mysql5.7 MGR配置

1. 规划

2 连接master报错， 使用主机名连接，所有需要修改主机 /etc/hosts

cat /etc/hosts
127.0.0.1    localhost
172.18.12.2     master
172.18.12.3     slave1
172.18.12.4     slave2
172.18.12.5     slave3
172.18.12.6     slave4

3.设置mysql启动参数

#关于group replication 一些参数

master_info_repository = TABLE
relay_log_info_repository = TABLE
relay_log_recovery=1
gtid_mode = on
enforce_gtid_consistency = 1
log_slave_updates = ON
binlog_checksum=NONE
slave_preserve_commit_order = ON #开启并行复制相关
server_id=1    #第一个节点是1 ，第二个节点是2 ......
slave_parallel_type=LOGICAL_CLOCK
slave_parallel_workers=2-8   #根据cpu数量

transaction_write_set_extraction=XXHASH64
loose-group_replication_group_name="aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
loose-group_replication_start_on_boot=off
loose-group_replication_local_address="172.18.12.2:23306"
loose-group_replication_group_seeds="172.18.12.2:23306,172.18.12.3:23306,172.18.12.4:23306,172.18.12.5:23306,172.18.12.6:23306"
loose-grup_replication_bootstrap_group=off

4.初始化：

mysqld --initialize

5.第一个节点启动

mysqld --defaults-file=/etc/my.cnf &

6. 登录到mysql， 创建一个账户，不记录到binlog中

mysql -u root -p 

set sql_log_bin=0;

create user 'feng'@'172.18.12.%';         

grant replication slave on *.* to 'feng'@'172.18.12.%' identified by '123456';

set sql_log_bin=1;

7.  channel 的名字不能改变

root@localhost 03:07:  [(none)]> change master to master_user='feng', master_password='123456' FOR CHANNEL  'group_replication_recovery';

8. 安装 加载group_replication插件

install plugin group_replication soname 'group_replication.so';
show plugins;

9. 查看group相关参数

root@localhost 03:11:  [(none)]> show global variables  like '%group%';

10. 设置第一个节点启动为on

 set global group_replication_bootstrap_group=on;   #只在第一个节点使用

11. 启动group_replication

start group_replication;

#启动后报错

  2020-03-25T13:31:11.636568+08:00 6 [Warning] Plugin group_replication reported: 'Group Replication requires slave-preserve-commit-order to be set to ON when using more than 1 applier threads.'

  slave_preserve_commit_order=on #并行复制相关 

12. 查看是否加入到组中

select * from performance_schema.replication_group_members;

root@localhost 05:50: [performance_schema]> select * from replication_group_members;
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| CHANNEL_NAME | MEMBER_ID | MEMBER_HOST | MEMBER_PORT | MEMBER_STATE |
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| group_replication_applier | a59bc963-5772-11ea-93aa-0242ac120c02 | master | 3306 | ONLINE |
+---------------------------+--------------------------------------+-------------+-------------+--------------+

二、加入二个节点172.18.12.3 节点

2.设置mysql启动参数

#关于group replication 一些参数

master_info_repository = TABLE
relay_log_info_repository = TABLE
relay_log_recovery=1
gtid_mode = on
enforce_gtid_consistency = 1
log_slave_updates = ON
server_id=2    #第一个节点是1 ，第二个节点是2 ......
binlog_checksum=NONE

slave_preserve_commit_order = ON #开启并行复制相关
slave_parallel_type=LOCIGCAL_CLOCK
slave_parallel_workers=2-8  #根据cpu数量

transaction_write_set_extraction=XXHASH64
loose-group_replication_group_name="aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
loose-group_replication_start_on_boot=off
loose-group_replication_local_address="172.18.12.3:23306"
loose-group_replication_group_seeds="172.18.12.2:23306,172.18.12.3:23306,172.18.12.4:23306,172.18.12.5:23306,172.18.12.6:23306"
loose-grup_replication_bootstrap_group=off

初始化mysql，启动mysql

mysql --initialize

mysqld --defaults-file=/etc/my.cnf &

 5. 登录到mysql， 创建一个账户，不记录到binlog中

mysql -u root -p 

set sql_log_bin=0;

create user 'feng'@'172.18.12.%';

grant replication slave on *.* to 'feng'@'172.18.12.%' identified by '123456';

set sql_log_bin=1;

