MySQL异步复制和半同步复制怎么实现

寻技术 Mysql 2023年07月11日 161

异步复制

在异步复制(async replication)中,Master不用关心Slave是否接收到二进制日志,所以Master与Slave没有任何的依赖关系。Master和Slave可以被视为两个独立地工作的服务器,在二进制日志的作用下,它们最终能够保持数据的一致性。

异步复制的性能最好,因为它对数据库本身几乎没有任何开销,除非主从延迟非常大,Dump Thread需要读取大量二进制日志文件。

如果业务对于数据一致性要求不高,当发生故障时,能容忍数据的丢失,甚至大量的丢失,推荐用异步复制,这样性能最好(比如像微博这样的业务,虽然它对性能的要求极高,但对于数据丢失,通常可以容忍)。但往往核心业务系统最关心的就是数据安全,比如监控业务、告警系统。

异步复制环境的规划:

  • master(docker),端口3310

  • slave(docker),端口3311

master的配置

配置文件my.cnf

$ sudo cat /home/mysql/docker-data/3311/conf/my.cnf
# For advice on how to change settings please see
# http://dev.mysql.com/doc/refman/5.7/en/server-configuration-defaults.html
[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
#
# 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
#datadir=/home/mysql/docker-data/3307/data
#socket=/home/mysql/docker-data/3307/mysql.sock
character_set_server=utf8
init_connect='SET NAMES utf8'
# Disabling symbolic-links is recommended to prevent assorted security risks
symbolic-links=0
#log-error=/home/mysql/docker-data/3307/logs/mysqld.log
#pid-file=/home/mysql/docker-data/3307/mysqld.pid
lower_case_table_names=1 # 表名是否小写
server-id=1403311
log-bin=mysql-bin # 开启binlog
binlog-format=ROW # binlog的格式
auto_increment_increment=1 # 自增的步长,适用于主主复制,为了避免id冲突,步长设置为master的个数
auto_increment_offset=1 # 自增的偏移,主主复制每个master的偏移需要不一致
# binlog-do-db=mstest      # 要同步的数据库
# binlog-ignore-db=mysql  # 要忽略的数据库
#rpl_semi_sync_master_enabled=1
#rpl_semi_sync_master_timeout=10000

创建用于同步的用户:

mysql> GRANT REPLICATION SLAVE,FILE,REPLICATION CLIENT ON *.* TO 'repluser'@'%' IDENTIFIED BY '123456';
Query OK, 0 rows affected, 1 warning (0.01 sec)
mysql> FLUSH PRIVILEGES;
Query OK, 0 rows affected (0.01 sec)

在master上查看master的二进制日志:

mysql> show master status;
+------------------+----------+--------------+------------------+-------------------+
| File             | Position | Binlog_Do_DB | Binlog_Ignore_DB | Executed_Gtid_Set |
+------------------+----------+--------------+------------------+-------------------+
| mysql-bin.000003 |      1147 |              |                  |                   |
+------------------+----------+--------------+------------------+-------------------+
1 row in set (0.00 sec)

查看master上的进程列表:

mysql> show processlist;
+----+----------+------------------+-------+-------------+------+---------------------------------------------------------------+------------------+
| Id | User     | Host             | db    | Command     | Time | State                                                         | Info             |
+----+----------+------------------+-------+-------------+------+---------------------------------------------------------------+------------------+
|  2 | root     | localhost        | order | Query       |    0 | starting                                                      | show processlist |
|  6 | repluser | 172.17.0.1:48450 | NULL  | Binlog Dump |  448 | Master has sent all binlog to slave; waiting for more updates | NULL             |
+----+----------+------------------+-------+-------------+------+---------------------------------------------------------------+------------------+
2 rows in set (0.00 sec)

查看slave的状态:

