在 DM MPP系统的运行过程中,某一 EP节点 发生故障停机,则整个 MPP 系统将处于不能正常服务的状态。当前所有的用户会话会被系统断开,不能进行全局登录,只能进行本地登录。 因此,为了保证 MPP 系统的高可用性,强烈建议采用 DM MPP 与数据守护相结合的部署方案。
一、集群规划
整个服务器集群需要三台服务器完成,其中A、B两台服务器互为主备,C服务器为监视器用于实现集群的故障自动切换。
服务器 | 业务IP | 系统监听IP | 实例名 | OGUID | 实例端口 | MAL端口 | MAL守护进程端口 | 守护进程端口 |
---|---|---|---|---|---|---|---|---|
A服务器 | 192.168.177.129 | 192.168.177.129 | 45331 | MPP1_01 | 5236 | 5336 | 5436 | 5536 |
B服务器 | 192.168.177.130 | 192.168.177.130 | 45332 | MPP1_02 | 5236 | 5336 | 5436 | 5536 |
A服务器 | 192.168.177.129 | 192.168.177.129 | 45332 | MPP1_02B | 5237 | 5337 | 5437 | 5537 |
B服务器 | 192.168.177.130 | 192.168.177.130 | 45331 | MPP1_01B | 5237 | 5337 | 5437 | 5537 |
C服务器 | 192.168.177.131 | 确认监视 |
安装目录 /dm8
实例目录 /dm8/data/
二、集群部署
1.配置A机器-MPP1_01库
–初始化实例
[dmdba@~]# /dm8/bin/dminit PATH=/dm8/data/ DB_NAME=MPP1_01 INSTANCE_NAME=MPP1_01 PORT_NUM=5236 PAGE_SIZE=32 LOG_SIZE=2048
–启动服务
[dmdba@~]# /dm8/bin/dmserver /dm8/data/MPP1_01/dm.ini
–开启归档
[dmdba@~]# /dm8/bin/disql SYSDBA/SYSDBA@172.16.1.1:5236
SQL> ALTER DATABASE MOUNT;
SQL> ALTER DATABASE ARCHIVELOG;
SQL> ALTER DATABASE ADD ARCHIVELOG 'DEST=/dm8/data/MPP1_01/arch, TYPE=LOCAL, FILE_SIZE=1024, SPACE_LIMIT=51200';
SQL> ALTER DATABASE OPEN;
–备份数据
SQL> BACKUP DATABASE BACKUPSET '/dm8/data/MPP1_01/BACKUP_FILE';
–修改dm.ini
SP_SET_PARA_VALUE (2,'PORT_NUM',5236);
SP_SET_PARA_VALUE (2,'DW_INACTIVE_INTERVAL',60);
SP_SET_PARA_VALUE (2,'ALTER_MODE_STATUS',0);
SP_SET_PARA_VALUE (2,'ENABLE_OFFLINE_TS',2);
SP_SET_PARA_VALUE (2,'MAL_INI',1);
SP_SET_PARA_VALUE (2,'ARCH_INI',1);
SP_SET_PARA_VALUE (2,'RLOG_SEND_APPLY_MON',64);
SP_SET_PARA_VALUE (2,'MPP_INI',1);
–配置dmarch.ini
[dmdba@~]# vi /dm8/data/MPP1_01/dmarch.ini
[ARCHIVE_LOCAL]
ARCH_TYPE = LOCAL #本地归档类型
ARCH_DEST = /dm8/data/MPP1_01/arch/ #本地归档存放路径
ARCH_FILE_SIZE = 1024 #单个归档大小,单位MB
ARCH_SPACE_LIMIT = 51200 #归档上限,单位MB
[ARCHIVE_REALTIME]
ARCH_TYPE = REALTIME #实时归档类型
ARCH_DEST = MPP1_01B #实时归档目标实例名
–创建dmmal.ini
[dmdba@~]# vi /dm8/data/MPP1_01/dmmal.ini
MAL_CHECK_INTERVAL = 10 #MAL链路检测时间间隔
MAL_CONN_FAIL_INTERVAL = 10 #判定MAL链路断开的时间
MAL_TEMP_PATH = /dm8/data/malpath/MPP1_01/ #临时文件目录
MAL_BUF_SIZE = 512 #单个MAL缓存大小,单位MB
MAL_SYS_BUF_SIZE = 2048 #MAL总大小限制,单位MB
