集群环境参考上篇
达梦数据守护集群部署
https://blog.csdn.net/qq_25045631/article/details/139900164

集群发散脑裂时，监视器显示如下，实例GRP1_RT_01发生脑裂

1. 关闭DW环境

因为Global守护类型的守护进程，会自动将数据库实例切换到Open状态，并将守护进程状态也切换为Open。因此在关闭DW系统时，请严格按照以下顺序：
1)如果启动了确认监视器，先关闭确认监视器（防止自动接管）
2)关闭主库守护进程（防止重启实例）
3)关闭备库守护进程（防止重启实例）
4)在关闭守护进程时会自动关闭对应的DM实例。但最好确认下ps -ef |grep dmserver，经常出现关不掉的情况

[root@node2 ~]# systemctl stop DmMonitorServiceDM
[root@node2 ~]# systemctl stop DmWatcherServiceDM
[root@node1 ~]# systemctl stop DmWatcherServiceDM

2. 删除分裂库dmwatcher.ctl

守护进程在检测到本地库(GRP1_RT_01)分裂时，自动创建dmwatcher.ctl文件，保存在本地库的 /opt/dmdbms/data/DAMENG/路径下，并且文件中记录的状态一定是 Split 分裂状态。
如果dmwatcher加载到dmwatcher.ctl文件，则认为对应的库一定是分裂状态。如果需要对分裂库进行重建，则需要手动将dmwatcher.ctl文件删除，否则守护进程仍然会认定本地库为分裂库。

脑裂节点dmwatcher.ctl文件显示如下

[dmdba@node1 ~]$ strings /opt/dmdbms/data/DAMENG/dmwatcher.ctl 
DMWATCHER8001
[!!! Local(GRP1_RT_01, PRIMARY & MOUNT & OK)'s sysopenhistory is included in remote(GRP1_RT_02, PRIMARY & OPEN & OK), but local apply info is larger or corssed with remote apply info in sysopenhistory, set local to split status.!!!]

删除dmwatcher.ctl文件
[dmdba@node1 ~]$ rm -f /opt/dmdbms/data/DAMENG/dmwatcher.ctl

3. 备份GRP1_RT_02

[dmdba@node2 ~]$ dmrman
RMAN> backup database '/opt/dmdbms/data/DAMENG/dm.ini' full backupset '/opt/dmdbms/data/DAMENG/bak/db_full_bak_split';

[dmdba@node2 ~]$ scp -r /opt/dmdbms/data/DAMENG/bak/db_full_bak_split dmdba@192.168.25.101:/opt/dmdbms/data/DAMENG/bak/

4. 恢复GRP1_RT_01

[dmdba@node1 ~]$ dmrman CTLSTMT="RESTORE DATABASE '/opt/dmdbms/data/DAMENG/dm.ini' FROM BACKUPSET '/opt/dmdbms/data/DAMENG/bak/db_full_bak_split'"

[dmdba@node1 ~]$ dmrman CTLSTMT="RECOVER DATABASE '/opt/dmdbms/data/DAMENG/dm.ini' FROM BACKUPSET '/opt/dmdbms/data/DAMENG/bak/db_full_bak_split'"

[dmdba@node1 ~]$ dmrman CTLSTMT="RECOVER DATABASE '/opt/dmdbms/data/DAMENG/dm.ini' UPDATE DB_MAGIC"

以mount模式前台启动GRP1_RT_01

[dmdba@node1 ~]$ dmserver /opt/dmdbms/data/DAMENG/dm.ini mount

打开一个新的终端
[dmdba@node1 ~]$ disql SYSDBA/SYSDBA@127.0.0.1:5236

SQL> SP_SET_PARA_VALUE(1, 'ALTER_MODE_STATUS', 1);
SQL> alter database standby;
SQL> SP_SET_PARA_VALUE(1, 'ALTER_MODE_STATUS', 0); 

操作完成后退出前台启动

5. 启动DW环境

启动备库守护进程
启动主库守护进程
启动确认监视器
[root@node1 ~]# systemctl start DmWatcherServiceDM
[root@node2 ~]# systemctl start DmWatcherServiceDM
[root@node2 ~]# systemctl start DmMonitorServiceDM

达梦社区地址
https://eco.dameng.com