights reserved.Oracle is a registered trademark of Oracle Corporation and/or itsaffiliates. Other names may be trademarks of their respectiveowners.Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.mysql> show databases;+--------------------+| Database |+--------------------+| information_schema || kevin || mysql || performance_schema || proxy_test || sys |+--------------------+6 rows in set (0.00 sec)删除上面创建的proxy_test测试库mysql> drop database proxy_test;Query OK, 1 row affected (0.16 sec)mysql> show databases;+--------------------+| Database |+--------------------+| information_schema || kevin || mysql || performance_schema || sys |+--------------------+5 rows in set (0.00 sec)mysql> select * from kevin.haha;+----+-----------+| id | name |+----+-----------+| 5 | hefei || 8 | beijing || 11 | changping || 12 | wangjing || 21 | zhongguo || 22 | xianggang || 23 | taiwan |+----+-----------+7 rows in set (0.00 sec)
至此，MySQL5.7基于GTID主从复制+并行复制+半同步复制+读写分离的环境部署工作已完成。