达梦DM8数据守护主备集群搭建
DM8守护集群 （类似于Oracle的DataGuard） 但是通过搭建过程 感觉 部署十分方便，搭建难度小于oracle的 DataGuard
其原理也是 操作系统平台，达梦版本相同的数据库 通过主库传输 归档日志到备库，备库通过实时应用主库的归档日志，使主库和备份 数据实时同步
并且，备库可以对外提供查询服务。
搭建的过程如下：
#####修改主机名 (两台机器都做)
hostnamectl set-hostname dmdbserver2
###修改操作系统限制 (两台机器都做)
cat >> /etc/security/limits.conf << EOF
dmdba soft nproc 16384
dmdba hard nproc 16384
dmdba soft nofile 65536
dmdba hard nofile 65536
EOF
####关闭防火墙 (两台机器都做)
systemctl stop firewalld
systemctl disable firewalld
sed -i "s/SELINUX=enforcing/SELINUX=disabled/g" /etc/selinux/config
####挂载DM光盘: (两台机器都做)
mount -o loop dm8_20220701_x86_rh6_64_ent_8.1.2.128.iso /mnt
######################## 配置 YUM (两台机器都做)
/dev/sr0 3.7G 3.7G 0 100% /run/media/root/RHEL-7.1 Server.x86_64
umount /run/media/root/RHEL-7.6\ Server.x86_64/
mount /dev/cdrom /media -o loop
cd /etc/yum.repos.d
vi rhel7.repo ----------------或者--------------- local.repo
[local]
baseurl=file:///media
name=this is a local repo
enabled=1
gpgcheck=0
yum clean all;
yum list all;
###安装依赖包
yum install *glibc* -y
yum install *libXp* -y
yum install *libXt* -y
yum install *libXtst* -y
#### 下一步 修改 主库 dmdw1 的dm.ini文件
vi /dmdata/dmdw/dm.ini
MAL_INI = 1 ## 0 变为 1
ARCH_INI = 1 ## 0 变为 1
ALTER_MODE_STATUS = 0 #Whether to permit database user to alter database mode and status by SQLs, 1: yes, 0: no
ENABLE_OFFLINE_TS = 2 #Whether tablespace can be offline
#### 新建文件 vi /dmdata/dmdw/dmmal.ini
MAL_CHECK_INTERVAL = 5 #MAL链路检测时间间隔
MAL_CONN_FAIL_INTERVAL = 5 #判定MAL链路断开的时间
[MAL_INST1]
MAL_INST_NAME = dmdw1 #实例名，和dm.ini中的INSTANCE_NAME一致
MAL_HOST = 10.10.10.150 #MAL系统监听TCP连接的IP地址 心跳地址
MAL_PORT = 61141 #MAL系统监听TCP连接的端口
MAL_INST_HOST = 192.168.115.150 #实例的对外服务IP地址
MAL_INST_PORT = 5236 #实例的对外服务端口，和dm.ini中的PORT_NUM一致
MAL_DW_PORT = 52141 #实例本地的守护进程监听TCP连接的端口
MAL_INST_DW_PORT = 33141 #实例监听守护进程TCP连接的端口
[MAL_INST2]
MAL_INST_NAME = dmdw2 #### 需要修改
MAL_HOST = 10.10.10.160
MAL_PORT = 61142
MAL_INST_HOST = 192.168.115.160
MAL_INST_PORT = 5236
MAL_DW_PORT = 52142
MAL_INST_DW_PORT = 33142
######## 新建立和 配置 vi /dmdata/dmdw/dmarch.ini
su - dmdba
mkdir -p /dmdata/arch
[ARCHIVE_REALTIME]
ARCH_TYPE = REALTIME #实时归档类型
ARCH_DEST = dmdw2 #实时归档目标实例名
[ARCHIVE_LOCAL1]
ARCH_TYPE = LOCAL #本地归档类型
ARCH_DEST = /dmdata/arch #本地归档文件存放路径
ARCH_FILE_SIZE = 128 #单位Mb，本地单个归档文件最大值
ARCH_SPACE_LIMIT = 0 #单位Mb，0表示无限制，范围1024~4294967294M
######## 新建立和 配置 vi /dmdata/dmdw/dmwatcher.ini
修改dmwatcher.ini配置守护进程，配置为全局守护类型，使用自动切换模式。
[GRP1]
DW_TYPE = GLOBAL #全局守护类型
DW_MODE = AUTO #自动切换模式
DW_ERROR_TIME = 10 #远程守护进程故障认定时间
INST_RECOVER_TIME = 60 #主库守护进程启动恢复的间隔时间
INST_ERROR_TIME = 10 #本地实例故障认定时间
INST_OGUID = 453331 #守护系统唯一OGUID值
INST_INI = /dmdata/dmdw/dm.ini #dm.ini配置文件路径
INST_AUTO_RESTART = 1 #打开实例的自动启动功能
INST_STARTUP_CMD = /dm/dmdbms/bin/dmserver #命令行方式启动
RLOG_SEND_THRESHOLD = 0 #指定主库发送日志到备库的时间阀值，默认关闭
RLOG_APPLY_THRESHOLD = 0 #指定备库重演日志的时间阀值，默认关闭
主库脱机备份:
cd /dm/dmdbms/bin
./dmrman
### 一号机 dmdw1 上执行脱机备份
### 备份出来的 dbfull 是个目录
backup database '/dmdata/dmdw/dm.ini' full backupset '/dmdata/dbfullbackup';
##把脱机备份集拷贝到2号机 dmdw2
scp -rp dbfullbackup/ 192.168.115.138:/dmdata/.