mysql> show slave statusG;*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: 192.168.54.214
                  Master_User: repluser
                  Master_Port: 3310
                Connect_Retry: 60
              Master_Log_File: mysql-bin.000003
          Read_Master_Log_Pos: 1147
               Relay_Log_File: 2da789531bf3-relay-bin.000002
                Relay_Log_Pos: 320
        Relay_Master_Log_File: mysql-bin.000003
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
              Replicate_Do_DB:
          Replicate_Ignore_DB:
           Replicate_Do_Table:
       Replicate_Ignore_Table:
      Replicate_Wild_Do_Table:
  Replicate_Wild_Ignore_Table:
                   Last_Errno: 0
                   Last_Error:
                 Skip_Counter: 0
          Exec_Master_Log_Pos: 1147
              Relay_Log_Space: 534
              Until_Condition: None
               Until_Log_File:
                Until_Log_Pos: 0
           Master_SSL_Allowed: No
           Master_SSL_CA_File:
           Master_SSL_CA_Path:
              Master_SSL_Cert:
            Master_SSL_Cipher:
               Master_SSL_Key:
        Seconds_Behind_Master: 0Master_SSL_Verify_Server_Cert: No
                Last_IO_Errno: 0
                Last_IO_Error:
               Last_SQL_Errno: 0
               Last_SQL_Error:
  Replicate_Ignore_Server_Ids:
             Master_Server_Id: 1403311
                  Master_UUID: cd2eaa0a-7a59-11ec-b3b4-0242ac110002
             Master_Info_File: /var/lib/mysql/master.info
                    SQL_Delay: 0
          SQL_Remaining_Delay: NULL
      Slave_SQL_Running_State: Slave has read all relay log; waiting for more updates
           Master_Retry_Count: 86400
                  Master_Bind:
      Last_IO_Error_Timestamp:
     Last_SQL_Error_Timestamp:
               Master_SSL_Crl:
           Master_SSL_Crlpath:
           Retrieved_Gtid_Set:
            Executed_Gtid_Set:
                Auto_Position: 0
         Replicate_Rewrite_DB:
                 Channel_Name:
           Master_TLS_Version:1 row in set (0.00 sec)

只有新增的数据才会进行复制,存量的数据需要手动使用工具(如mysqldump)进行同步。

查看slave上的进程列表:

mysql> show processlist;
+----+-------------+-----------+-------+---------+------+--------------------------------------------------------+------------------+
| Id | User        | Host      | db    | Command | Time | State                                                  | Info             |
+----+-------------+-----------+-------+---------+------+--------------------------------------------------------+------------------+
|  4 | root        | localhost | order | Query   |    0 | starting                                               | show processlist |
|  7 | system user |           | NULL  | Connect |  533 | Waiting for master to send event                       | NULL             |
|  8 | system user |           | NULL  | Connect |  127 | Slave has read all relay log; waiting for more updates | NULL             |
+----+-------------+-----------+-------+---------+------+--------------------------------------------------------+------------------+
3 rows in set (0.00 sec)

半同步复制

半同步复制是MySQL 5.7版本前的半同步复制机制。

半同步复制要求Master事务提交过程中,至少有N个Slave接收到二进制日志,这样就能保证当Master发生宕机,至少有N 台Slave服务器中的数据是完整的。

半同步复制并不是MySQL内置的功能,而是要安装半同步插件,并启用半同步复制功能,设置N个Slave接受二进制日志成功。

缺点:假设user1在主库插入了一条数据,正在等待数据返回,user2这时就能查询到这条数据了,如果此时master挂了,user2又查不到这条数据了,产生了类似幻读的现象。

增强半同步复制

增强半同步复制解决了半同步复制的缺点,WAIT和ACK发生在事务提交之前,这样即便Slave没有收到二进制日志,但是 Master宕机了,由于最后一个事务还没有提交,所以本身这个数据对外也不可见,不存在丢失的问题。

所以,对于任何有数据一致性要求的业务,如电商的核心订单业务、银行、保险、证券等与资金密切相关的业务,务必使用增强半同步复制。即使出现宕机,从机也会有完整的数据备份,因此数据将是安全且可靠的。

增强半同步复制环境是基于异步复制的基础上进行的。

安装插件,建议master和slave都安装,因为会有主从切换的情景:

# master
mysql> install plugin rpl_semi_sync_master soname 'semisync_master.so';
# slave
mysql> install plugin rpl_semi_sync_slave soname 'semisync_slave.so';

确保插件安装成功:

mysql> show plugins;
... ...
| rpl_semi_sync_master       | ACTIVE   | REPLICATION        | semisync_master.so | GPL     |
| rpl_semi_sync_slave        | ACTIVE   | REPLICATION        | semisync_slave.so  | GPL     |
+----------------------------+----------+--------------------+--------------------+---------+
46 rows in set (0.00 sec)

再启动master上的半同步:

mysql> set global rpl_semi_sync_master_enabled = {0|1};  # 1:启用,0:禁止
# mysql> set global rpl_semi_sync_master_timeout = 10000;    # 单位为ms,默认为10s

从库重启io_thread:

mysql> stop slave io_thread;
mysql> start slave io_thread;