6.  channel 的名字不能改变

root@localhost 03:07:  [(none)]> change master to master_user='feng', master_password='123456' FOR CHANNEL  'group_replication_recovery';

7. 安装 加载group_replication插件

install plugin group_replication soname 'group_replication.so';
show plugins;

10. 启动group_replication

set global group_replication_allow_local_disjoint_gtids_join=ON;
start group_replication;

11. 查看状态：

root@localhost 07:12:  [(none)]> show slave status for channel 'group_replication_recovery' G;
*************************** 1. row ***************************
               Slave_IO_State: 
                  Master_Host: master
                  Master_User: rpl_user
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: 
          Read_Master_Log_Pos: 4
               Relay_Log_File: slave1-relay-bin-group_replication_recovery.000001
                Relay_Log_Pos: 4
        Relay_Master_Log_File: 
             Slave_IO_Running: No
            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: 0
              Relay_Log_Space: 1534
              Until_Condition: SQL_VIEW_ID
               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: 0
Master_SSL_Verify_Server_Cert: No
                Last_IO_Errno: 2005
                Last_IO_Error: error connecting to master 'rpl_user@master:3306' - retry-time: 60  retries: 1
               Last_SQL_Errno: 0
               Last_SQL_Error: 
  Replicate_Ignore_Server_Ids: 
             Master_Server_Id: 0
                  Master_UUID: 
             Master_Info_File: mysql.slave_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: 1
                  Master_Bind: 
      Last_IO_Error_Timestamp: 200225 07:15:55
     Last_SQL_Error_Timestamp: 
               Master_SSL_Crl: 
           Master_SSL_Crlpath: 
           Retrieved_Gtid_Set: 
            Executed_Gtid_Set: 
                Auto_Position: 1
         Replicate_Rewrite_DB: 
                 Channel_Name: group_replication_recovery
           Master_TLS_Version: 
1 row in set (0.00 sec)

ERROR: 
No query specified

再次查看：

root@localhost 07:20:  [(none)]> show slave status for channel 'group_replication_recovery' G;
*************************** 1. row ***************************
               Slave_IO_State: 
                  Master_Host: <NULL>
                  Master_User: rpl_user
                  Master_Port: 0
                Connect_Retry: 60
              Master_Log_File: 
          Read_Master_Log_Pos: 4
               Relay_Log_File: slave1-relay-bin-group_replication_recovery.000001
                Relay_Log_Pos: 4
        Relay_Master_Log_File: 
             Slave_IO_Running: No
            Slave_SQL_Running: No
              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: 0
              Relay_Log_Space: 496
              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: NULL
Master_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: 0
                  Master_UUID: 
             Master_Info_File: mysql.slave_master_info
                    SQL_Delay: 0
          SQL_Remaining_Delay: NULL
      Slave_SQL_Running_State: 
           Master_Retry_Count: 1
                  Master_Bind: 
      Last_IO_Error_Timestamp: 
     Last_SQL_Error_Timestamp: 
               Master_SSL_Crl: 
           Master_SSL_Crlpath: 
           Retrieved_Gtid_Set: 
            Executed_Gtid_Set: 09440083-2b74-11ea-838b-549f3510fe78:1-4
                Auto_Position: 1
         Replicate_Rewrite_DB: 
                 Channel_Name: group_replication_recovery
           Master_TLS_Version: 
1 row in set (0.00 sec)

ERROR: 
No query specified

11. 查看是否加入到组中

root@localhost 07:21:  [performance_schema]> select * from replication_group_members;
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| CHANNEL_NAME              | MEMBER_ID                            | MEMBER_HOST | MEMBER_PORT | MEMBER_STATE |
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| group_replication_applier | 66576b31-579b-11ea-b037-0242ac120c02 | master      |        3306 | ONLINE       |
| group_replication_applier | 6808a18a-579b-11ea-97cd-0242ac120c03 | slave1      |        3306 | ONLINE       |
| group_replication_applier | 695065b9-579b-11ea-a932-0242ac120c05 | slave3      |        3306 | ONLINE       |
| group_replication_applier | 6bc82210-579b-11ea-bb3b-0242ac120c04 | slave2      |        3306 | ONLINE       |
+---------------------------+--------------------------------------+-------------+-------------+--------------+
4 rows in set (0.00 sec)

配置出现的问题：

1 [ERROR] Plugin group_replication reported: 'Gtid mode should be ON for Group Replication'

没有打开GtID

2 Last_IO_Error: error connecting to master 'rpl_user@master:3306' - retry-time: 60  retries: 1

添加所有节点的主机名和ip的对应关系到/etc/hosts

3.  [ERROR] Plugin group_replication reported: '[GCS] Error on opening a connection to 172.17.12.6:23306 on local port: 23306.