MAL_COMPRESS_LEVEL = 0 #MAL消息压缩等级,0表示不压缩
[MAL_INST1]
MAL_INST_NAME = MPP1_01 #实例名,和 dm.ini中INSTANCE_NAME一致
MAL_HOST = 192.168.177.129 #MAL系统监听TCP连接的IP地址
MAL_PORT = 5336 #MAL系统监听TCP连接的端口
MAL_INST_HOST = 192.168.177.129 #实例的对外服务IP地址
MAL_INST_PORT = 5236 #实例对外服务端口,和dm.ini中PORT_NUM一致
MAL_DW_PORT = 5436 #实例对应的守护进程监听TCP连接的端口
MAL_INST_DW_PORT = 5536 #实例监听守护进程TCP连接的端口
[MAL_INST2]
MAL_INST_NAME = MPP1_01B
MAL_HOST = 192.168.177.130
MAL_PORT = 5337
MAL_INST_HOST = 192.168.177.130
MAL_INST_PORT = 5237
MAL_DW_PORT = 5437
MAL_INST_DW_PORT = 5537
[MAL_INST3]
MAL_INST_NAME = MPP1_02
MAL_HOST = 192.168.177.130
MAL_PORT = 5336
MAL_INST_HOST = 192.168.177.130
MAL_INST_PORT = 5236
MAL_DW_PORT = 5436
MAL_INST_DW_PORT = 5536
[MAL_INST4]
MAL_INST_NAME = MPP1_02B
MAL_HOST = 192.168.177.129
MAL_PORT = 5337
MAL_INST_HOST = 192.168.177.129
MAL_INST_PORT = 5237
MAL_DW_PORT = 5437
MAL_INST_DW_PORT = 5537
–创建dmmpp.ini
[dmdba@~]# vi /dm8/data/MPP1_01/dmmpp.ini
[service_name1]
mpp_seq_no = 0
mpp_inst_name = MPP1_01
[service_name2]
mpp_seq_no = 1
mpp_inst_name = MPP1_02
–创建dmmpp.ctl
[dmdba@~]# /dm8/bin/dmctlcvt t2c /dm8/data/MPP1_01/dmmpp.ini /dm8/data/MPP1_01/dmmpp.ctl
–创建dmwatcher.ini
[dmdba@~]# vi /dm8/data/MPP1_01/dmwatcher.ini
[GMPP1_01]
DW_TYPE = GLOBAL #全局守护类型
DW_MODE = MANUAL #故障手动切换模式
DW_ERROR_TIME = 20 #远程守护进程故障认定时间
INST_ERROR_TIME = 20 #本地实例故障认定时间
INST_RECOVER_TIME = 60 #主库守护进程启动恢复的间隔时间
INST_OGUID = 45331 #守护系统唯一OGUID值
INST_INI = /dm8/data/MPP1_01/dm.ini #dm.ini文件路径
INST_AUTO_RESTART = 1 #打开实例的自动启动功能
INST_STARTUP_CMD = /dm8/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阈值,默认关闭
RLOG_APPLY_THRESHOLD = 0 #指定备库重演日志的时间阈值,默认关闭
[GMPP1_02]
DW_TYPE = GLOBAL
DW_MODE = MANUAL
DW_ERROR_TIME = 20
INST_ERROR_TIME = 20
INST_RECOVER_TIME = 60
INST_OGUID = 45332
INST_INI = /dm8/data/MPP1_02/dm.ini
INST_AUTO_RESTART = 1
INST_STARTUP_CMD = /dm8/bin/dmserver
RLOG_SEND_THRESHOLD = 0
RLOG_APPLY_THRESHOLD = 0
–拷贝实例到B机器
[dmdba@~]# scp -r /dm8/data/MPP1_01 dmdba@192.168.177.139:/dm8/data/
–注册服务
[root@~]# /dm8/script/root/dm_service_installer.sh -t dmserver -p MPP1_01 -dm_ini /dm8/data/MPP1_01/dm.ini -m mount
[root@~]# /dm8/script/root/dm_service_installer.sh -t dmwatcher -p MPP1_01 -watcher_ini /dm8/data/MPP1_01/dmwatcher.ini