### 在2号机器dmdw2进行还原
cd /dm/dmdbms/bin
./dmrman
restore database '/dmdata/dmdw/dm.ini' from backupset '/dmdata/dbfullbackup';
recover database '/dmdata/dmdw/dm.ini' from backupset '/dmdata/dbfullbackup';
recover database '/dmdata/dmdw/dm.ini' update db_magic;
#### 启动 主库 dmdw1 到 mount状态 设置 OGUID
cd /dm/dmdbms/bin
./dmserver /dmdata/dmdw/dm.ini mount
SP_SET_PARA_VALUE(1, 'ALTER_MODE_STATUS', 1);
sp_set_oguid(453331);
SP_SET_PARA_VALUE(1, 'ALTER_MODE_STATUS', 0);
alter database primary;
#### 启动 主库 dmdw2 到 mount状态 设置 OGUID
cd /dm/dmdbms/bin
./dmserver /dmdata/dmdw/dm.ini mount
SP_SET_PARA_VALUE(1, 'ALTER_MODE_STATUS', 1);
sp_set_oguid(453331);
SP_SET_PARA_VALUE(1, 'ALTER_MODE_STATUS', 0);
alter database standby;
#######主库启动守护进程
/dm/dmdbms/bin/dmwatcher /dmdata/dmdw/dmwatcher.ini
#######备库启动守护进程
/dm/dmdbms/bin/dmwatcher /dmdata/dmdw/dmwatcher.ini
######## 注册守护进程为服务的形式
su - root
cd /dm/dmdbms/script/root
./dm_service_installer.sh -t dmwatcher -watcher_ini /dmdata/dmdw/dmwatcher.ini -p dmdw1
su - root
cd /dm/dmdbms/script/root
./dm_service_installer.sh -t dmwatcher -watcher_ini /dmdata/dmdw/dmwatcher.ini -p dmdw2
######## 注册 监视器 服务的形式
su - root
vi /dmdata/dmdw/dmmonitor.ini
cd /dm/dmdbms/script/root
./dm_service_installer.sh -t dmmonitor -monitor_ini /dmdata/dmmonitor.ini -p dmdw150160
7、启动监视器
/dm/dmdbms/bin/dmmonitor /dmdata/dmmonitor_0.ini
/dm/dmdbms/bin/dmmonitor /dmdata/dmmonitor.ini
MON_DW_CONFIRM = 1 #确认监视器模式
MON_LOG_PATH = /dmdata/log #监视器日志文件存放路径
MON_LOG_INTERVAL = 60 #每隔60s定时记录系统信息到日志文件
MON_LOG_FILE_SIZE = 32 #每个日志文件最大32
MON_LOG_SPACE_LIMIT = 0 #不限定日志文件总占用空间
[GRP1]
MON_INST_OGUID = 453331 #组GRP1的唯一OGUID值
#以下配置为监视器到组GRP1的守护进程的连接信息，以―IP:PORT‖的形式配置
#IP对应dmmal.ini中的MAL_HOST，PORT对应dmmal.ini中的MAL_DW_PORT
MON_DW_IP = 192.168.115.150:52141
MON_DW_IP = 192.168.115.160:52142
其他命令:
监视器里输入:
show global info
show arch send info DMDW2
show apply stat DMDW2
#### 5 s 自动刷新显示
show i 5
stop dmwatcher grp1
startup dmwatcher grp1
#### 停止 组里所有的数据库
stop group grp1
startup group grp1
### 停止单库
login
detach database dmdw2
stop database dmdw2
startup database dmdw2
attach database dmdw2
###查看 系统运行状态
tip
切换命令 :
正常切换 switchover
login
choose switchover grp1
switchover grp1.dmdw2
switchover grp1.dmdw1
choose takeover grp1
check recover dmdw2
#=============================================================================================#
#------------------------------------FOR GLOBAL COMMAND---------------------------------------#
1.help --show help information
2.exit --exit dmmonitor
3.show version --show self dmmonitor version
4.show global info --show global database information of all groups
5.show database [group_name.]db_name --show detail database information of specified database
6.show [group_name] --show detail database information of specified group, if not specified, it will show all groups
7.show i[nterval] n --auto show database information in console every n seconds
8.q --cancel 'auto show' in console
9.list [[group_name.]db_name] --list configuration of specified dmwatcher, if not specified, it will show all dmwatchers
10.show open info [group_name.]db_name --show open history of specified database
11.show arch send info [group_name.]db_name --show source database arch send info to specified database(include recover time info)
12.show apply stat [group_name.]db_name --show specified database apply stat
13.show monitor [group_name[.]] [db_name] --show all dmmonitor connection info of specified dmwatcher
14 show state --show state of all monitors which are in the same monitor group with current monitor
15.tip --show system currently working status
16.login --login dmmonitor
17.logout --logout dmmonitor
18.get takeover time --get how much time will delay before standby instance will do takeover
---Type q/Q <return> to quit, or <return> to continue---
#=============================================================================================#
#------------------------------------FOR GROUP COMMAND----------------------------------------#
19.startup dmwatcher [group_name] --startup watching specified group
20.stop dmwatcher [group_name] --stop watching specified group
21.startup group [group_name] --startup all databases of specified group
22.stop group [group_name] --exit all databases of specified group
23.kill group [group_name] --kill all active databases of specified group
24.choose switchover [group_name] --choose databases which can do switchover
25.choose takeover [group_name] --choose databases which can do takeover
26.choose takeover force [group_name] --choose databases which can do takeover force
27.set group [group_name] auto restart on --set all instances of specified group auto restart on
