MySQL高可用架构之MySQL5.7组复制MGR
###########################################################################################
规划MGR:
host1.us.oracle.com 10.10.20.95
host2.us.oracle.com 10.10.20.97
host3.us.oracle.com 10.10.20.132
MySQL版本：5.7.21
MySQL实例是33061端口
server-id=330611 330612330613
###########################################################################################
host1：
MySQL实例是33061端口
mysql> select @@port;
+——–+
| @@port |
+——–+
| 33061 |
+——–+
1 row in set (0.01 sec)
[root@host1 ]#cat /etc/hosts
#FOR TEST MGR
10.10.20.132 host3.us.oracle.com
10.10.20.95 host1.us.oracle.com
10.10.20.97 host2.us.oracle.com
vi /data/mysql_33061/my.cnf
[client]
port = 33061
socket = /data/mysql_33061/run/mysql.sock
[mysqld]
port = 33061
socket = /data/mysql_33061/run/mysql.sock
pid_file = /data/mysql_33061/run/mysql.pid
basedir = /usr/local/mysql
datadir = /data/mysql_33061/data
default_storage_engine = InnoDB
max_allowed_packet = 128M
max_connections = 1024
open_files_limit = 65535 server-id=330611
skip-name-resolve
lower_case_table_names=1 character-set-server = utf8mb4
collation-server = utf8mb4_unicode_ci
init_connect='SET NAMES utf8mb4' innodb_buffer_pool_size = 4096M
innodb_buffer_pool_instances = 8
innodb_buffer_pool_load_at_startup = 1
innodb_buffer_pool_dump_at_shutdown = 1
innodb_file_per_table = 1
innodb_flush_log_at_trx_commit = 2
innodb_data_home_dir=/data/mysql_33061/data/
innodb_log_group_home_dir=/data/mysql_33061/logs/
innodb_log_files_in_group=3
innodb_log_file_size=512M
innodb_data_file_path=ibdata1:512M:autoextend
log_slave_updates=1
binlog_format=ROW
key_buffer_size = 16M
innodb_log_buffer_size = 32M
innodb_io_capacity = 4000
innodb_io_capacity_max = 8000
innodb_flush_method = O_DIRECT
innodb_print_all_deadlocks = 1 log-error = /data/mysql_33061/logs/mysql_error.log
log-bin = /data/mysql_33061/logs/mysql_bin
slow_query_log = 1
slow_query_log_file = /data/mysql_33061/logs/mysql_slow_query.log
long_query_time = 2
sync_binlog = 600
relay_log_recovery = 1 tmp_table_size = 64M
max_heap_table_size = 64M #FOR MGR
transaction_write_set_extraction = XXHASH64
loose-group_replication_group_name = 'ce9be252-2b71-11e6-b8f4-00212889f856' #不能用主机的GTID
loose-group_replication_start_on_boot = off
loose-group_replication_bootstrap_group = off
loose-group_replication_ip_whitelist = '10.10.20.95,10.10.20.132,10.10.20.97'
loose-group_replication_local_address = '10.10.20.95:33066' #写自己主机所在IP，端口自选
loose-group_replication_group_seeds ='10.10.20.95:33066,10.10.20.97:33066,10.10.20.132:33066'#写MGR组内主机所在IP，端口自选
loose-group_replication_single_primary_mode = off #关闭单主模式的参数
loose-group_replication_enforce_update_everywhere_checks = off #开启多主模式的参数
init_slave = 'set sql_mode=STRICT_ALL_TABLES'
#自动启动组复制
#loose-group_replication_start_on_boot = on
#group_replication_allow_local_disjoint_gtids_join = on #FOR GTID
gtid_mode=on
enforce_gtid_consistency=on
binlog_gtid_simple_recovery=1
log_slave_updates=1
binlog_checksum=NONE
master_info_repository=TABLE
relay_log_info_repository=TABLE
my.cnf
mysql>INSTALL PLUGIN group_replication SONAME ‘group_replication.so’;
mysql> show plugins;
+—————————-+———-+——————–+———————-+———+
……
| BLACKHOLE | ACTIVE | STORAGE ENGINE | NULL | GPL |
| FEDERATED | DISABLED | STORAGE ENGINE | NULL | GPL |