备注:删除自启
[root@~]# /dm8/script/root/dm_service_uninstaller.sh -n DmServiceMPP1_01
[root@~]# /dm8/script/root/dm_service_uninstaller.sh -n DmWatcherServiceMPP1_01
2.配置B机器-MPP1_02库
–初始化实例
[dmdba@~]# /dm8/bin/dminit PATH=/dm8/data/ DB_NAME=MPP1_02 INSTANCE_NAME=MPP1_02 PORT_NUM=5236 PAGE_SIZE=32 LOG_SIZE=2048
–启动服务
[dmdba@~]# /dm8/bin/dmserver /dm8/data/MPP1_02/dm.ini
–开启归档
[dmdba@~]# /dm8/bin/disql SYSDBA/SYSDBA@172.16.1.2:5236
SQL> ALTER DATABASE MOUNT;
SQL> ALTER DATABASE ARCHIVELOG;
SQL> ALTER DATABASE ADD ARCHIVELOG 'DEST=/dm8/data/MPP1_02/arch, TYPE=LOCAL, FILE_SIZE=1024, SPACE_LIMIT=51200';
SQL> ALTER DATABASE OPEN;
–备份数据
SQL> BACKUP DATABASE BACKUPSET '/dm8/data/MPP1_02/bak/BACKUP_FILE';
–修改dm.ini
SQL> SP_SET_PARA_VALUE (2,'PORT_NUM',5236);
SQL> SP_SET_PARA_VALUE (2,'DW_INACTIVE_INTERVAL',60);
SQL> SP_SET_PARA_VALUE (2,'ALTER_MODE_STATUS',0);
SQL> SP_SET_PARA_VALUE (2,'ENABLE_OFFLINE_TS',2);
SQL> SP_SET_PARA_VALUE (2,'MAL_INI',1);
SQL> SP_SET_PARA_VALUE (2,'MPP_INI',1);
SQL> SP_SET_PARA_VALUE (2,'RLOG_SEND_APPLY_MON',64);
—替换dmarch.ini
[dmdba@~]# vi /dm8/data/MPP1_02/dmarch.ini
[ARCHIVE_LOCAL]
ARCH_TYPE = LOCAL #本地归档类型
ARCH_DEST = /dm8/data/MPP1_02/arch/ #本地归档存放路径
ARCH_FILE_SIZE = 1024 #单个归档大小,单位MB
ARCH_SPACE_LIMIT = 51200 #归档上限,单位MB
[ARCHIVE_REALTIME]
ARCH_TYPE = REALTIME #实时归档类型
ARCH_DEST = MPP1_02B #实时归档目标实例名
—创建dmmal.ini
[dmdba@~]# vi /dm8/data/MPP1_01/dmmal.ini
MAL_CHECK_INTERVAL = 10 #MAL链路检测时间间隔
MAL_CONN_FAIL_INTERVAL = 10 #判定MAL链路断开的时间
MAL_TEMP_PATH = /dm8/data/malpath/MPP1_01/ #临时文件目录
MAL_BUF_SIZE = 512 #单个MAL缓存大小,单位MB
MAL_SYS_BUF_SIZE = 2048 #MAL总大小限制,单位MB
MAL_COMPRESS_LEVEL = 0 #MAL消息压缩等级,0表示不压缩
[MAL_INST1]
MAL_INST_NAME = MPP1_01 #实例名,和 dm.ini中INSTANCE_NAME一致
MAL_HOST = 192.168.177.129 #MAL系统监听TCP连接的IP地址
MAL_PORT = 5336 #MAL系统监听TCP连接的端口
MAL_INST_HOST = 192.168.177.129 #实例的对外服务IP地址
MAL_INST_PORT = 5236 #实例对外服务端口,和dm.ini中PORT_NUM一致
MAL_DW_PORT = 5436 #实例对应的守护进程监听TCP连接的端口
MAL_INST_DW_PORT = 5536 #实例监听守护进程TCP连接的端口
[MAL_INST2]
MAL_INST_NAME = MPP1_01B
MAL_HOST = 192.168.177.130
MAL_PORT = 5337
MAL_INST_HOST = 192.168.177.130
MAL_INST_PORT = 5237
MAL_DW_PORT = 5437
MAL_INST_DW_PORT = 5537
[MAL_INST3]
