由於的我mgr建立在虛擬機器上面（即使是正式環境，如果計劃內的停機或者斷電都需要關閉所有的節點），如何關閉所有的組成員，關閉的順序還是比較重要的。

我的環境是一個primary，多個slave的架構，qht131為parmary,其它qht132,qht133,qht134為slave資料庫。

1.關閉所有的組成員必須先先關閉slave，最後關閉master.

qht132,qht133,qht134：

[[email protected] mysql]# service mysql stop
Shutting down MySQL..............                          [  OK  ]

qht131:

所有的slave關閉後都被移出了組成員：

mysql> select * from performance_schema.replication_group_members ;
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| CHANNEL_NAME              | MEMBER_ID                            | MEMBER_HOST | MEMBER_PORT | MEMBER_STATE |
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| group_replication_applier | bb0dea82-58ed-11e8-94e5-000c29e8e89b | qht131      |        3306 | ONLINE       |
+---------------------------+--------------------------------------+-------------+-------------+--------------+
1 row in set (0.00 sec)
 

最後也關閉primary資料庫

[[email protected] mysql]# service mysql stop
Shutting down MySQL..............                          [  OK  ]

2.開啟組複製需要先開始primary，接著一個啟動slave server.

[[email protected] mysql]# service mysql start
Starting MySQL...........                                  [  OK  ]

由於group_replication_bootstrap_group引數 我們在配置檔案中設為了OFF，所以需要手動開啟設為ON.

mysql> show variables like 'group_replication_bootstrap_group';
+-----------------------------------+-------+
| Variable_name                     | Value |
+-----------------------------------+-------+
| group_replication_bootstrap_group | OFF   |
+-----------------------------------+-------+
1 row in set (0.24 sec)

mysql> set global group_replication_bootstrap_group=on;
Query OK, 0 rows affected (0.00 sec)

mysql> select * from performance_schema.replication_group_members ;
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| CHANNEL_NAME              | MEMBER_ID                            | MEMBER_HOST | MEMBER_PORT | MEMBER_STATE |
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| group_replication_applier | bb0dea82-58ed-11e8-94e5-000c29e8e89b | qht131      |        3306 | OFFLINE      |
+---------------------------+--------------------------------------+-------------+-------------+--------------+
1 row in set (0.02 sec)

在沒有開啟組複製之前member_state為OFFLINE

開啟組複製之後記得將group_replication_bootstrap_group再設為off

mysql> start group_replication;
Query OK, 0 rows affected (2.12 sec)

mysql> set global group_replication_bootstrap_group=off;
Query OK, 0 rows affected (0.00 sec)

mysql>  select * from performance_schema.replication_group_members ;
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| CHANNEL_NAME              | MEMBER_ID                            | MEMBER_HOST | MEMBER_PORT | MEMBER_STATE |
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| group_replication_applier | bb0dea82-58ed-11e8-94e5-000c29e8e89b | qht131      |        3306 | ONLINE       |
+---------------------------+--------------------------------------+-------------+-------------+--------------+
1 rows in set (0.00 sec)

待primary server準備好後，其它的slave資料庫可以一個個開啟並開啟組複製

qht132,qht133,qht134:

[[email protected] mysql]# service mysql start
Starting MySQL.......                                      [  OK  ]
mysql> start group_replication;
Query OK, 0 rows affected (6.52 sec)

可能出現的問題：

mysql> start group_replication;
ERROR 3092 (HY000): The server is not configured properly to be an active member of the group. Please see more details on error log.

同時錯誤日誌：

2018-05-17T03:08:40.654242Z 2 [ERROR] Plugin group_replication reported: '[GCS] The member is leaving a group without being on one.'
2018-05-17T03:08:40.654369Z 2 [Note] Plugin group_replication reported: 'auto_increment_increment is reset to 1'
2018-05-17T03:08:40.654378Z 2 [Note] Plugin group_replication reported: 'auto_increment_offset is reset to 1'
2018-05-17T03:08:40.654551Z 7 [Note] Error reading relay log event for channel 'group_replication_applier': slave SQL thread was killed

2018-05-17T03:08:40.657610Z 4 [Note] Plugin group_replication reported: 'The group replication applier thread was killed'

解決辦法：檢查一下primary server組複製的狀態，group_replication_bootstrap_group是否為on，performance_schema.replication_group_members是否為ONLINE。

全部啟動好後：

mysql> select * from performance_schema.replication_group_members ;
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| CHANNEL_NAME              | MEMBER_ID                            | MEMBER_HOST | MEMBER_PORT | MEMBER_STATE |
+---------------------------+--------------------------------------+-------------+-------------+--------------+
| group_replication_applier | bb0dea82-58ed-11e8-94e5-000c29e8e89b | qht131      |        3306 | ONLINE       |
| group_replication_applier | bb0dea82-58ed-11e8-94e5-000c29e8e89c | qht134      |        3306 | ONLINE       |
| group_replication_applier | bb0dea82-58ed-11e8-94e5-000c29e8e90b | qht132      |        3306 | ONLINE       |
| group_replication_applier | bb0dea82-58ed-11e8-94e5-000c29e8e91b | qht133      |        3306 | ONLINE       |
+---------------------------+--------------------------------------+-------------+-------------+--------------+
4 rows in set (0.00 sec)