| ngram | ACTIVE | FTPARSER | NULL | GPL |
| group_replication | ACTIVE | GROUP REPLICATION | group_replication.so | GPL |
+—————————-+———-+——————–+———————-+———+
45 rows in set (0.00 sec)
mysql>GRANT REPLICATION SLAVE ON *.* TO ‘dbsync’@’10.10.%.%’ IDENTIFIED BY ‘xxxxxxx’;
Query OK, 0 rows affected, 1 warning (0.00 sec)
mysql> CHANGE MASTER TO MASTER_USER=’dbsync’,MASTER_PASSWORD=’xxxxxxx’ FOR CHANNEL ‘group_replication_recovery’;
Query OK, 0 rows affected, 2 warnings (0.27 sec)
mysql> set global group_replication_ip_whitelist=”10.10.20.95,10.10.20.132,10.10.20.97″;
Query OK, 0 rows affected (0.00 sec)
mysql> SET GLOBAL group_replication_bootstrap_group=ON;
Query OK, 0 rows affected (0.00 sec)
mysql> set auto_increment_offset=5555;
Query OK, 0 rows affected (0.00 sec)
mysql> START GROUP_REPLICATION;
Query OK, 0 rows affected (2.04 sec)
mysql> SELECT * FROM performance_schema.replication_group_members;
+—————————+————————————–+———————–+————-+————–+
| CHANNEL_NAME | MEMBER_ID | MEMBER_HOST | MEMBER_PORT | MEMBER_STATE |
+—————————+————————————–+———————–+————-+————–+
| group_replication_applier | 57bc1e44-a447-11e8-8e78-5254000f067d | host1.us.oracle.com | 33061 | ONLINE |
+—————————+————————————–+———————–+————-+————–+
1 row in set (0.00 sec)
###########################################################################################
host2:
[root@host2 ]#cat /etc/hosts
#FOR TEST MGR
10.10.20.132 host3.us.oracle.com
10.10.20.95 host1.us.oracle.com
10.10.20.97 host2.us.oracle.com
vi /data/mysql_33061/my.cnf
[client]
port = 33061
socket = /data/mysql_33061/run/mysql.sock [mysqld]
port = 33061
socket = /data/mysql_33061/run/mysql.sock
pid_file = /data/mysql_33061/run/mysql.pid
basedir = /usr/local/mysql
datadir = /data/mysql_33061/data
default_storage_engine = InnoDB
max_allowed_packet = 128M
max_connections = 1024
open_files_limit = 65535 skip-name-resolve
lower_case_table_names=1 character-set-server = utf8mb4
collation-server = utf8mb4_unicode_ci
init_connect='SET NAMES utf8mb4' innodb_buffer_pool_size = 4096M
innodb_buffer_pool_instances = 8
innodb_buffer_pool_load_at_startup = 1
innodb_buffer_pool_dump_at_shutdown = 1
innodb_file_per_table = 1
innodb_flush_log_at_trx_commit = 2
innodb_data_home_dir=/data/mysql_33061/data/
innodb_log_group_home_dir=/data/mysql_33061/logs/
innodb_log_files_in_group=3
innodb_log_file_size=512M
innodb_data_file_path=ibdata1:512M:autoextend
log_slave_updates=1
binlog_format=ROW
key_buffer_size = 16M
innodb_log_buffer_size = 32M
innodb_io_capacity = 4000
innodb_io_capacity_max = 8000
innodb_flush_method = O_DIRECT
innodb_print_all_deadlocks = 1 log-error = /data/mysql_33061/logs/mysql_error.log
log-bin = /data/mysql_33061/logs/mysql_bin.log
slow_query_log = 1
slow_query_log_file = /data/mysql_33061/logs/mysql_slow_query.log
long_query_time = 2
sync_binlog = 600
relay_log_recovery = 1 tmp_table_size = 64M
max_heap_table_size = 64M server-id=330612 #FOR MGR
transaction_write_set_extraction = XXHASH64
loose-group_replication_group_name = 'ce9be252-2b71-11e6-b8f4-00212889f856' #不能用主机的GTID
loose-group_replication_start_on_boot = off
loose-group_replication_bootstrap_group = off
loose-group_replication_ip_whitelist = '10.10.20.95,10.10.20.132,10.10.20.97'
loose-group_replication_local_address = '10.10.20.97:33066' #写自己主机所在IP，端口自选