MAL_INST_NAME = MPP1_02
MAL_HOST = 192.168.177.130
MAL_PORT = 5336
MAL_INST_HOST = 192.168.177.130
MAL_INST_PORT = 5236
MAL_DW_PORT = 5436
MAL_INST_DW_PORT = 5536
[MAL_INST4]
MAL_INST_NAME = MPP1_02B
MAL_HOST = 192.168.177.129
MAL_PORT = 5337
MAL_INST_HOST = 192.168.177.129
MAL_INST_PORT = 5237
MAL_DW_PORT = 5437
MAL_INST_DW_PORT = 5537
—创建dmmpp.ini
[dmdba@~]# vi /dm8/data/MPP1_02/dmmpp.ini
[service_name1]
mpp_seq_no = 0
mpp_inst_name = MPP1_01
[service_name2]
mpp_seq_no = 1
mpp_inst_name = MPP1_02
—创建dmmpp.ctl
[dmdba@~]# /dm8/bin/dmctlcvt t2c /dm8/data/MPP1_02/dmmpp.ini /dm8/data/MPP1_02/dmmpp.ctl
—创建dmwatcher.ini
[dmdba@~]# vi /dm8/data/MPP1_02/dmwatcher.ini
[GMPP1_01]
DW_TYPE = GLOBAL #全局守护类型
DW_MODE = MANUAL #故障手动切换模式
DW_ERROR_TIME = 20 #远程守护进程故障认定时间
INST_ERROR_TIME = 20 #本地实例故障认定时间
INST_RECOVER_TIME = 60 #主库守护进程启动恢复的间隔时间
INST_OGUID = 45331 #守护系统唯一OGUID值
INST_INI = /dm8/data/MPP1_01/dm.ini #dm.ini文件路径
INST_AUTO_RESTART = 1 #打开实例的自动启动功能
INST_STARTUP_CMD = /dm8/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阈值,默认关闭
RLOG_APPLY_THRESHOLD = 0 #指定备库重演日志的时间阈值,默认关闭
[GMPP1_02]
DW_TYPE = GLOBAL
DW_MODE = MANUAL
DW_ERROR_TIME = 20
INST_ERROR_TIME = 20
INST_RECOVER_TIME = 60
INST_OGUID = 45332
INST_INI = /dm8/data/MPP1_02/dm.ini
INST_AUTO_RESTART = 1
INST_STARTUP_CMD = /dm8/bin/dmserver
RLOG_SEND_THRESHOLD = 0
RLOG_APPLY_THRESHOLD = 0
–拷贝实例到A机器
[dmdba@~]# scp -r /dm8/data/MPP1_02 dmdba@192.168.177.129:/dm8/data/
—注册服务
[root@~]# /dm8/script/root/dm_service_installer.sh -t dmserver -p MPP1_02 -dm_ini /dm8/data/MPP1_02/dm.ini -m mount
[root@~]# /dm8/script/root/dm_service_installer.sh -t dmwatcher -p MPP1_02 -watcher_ini /dm8/data/MPP1_02/dmwatcher.ini
备注:删除自启
[root@~]# /dm8/script/root/dm_service_uninstaller.sh -n DmServiceMPP1_02
[root@~]# /dm8/script/root/dm_service_uninstaller.sh -n DmWatcherServiceMPP1_02
3.配置B机器-MPP1_01B库
----修改dm.ini
[dmdba@~]# vi /dm8/data/MPP1_01/dm.ini
INSTANCE_NAME = MPP1_01B #数据库实例名
PORT_NUM = 5237 #数据库实例监听端口
—替换dmarch.ini
[dmdba@~]# vi /dm8/data/MPP1_01/dmarch.ini
[ARCHIVE_LOCAL]
ARCH_TYPE = LOCAL #本地归档类型
ARCH_DEST = /dm8/data/MPP1_01/arch/ #本地归档存放路径
ARCH_FILE_SIZE = 1024 #单个归档大小,单位MB
ARCH_SPACE_LIMIT = 51200 #归档上限,单位MB