/etc/my.cnf 配置文件 loose-group_replication_group_seeds ip地址写错了

4. 第二个节点加入MGR时报错[

ERROR] Plugin group_replication reported: 'This member has more executed transactions than those present in the group. Local transactions: f9418f39-8f6a-11ea-8e60-005056841c10:1 > Group transactions: 6c0d784d-8f66-11ea-bd2e-00505684742a:1,
aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa:1-5'
2020-05-06T16:09:46.375914+08:00 0 [ERROR] Plugin group_replication reported: 'The member contains transactions not present in the group. The member will now exit the group.'

解决办法 需要设置参数,也就是兼容加入组
set global group_replication_allow_local_disjoint_gtids_join=ON;

12 创建测试数据库（master可以写入，其他节点不能写入）

master:

root@localhost 05:52:  [fengjian]> create database fengjian;
root@localhost 05:52:  [fengjian]> use fengjian;
root@localhost 05:52:  [fengjian]> create table test(id int, name varchar(60), primary key(id));
root@localhost 05:52:  [fengjian]> insert into test values(1,'feng');


slave: 报错，无法写入

  root@localhost 08:18: [fengjian]> insert into test values(2,'su');
  ERROR 1290 (HY000): The MySQL server is running with the --super-read-only option so it cannot execute this statement

查看其他节点所有节点的read_only 选项

master

slave1

Group replication 工作模式

1. 基于single-master 环境：

在group replication 配置中，默认模式是 single-master 模式， 只有一个节点可以进行写操作，其他节点开启 --super-read-only 模式， 当主几点挂了， 其他节点会选出来一个新的master

在 single-master 上查看那个节点是 主节点：

root@localhost 08:25:  [performance_schema]> select variable_value from global_status where variable_name='group_replication_primary_member';
+--------------------------------------+
| variable_value                       |
+--------------------------------------+
| 43567e06-6e49-11ea-a1ef-0242ac110003 |
+--------------------------------------+

登录到 其他节点上，查看 uuid

root@localhost 08:27:  [fengjian]> select @@server_uuid;
+--------------------------------------+
| @@server_uuid                        |
+--------------------------------------+
| 43567e06-6e49-11ea-a1ef-0242ac110003 |
+--------------------------------------+

multi-master 模式 （多主，性能有15%的损失）

如果要开启multi-master 模式，需要在配置文件/etc/my.cnf 中指定

loose-group_replication_single_primary_mode=OFF; 
loose-group_replication_enforce_update_everywhere_checks=ON;

不管是 single-master 还是 multi-master ，我们只提供一台负责写操作

集群重启：

1. 滚动重启

节点上执行 ：  start group_replication;

2. 全部一起重启（集体掉电）

第一个节点启动(single-master 和 multi-master 都一样)

要在每个节点执行 show master stauts; 如果那个gtid多，则设置成主节点。

set global_replication_bootstrap_group=on;

start group_replication;

其他节点 start  group-replication;

MGR 监控点讲解：

可用性监控

root@localhost 09:28:  [performance_schema]> select member_id,member_host,member_state from replication_group_members where member_id=@@server_uuid;
+--------------------------------------+-------------+--------------+
| member_id                            | member_host | member_state |
+--------------------------------------+-------------+--------------+
| 66576b31-579b-11ea-b037-0242ac120c02 | master      | ONLINE       |
+--------------------------------------+-------------+--------------+

查看当前是不是可读

root@localhost 09:32:  [performance_schema]> select * from performance_schema.global_variables where variable_name in ('read_only', 'super_read_only');
+-----------------+----------------+
| VARIABLE_NAME   | VARIABLE_VALUE |
+-----------------+----------------+
| read_only       | OFF            |
| super_read_only | OFF            |
+-----------------+----------------+

为可读节点

复制是不是存在延迟：
对比获得到的GTID和本节点执行的GTID是不是一致：

从远程获取的GTID：

SELECT Received_transaction_set FROM performance_schema.replication_connection_status WHERE Channel_name = 'group_replication_applier';