28.set group [group_name] auto restart off --set all instances of specified group auto restart off
29.set group [group_name] para_name para_value --set all dmwatchers of specified group configuration(set both ini file and value in memory)
--if group_name not specified, it will notify all groups to execute
--para_name: specified parameter name, para_value: specified parameter value
--can set parameters: DW_ERROR_TIME/INST_RECOVER_TIME/INST_ERROR_TIME/INST_AUTO_RESTART/
INST_SERVICE_IP_CHECK/RLOG_SEND_THRESHOLD/RLOG_APPLY_THRESHOLD
30.set group [group_name] recover time time_value --set all standby database of specified group recover time with time_value(time_value is an integer: 3~86400(s))(only set value in dmwatcher memory)
--if group_name not specified, it will notify all groups to execute
31.set group [group_name] arch invalid --set all standby databases of specified group arch status invalid
--if group_name not specified, it will notify all groups to execute
32.clear group [group_name] arch send info --clear source database recent arch send info to all standby databases of specified group(notify source database to execute)
--if group_name not specified, it will notify all groups to execute
33.clear group [group_name] apply stat --clear all standby databases of specified group recent arch apply stat(notify all standby databases to execute)
--if group_name not specified, it will notify all groups to execute
---Type q/Q <return> to quit, or <return> to continue---
#=============================================================================================#
#------------------------------------FOR DATABASE COMMAND-------------------------------------#
34.check recover [group_name.]db_name --check specified database of specified group can be auto recovered or not
35.check open [group_name.]db_name --check specified database of specified group can be auto opened or not
36.open database [group_name.]db_name --open specified database of specified group
37.switchover [group_name[.]] [db_name] --switchover specified database of specified group as primary database
38.takeover [group_name[.]] [db_name] --use specified database of specified group to takeover as primary database
39.takeover force [group_name[.]] [db_name] --use specified database of specified group to takeover as primary database forcefully
40.set database [group_name.]db_name recover time time_value --set specified database of specified group recover time with time_value(time_value is an integer: 3~86400(s))(only set in dmwatcher memory)
41.set database [group_name.]db_name arch invalid --set specified database of specified group arch status invalid
42.detach database [group_name.]db_name --detach specified standby database from specified group
43.attach database [group_name.]db_name --attach specified standby database to specified group
44.startup dmwatcher database [group_name.]db_name --startup watching specified database
45.stop dmwatcher database [group_name.]db_name --stop watching specified database
46.startup database [group_name.]db_name --startup specified database of specified group
47.stop database [group_name.]db_name --stop specified database of specified group
48.kill database [group_name.]db_name --kill specified database of specified group
49.clear database [group_name.]db_name arch send info --clear primary database recent arch send info to specified standby database(notify primary database to execute)
50.clear database [group_name.]db_name apply stat --clear specified standby database recent arch apply stat(notify standby database to execute)
51.set database [group_name.]db_name until time time_val --set specified database of specified group archive send until time with time_val
52.cancel database [group_name.]db_name until time --cancel specified database of specified group archive send until time
---Type q/Q <return> to quit, or <return> to continue---
#=============================================================================================#
#------------------------------------FOR MPP COMMAND------------------------------------------#
53.show mpp --show mpp site info
54.startup dmwatcher all --startup watching all groups
55.stop dmwatcher all --stop watching all groups
56.startup group all --startup all database of all groups
57.stop group all --exit all database of all groups
58.kill group all --kill active database of all groups
59.check mppctl --check all active primary databases' dmmpp.ctl files are same or not
60.recover mppctl --recover all active primary databases' dmmpp.ctl files to be same
#=============================================================================================#
达梦社区地址：https://eco.dameng.com