半同步的参数为什么不建议写入配置文件

参数写入配置文件的话,会使实例启动后立即进入半同步模式,如果发生长时间断连的实例重新运行启动,有可能导致主库被拖垮。

长时间断开的从库,重新连接后,要等待追完全部事务后,手动开启半同步模式,而不是启动后直接切换,防止冲击主库。

查询主库状态信息

mysql> show global status like "%semi%";
+--------------------------------------------+-------+
| Variable_name                              | Value |
+--------------------------------------------+-------+
| Rpl_semi_sync_master_clients               | 0     |
| 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     |
| Rpl_semi_sync_slave_status                 | ON    |
+--------------------------------------------+-------+
15 rows in set (0.00 sec)

重要参数说明:

  • Rpl_semi_sync_master_clients:支持和注册半同步复制的已连Slave数

  • Rpl_semi_sync_master_no_times:master关闭半同步复制的次数

  • Rpl_semi_sync_master_no_tx:master没有收到slave的回复而提交的次数,可以理解为master等待超时的次数,即半同步模式不成功提交数量

  • Rpl_semi_sync_master_status:ON是活动状态(半同步),OFF是非活动状态(异步),用于表示主服务器使用的是异步复制模式,还是半同步复制模式

  • Rpl_semi_sync_master_tx_avg_wait_time:master花在每个事务上的平均等待时间

  • Rpl_semi_sync_master_tx_waits:master等待成功的次数,即master没有等待超时的次数,也就是成功提交的次数

  • Rpl_semi_sync_master_yes_tx:master成功接收到slave的回复的次数,即半同步模式成功提交数量。

查询主库参数信息

mysql> show global variables like '%sync%';
+-------------------------------------------+------------+
| Variable_name                             | Value      |
+-------------------------------------------+------------+
| binlog_group_commit_sync_delay            | 0          |
| binlog_group_commit_sync_no_delay_count   | 0          |
| innodb_flush_sync                         | ON         |
| innodb_sync_array_size                    | 1          |
| innodb_sync_spin_loops                    | 30         |
| rpl_semi_sync_master_enabled              | ON         |
| rpl_semi_sync_master_timeout              | 10000      |
| rpl_semi_sync_master_trace_level          | 32         |
| rpl_semi_sync_master_wait_for_slave_count | 1          |
| rpl_semi_sync_master_wait_no_slave        | ON         |
| rpl_semi_sync_master_wait_point           | AFTER_SYNC |
| rpl_semi_sync_slave_enabled               | ON         |
| rpl_semi_sync_slave_trace_level           | 32         |
| sync_binlog                               | 1          |
| sync_frm                                  | ON         |
| sync_master_info                          | 10000      |
| sync_relay_log                            | 10000      |
| sync_relay_log_info                       | 10000      |
+-------------------------------------------+------------+
18 rows in set (0.01 sec)

会发现这条语句会阻塞10s(对应上面的参数rpl_semi_sync_master_timeout),然后执行成功,最后看看看master的日志描述:

2022-01-25T02:31:57.016430Z 4 [Note] Start binlog_dump to master_thread_id(4) slave_server(1403312), pos(mysql-bin.000005, 154)
2022-01-25T02:31:57.016515Z 4 [Note] Start asynchronous binlog_dump to slave (server_id: 1403312), pos(mysql-bin.000005, 154)
2022-01-25T02:33:32.183819Z 2 [Note] Semi-sync replication initialized for transactions.
2022-01-25T02:33:32.183865Z 2 [Note] Semi-sync replication enabled on the master.
2022-01-25T02:33:32.184004Z 0 [Note] Starting ack receiver thread
2022-01-25T02:33:59.644444Z 5 [Note] While initializing dump thread for slave with UUID <aba2eb12-7cbc-11ec-9c1d-0242ac110003>, found a zombie dump thread with the same UUID. Master is killing the zombie dump thread(4).
2022-01-25T02:33:59.644612Z 5 [Note] Start binlog_dump to master_thread_id(5) slave_server(1403312), pos(mysql-bin.000005, 154)
2022-01-25T02:33:59.644632Z 4 [Note] Stop asynchronous binlog_dump to slave (server_id: 1403312)
2022-01-25T02:33:59.644727Z 5 [Note] Start semi-sync binlog_dump to slave (server_id: 1403312), pos(mysql-bin.000005, 154)
2022-01-25T02:38:02.847879Z 0 [ERROR] mysqld: Got an error reading communication packets
2022-01-25T02:38:27.228952Z 2 [Warning] Timeout waiting for reply of binlog (file: mysql-bin.000005, pos: 684), semi-sync up to file mysql-bin.000005, position 419.
2022-01-25T02:38:27.229063Z 2 [Note] Semi-sync replication switched OFF.
2022-01-25T02:39:47.230189Z 5 [Note] Stop semi-sync binlog_dump to slave (server_id: 1403312)

可以发现半同步关闭了,变成异步模式。

关闭

用微信“扫一扫”