本节点执行的GTID：

select  @@global.gtid_executed;   不显示结果

root@localhost 09:47: [performance_schema]> show global variables like '%gtid_executed%';
+----------------------------------+--------------------------------------------+
| Variable_name | Value |
+----------------------------------+--------------------------------------------+
| gtid_executed | 09440083-2b74-11ea-838b-549f3510fe78:1-977 |

远程获取的GTID - 本节点执行的GTID = 延迟的GTID数

本节点执行队列是不是有堆积(大于0表示有延迟）：

select count_transactions_in_queue from replication_group_member_stats where member_id=@@server_uuid;

流控(flow control)
在MGR中如果节点落后集群中其它成员太多，就会发起让其它节点等他完成在做的控制，这个叫流控。

当启用： 
group_replication_flow_control_mode=QUOTA 是表示启用流控。 流控默认通过两个参数控制：
group_replication_flow_control_applier_threshold (默认: 25000)
group_replication_flow_control_certifier_threshold (默认: 25000)

也就说默认延迟在25000个GTID时，会对整个集群Block住写操作。

当然，也可以允许，节点延迟，就如同我们主从结构，从节点延迟，不往上面发请求就可以。

关闭流控：

set global group_replication_flow_control_mode='DISABLED';

提示： 关闭流控制，注意查看是不是存在延迟，如果延迟，自已控制阀值不向上面发请求即可。 多IDC结构的MGR，建议关闭流控。

MGR调优参数

因为基本复制结构，所有的数据复制，还是逻辑的重放，所以优化也是复制优化点。
更改：

slave_parallel_type -> LOGICAL_CLOCK   #增强sql_thread个数：

slave_parallel_workers -> 2-8          #如果CPU瓶颈，网络没问题，减少CPU压缩：

group_replication_compression_threshold = 1000000 -> 2000000
由原来的1M变成2M，再进行压缩（主要针对大事务传述优化）

 

MGR备份及新节点的加入

1. 通过备份工具，建立备份
2. 加载MGR插件
3. mysqldump -u root --single-transaction -A > full_bak.sql  其他从库，备份
4. 新加入的mgr节点配置文件

　　transaction_write_set_extraction=XXHASH64
　　loose-group_replication_group_name="aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
　　loose-group_replication_start_on_boot=off
　　loose-group_replication_local_address="172.18.12.7:23306"
　　loose-group_replication_group_seeds="172.18.12.2:23306,172.18.12.3:23306,172.18.12.4:23306,172.18.12.5:23306,172.18.12.6:23306,172.18.12.7:23306"
　　loose-group_replication_bootstrap_group=off

5. mysql -u root -p  ; reset master ; mysql -u root -p < full_bak.sql
6. 可以加入集群的关键点： set global gtid_purged='xxx-xxxx-xxxx:id'
7. start group_replication;
8. use  performance_schema;  select * from replication_group_members;

root@localhost 08:52: [performance_schema]> show slave status FOR CHANNEL 'group_replication_recovery' G
*************************** 1. row ***************************
Slave_IO_State:
Master_Host: <NULL>
Master_User: rpl_user
Master_Port: 0
Connect_Retry: 60
Master_Log_File:
Read_Master_Log_Pos: 4
Relay_Log_File: slave3-relay-bin-group_replication_recovery.000001
Relay_Log_Pos: 4
Relay_Master_Log_File:
Slave_IO_Running: No
Slave_SQL_Running: No
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: 0
Relay_Log_Space: 496
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: NULL
Master_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: 0
Master_UUID:
Master_Info_File: mysql.slave_master_info
SQL_Delay: 0
SQL_Remaining_Delay: NULL
Slave_SQL_Running_State:
Master_Retry_Count: 1
Master_Bind:
Last_IO_Error_Timestamp:
Last_SQL_Error_Timestamp:
Master_SSL_Crl:
Master_SSL_Crlpath:
Retrieved_Gtid_Set:
Executed_Gtid_Set: 09440083-2b74-11ea-838b-549f3510fe78:1-896
Auto_Position: 1
Replicate_Rewrite_DB:
Channel_Name: group_replication_recovery
Master_TLS_Version:

如果binlog已经过期， start group_replication 启动报错，因为从库的gtid信息已经删除了，

引用

https://www.cnblogs.com/manger/p/7212039.html