loose-group_replication_group_seeds ='10.10.20.95:33066,10.10.20.97:33066,10.10.20.132:33066'#写MGR组内主机所在IP，端口自选
loose-group_replication_single_primary_mode = off #关闭单主模式的参数
loose-group_replication_enforce_update_everywhere_checks = off #开启多主模式的参数
init_slave = 'set sql_mode=STRICT_ALL_TABLES' #FOR GTID
gtid_mode=on
enforce_gtid_consistency=on
binlog_gtid_simple_recovery=1
log_slave_updates=1
binlog_checksum=NONE
master_info_repository=TABLE
relay_log_info_repository=TABLE
my.cnf
mysql>INSTALL PLUGIN group_replication SONAME ‘group_replication.so’;
mysql> CHANGE MASTER TO MASTER_USER=’dbsync’,MASTER_PASSWORD=’xxxxxxx’ FOR CHANNEL ‘group_replication_recovery’;
Query OK, 0 rows affected, 2 warnings (0.56 sec)
mysql> set global group_replication_ip_whitelist=”10.10.20.95,10.10.20.132,10.10.20.97″;
Query OK, 0 rows affected (0.00 sec)
mysql> set auto_increment_offset=5555;
Query OK, 0 rows affected (0.00 sec)
mysql> set global group_replication_allow_local_disjoint_gtids_join=ON;
Query OK, 0 rows affected, 1 warning (0.00 sec)
mysql> START GROUP_REPLICATION;
Query OK, 0 rows affected, 1 warning (6.88 sec)
###########################################################################################
host3:
[root@host3 ]#cat /etc/hosts
#FOR TEST MGR
10.10.20.132 host3.us.oracle.com
10.10.20.95 host1.us.oracle.com
10.10.20.97 host2.us.oracle.com
vi /data/mysql_33061/my.cnf
[client]
port = 33061
socket = /data/mysql_33061/run/mysql.sock [mysqld]
port = 33061
socket = /data/mysql_33061/run/mysql.sock
pid_file = /data/mysql_33061/run/mysql.pid
basedir = /usr/local/mysql
datadir = /data/mysql_33061/data
default_storage_engine = InnoDB
max_allowed_packet = 128M
max_connections = 1024
open_files_limit = 65535 skip-name-resolve
lower_case_table_names=1 character-set-server = utf8mb4
collation-server = utf8mb4_unicode_ci
init_connect='SET NAMES utf8mb4' innodb_buffer_pool_size = 4096M
innodb_buffer_pool_instances = 8
innodb_buffer_pool_load_at_startup = 1
innodb_buffer_pool_dump_at_shutdown = 1
innodb_file_per_table = 1
innodb_flush_log_at_trx_commit = 2
innodb_data_home_dir=/data/mysql_33061/data/
innodb_log_group_home_dir=/data/mysql_33061/logs/
innodb_log_files_in_group=3
innodb_log_file_size=512M
innodb_data_file_path=ibdata1:512M:autoextend
log_slave_updates=1
binlog_format=ROW
key_buffer_size = 16M
innodb_log_buffer_size = 32M
innodb_io_capacity = 4000
innodb_io_capacity_max = 8000
innodb_flush_method = O_DIRECT
innodb_print_all_deadlocks = 1 log-error = /data/mysql_33061/logs/mysql_error.log
log-bin = /data/mysql_33061/logs/mysql_bin.log
slow_query_log = 1
slow_query_log_file = /data/mysql_33061/logs/mysql_slow_query.log
long_query_time = 2
sync_binlog = 600
relay_log_recovery = 1 tmp_table_size = 64M
max_heap_table_size = 64M server-id=330613 #FOR MGR
transaction_write_set_extraction = XXHASH64
loose-group_replication_group_name = 'ce9be252-2b71-11e6-b8f4-00212889f856' #不能用主机的GTID
loose-group_replication_start_on_boot = off
loose-group_replication_bootstrap_group = off
loose-group_replication_ip_whitelist = '10.10.20.95,10.10.20.132,10.10.20.97'
loose-group_replication_local_address = '10.10.20.132:33066' #写自己主机所在IP，端口自选
loose-group_replication_group_seeds ='10.10.20.95:33066,10.10.20.97:33066,10.10.20.132:33066'#写MGR组内主机所在IP，端口自选