[ARCHIVE_REALTIME]
ARCH_TYPE = REALTIME #实时归档类型
ARCH_DEST = MPP1_01 #实时归档目标实例名
—相同配置项
与A机器MPP1_01的dmmal.ini、dmmpp.ini、dmmpp.ctl相同
—注册服务
[root@~]# /dm8/script/root/dm_service_installer.sh -t dmserver -p MPP1_01B -dm_ini /dm8/data/MPP1_01/dm.ini -m mount
备注:删除自启
[root@~]# /dm8/script/root/dm_service_uninstaller.sh -n DmServiceMPP1_01B
—恢复数据
[dmdba@~]# /dm8/bin/dmrman CTLSTMT="RESTORE DATABASE '/dm8/data/MPP1_01/dm.ini' FROM BACKUPSET '/dm8/data/MPP1_01/bak/BACKUP_FILE'"
[dmdba@~]# /dm8/bin/dmrman CTLSTMT="RECOVER DATABASE '/dm8/data/MPP1_01/dm.ini' FROM BACKUPSET '/dm8/data/MPP1_01/bak/BACKUP_FILE'"
[dmdba@~]# /dm8/bin/dmrman CTLSTMT="RECOVER DATABASE '/dm8/data/MPP1_01/dm.ini' UPDATE DB_MAGIC"
4.配置A机器-MPP1_02B库
— 修改dm.ini
[dmdba@~]# vi /dm8/data/MPP1_02/dm.ini
INSTANCE_NAME = MPP1_02B #数据库实例名
PORT_NUM = 5237 #数据库实例监听端口
—替换dmarch.ini
[dmdba@~]# vi /dm8/data/MPP1_02/dmarch.ini
[ARCHIVE_LOCAL]
ARCH_TYPE = LOCAL #本地归档类型
ARCH_DEST = /dm8/data/MPP1_02/arch/ #本地归档存放路径
ARCH_FILE_SIZE = 1024 #单个归档大小,单位MB
ARCH_SPACE_LIMIT = 51200 #归档上限,单位MB
[ARCHIVE_REALTIME]
ARCH_TYPE = REALTIME #实时归档类型
ARCH_DEST = MPP1_02 #实时归档目标实例名
----配置相同项
与B机器MPP1_02的dmmal.ini、dmmpp.ini、dmmpp.ctl相同
—注册服务
[root@~]# /dm8/script/root/dm_service_installer.sh -t dmserver -p MPP1_02B -dm_ini /dm8/data/MPP1_02/dm.ini -m mount
备注:删除自启
[root@~]# /dm8/script/root/dm_service_uninstaller.sh -n DmServiceMPP1_02B
—恢复数据
[dmdba@~]# /dm8/bin/dmrman CTLSTMT="RESTORE DATABASE '/dm8/data/MPP1_02/dm.ini' FROM BACKUPSET '/dm8/data/MPP1_02/bak/BACKUP_FILE'"
[dmdba@~]# /dm8/bin/dmrman CTLSTMT="RECOVER DATABASE '/dm8/data/MPP1_02/dm.ini' FROM BACKUPSET '/dm8/data/MPP1_02/bak/BACKUP_FILE'"
[dmdba@~]# /dm8/bin/dmrman CTLSTMT="RECOVER DATABASE '/dm8/data/MPP1_02/dm.ini' UPDATE DB_MAGIC"
5.配置监视器
–创建dmmonitor.ini
[dmdba@~]# vi /dm8/bin/dmmonitor.ini
MON_DW_CONFIRM = 0 #0为非确认,1为确认
MON_LOG_PATH = ../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 = 512 #单个日志大小,单位MB
MON_LOG_SPACE_LIMIT = 2048 #日志上限,单位MB
[GMPP1_01]
MON_INST_OGUID = 45331 #组GMPP1_01的唯一OGUID 值
MON_DW_IP = 192.168.177.129:5436 #IP对应MAL_HOST,PORT对应MAL_DW_PORT
MON_DW_IP = 192.168.177.130:5437
[GMPP1_02]
MON_INST_OGUID = 45332 #组GMPP1_02的唯一OGUID 值