loose-group_replication_single_primary_mode = off #关闭单主模式的参数
loose-group_replication_enforce_update_everywhere_checks = off #开启多主模式的参数
init_slave = 'set sql_mode=STRICT_ALL_TABLES' #FOR GTID
gtid_mode=on
enforce_gtid_consistency=on
binlog_gtid_simple_recovery=1
log_slave_updates=1
binlog_checksum=NONE
master_info_repository=TABLE
relay_log_info_repository=TABLE
my.cnf
mysql>INSTALL PLUGIN group_replication SONAME ‘group_replication.so’;
mysql> CHANGE MASTER TO MASTER_USER=’dbsync’,MASTER_PASSWORD=’xxxxxxx’ FOR CHANNEL ‘group_replication_recovery’;
Query OK, 0 rows affected, 2 warnings (0.56 sec)
mysql> set global group_replication_ip_whitelist=”10.10.20.95,10.10.20.132,10.10.20.97″;
Query OK, 0 rows affected (0.00 sec)
mysql> set auto_increment_offset=5555;
Query OK, 0 rows affected (0.00 sec)
mysql> set global group_replication_allow_local_disjoint_gtids_join=ON;
Query OK, 0 rows affected, 1 warning (0.00 sec)
mysql> START GROUP_REPLICATION;
Query OK, 0 rows affected, 1 warning (3.27 sec)
mysql> SELECT * FROM performance_schema.replication_group_members;
+—————————+————————————–+———————–+————-+————–+
| CHANNEL_NAME | MEMBER_ID | MEMBER_HOST | MEMBER_PORT | MEMBER_STATE |
+—————————+————————————–+———————–+————-+————–+
| group_replication_applier | 041eac3b-a40a-11e8-b11f-545200297bd8 | host2.us.oracle.com | 33061 | ONLINE |
| group_replication_applier | 57bc1e44-a447-11e8-8e78-5254000f067d | host1.us.oracle.com | 33061 | ONLINE |
| group_replication_applier | e3d6b2db-a44e-11e8-8769-545200494892 | host3.us.oracle.com | 33061 | ONLINE |
+—————————+————————————–+———————–+————-+————–+
3 rows in set (0.00 sec)
host1:
mysql> SELECT * FROM performance_schema.replication_group_members;
+—————————+————————————–+———————–+————-+————–+
| CHANNEL_NAME | MEMBER_ID | MEMBER_HOST | MEMBER_PORT | MEMBER_STATE |
+—————————+————————————–+———————–+————-+————–+
| group_replication_applier | 041eac3b-a40a-11e8-b11f-545200297bd8 | host2.us.oracle.com | 33061 | ONLINE |
| group_replication_applier | 57bc1e44-a447-11e8-8e78-5254000f067d | host1.us.oracle.com | 33061 | ONLINE |
| group_replication_applier | e3d6b2db-a44e-11e8-8769-545200494892 | host3.us.oracle.com | 33061 | ONLINE |
+—————————+————————————–+———————–+————-+————–+
3 rows in set (0.00 sec)
host2:
mysql> SELECT * FROM performance_schema.replication_group_members;
+—————————+————————————–+———————–+————-+————–+
| CHANNEL_NAME | MEMBER_ID | MEMBER_HOST | MEMBER_PORT | MEMBER_STATE |
+—————————+————————————–+———————–+————-+————–+
| group_replication_applier | 041eac3b-a40a-11e8-b11f-545200297bd8 | host2.us.oracle.com | 33061 | ONLINE |
| group_replication_applier | 57bc1e44-a447-11e8-8e78-5254000f067d | host1.us.oracle.com | 33061 | ONLINE |
| group_replication_applier | e3d6b2db-a44e-11e8-8769-545200494892 | host3.us.oracle.com | 33061 | ONLINE |
+—————————+————————————–+———————–+————-+————–+
3 rows in set (0.00 sec)
查询主节点：
select variable_value from performance_schema.global_status where variable_name=’group_replication_primary_member’;
查看当前实例是否有延迟等性能问题（通过队列深度判断）的方法：
select * from performance_schema.replication_group_member_stats\G;
通过查看COUNT_TRANSACTIONS_IN_QUEUE的值，可以判断等待处理的事务数。
参考：
http://www.cnblogs.com/xinysu/p/6674832.html
https://www.cnblogs.com/manger/p/7211932.html