MON_DW_IP = 192.168.177.130:5436
MON_DW_IP = 192.168.177.129:5437
—注册服务(选做)
[root@~]# /dm8/script/root/dm_service_installer.sh -t dmmonitor -p Monitor -monitor_ini /dm8/bin/dmmonitor.ini
备注:删除自启
[root@~]# /dm8/script/root/dm_service_uninstaller.sh -n DmMonitorServiceMonitor
5.启动服务及查看信息
----启动数据库并修改参数
-----A机器-MPP1_01
[dmdba@~]# /dm8/bin/DmServiceMPP1_01 start
[dmdba@~]# /dm8/bin/disql SYSDBA/SYSDBA
SQL> SP_SET_OGUID(45331);
SQL> ALTER DATABASE PRIMARY;
----B机器-MPP1_02
[dmdba@~]# /dm8/bin/DmServiceMPP1_02 start
[dmdba@~]# /dm8/bin/disql SYSDBA/SYSDBA
SQL> SP_SET_OGUID(45332);
SQL> ALTER DATABASE PRIMARY;
-----A机器-MPP1_02B
[dmdba@~]# /dm8/bin/DmServiceMPP1_02B start
[dmdba@~]# /dm8/bin/disql SYSDBA/SYSDBA@192.168.177.129:5237
SQL> SP_SET_OGUID(45332);
SQL> SP_SET_PARA_VALUE(1, 'ALTER_MODE_STATUS', 1);
SQL> ALTER DATABASE STANDBY;
SQL> SP_SET_PARA_VALUE(1, 'ALTER_MODE_STATUS', 0);
----B机器-MPP1_01B
[dmdba@~]# /dm8/bin/DmServiceMPP1_01B start
[dmdba@~]# /dm8/bin/disql SYSDBA/SYSDBA@@192.168.177.130:5237
SQL> SP_SET_OGUID(45331);
SQL> SP_SET_PARA_VALUE(1, 'ALTER_MODE_STATUS', 1);
SQL> ALTER DATABASE STANDBY;
SQL> SP_SET_PARA_VALUE(1, 'ALTER_MODE_STATUS', 0);
-----启动守护进程
A机器:
[dmdba@~]# /dm8/bin/DmWatcherServiceMPP1_01 start
B机器:
[dmdba@~]# /dm8/bin/DmWatcherServiceMPP1_02 start
-----启动监视器
[dmdba@~]# /dm8/bin/DmMonitorServiceMonitor start
前台启动:
[dmdba@~]# /dm8/bin/dmmonitor /dm8/bin/dmmonitor.ini
----停止
A机器:[dmdba@~]# /dm8/bin/DmWatcherServiceMPP1_01 stop
B机器:[dmdba@~]# /dm8/bin/DmWatcherServiceMPP1_02 stop
A机器:[dmdba@~]# /dm8/bin/DmServiceMPP1_01 stop
B机器:[dmdba@~]# /dm8/bin/DmServiceMPP1_02 stop
B机器:[dmdba@~]# /dm8/bin/DmServiceMPP1_01B stop
A机器:[dmdba@~]# /dm8/bin/DmServiceMPP1_02B stop
三、配置应用连接
Linux环境:dm_svc.conf放在应用服务器/etc目录下。
Windows32环境:dm_svc.conf放在应用服务器System32目录下。
Windows64环境:dm_svc.conf放在应用服务器System32和SysWOW64目录下。
-----创建dm_svc.conf
[root@~]# vi /etc/dm_svc.conf
TIME_ZONE=(+8:00)
LANGUAGE=(cn)
MPP1=(192.168.177.129:5236,192.168.177.130:5236,192.168.177.129:5237,192.168.177.130:5237)
[MPP1]
LOGIN_MODE=(1)
SWITCH_TIME=(300)
SWITCH_INTERVAL=(200)
------应用连接
<DRIVER>dm.jdbc.driver.DmDriver</DRIVER>
<URL>jdbc:dm://MPP1</URL>
达梦技术社区地址:https://eco.dameng.com