问题描述
开发同事让在一个服务器上查找下先前库的备份文件是否存在，如果存在进行下恢复。翻了服务器发现备份文件存在，多愁了一眼竟翻到了该备份文件于2024.6.17日恢复过的日志，赶紧和开发沟通说2024.6.17号已经恢复过了为啥还要恢复，开发这才说出关键：前几天清理磁盘，不小心将log文件删除了现在库不能启动，咨询了另一个DBA说是得重新恢复。可见沟通很关键啊，沟通后可不敢直接开干，多问问多翻翻多核实。
分析过程
查看服务器上备份文件
说明：db_20240617_TOPICIS_5238.dmp是由其他服务器上备份后上传到该服务器上并进行恢复的，db_20240617_TOPICIS_5238_imp.log是恢复日志。
[dmdba@db191 dmbak]$ ls -lhtr TOPICIS
-rw------- 1 dmdba dinstall 91G 6月 17 23:06 db_20240617_TOPICIS_5238.dmp
-rw------- 1 dmdba dinstall 1.8M 6月 17 23:06 db_20240617_TOPICIS_5238_exp.log
-rw-r--r-- 1 dmdba dinstall 3.1M 6月 18 04:37 db_20240617_TOPICIS_5238_imp.log
启动实例
查看配置文件
--查找配置文件
[root@db191 dmbak]# find / -name dm.ini
/opt/dmdb/topicis/topicis/dm.ini
/opt/dmdb/dzzzk/dzzzk/dm.ini
/opt/dmdb/DMSERVER/DAMENG/dm.ini
/opt/dmdb/xzspkp5555/xzspkp/dm.ini
--查看配置文件内容
[root@db191 dmbak]# cat /opt/dmdb/topicis/topicis/dm.ini | grep PORT
PORT_NUM = 5238 #Port number on which the database server will listen
ELOG_REPORT_LINK_SQL = 0 #Whether to write the SQLs that sent to remote database by DBLINKs into error log file
DFS_HUGE_SUPPORT = 1 #Whether support huge table operation in DDFS
EXTERNAL_JFUN_PORT = 6363 #DmAgent port for external java fun.
EXTERNAL_AP_PORT = 4236 #DmAp port for external fun.
DCRS_PORT_NUM = 6236 #Port number on which the database dcrs will listen
AP_PORT_NUM = 0 #Port number on which the database ap will listen
DW_PORT = 0 #Instance tcp port for watch2
HA_INST_CHECK_PORT = 65534 #HA instance check port
查看实例服务
db_20240617_TOPICIS_5238.dmp对应的是DmServiceTOPICIS.service
[root@db191 dmbak]# systemctl list-units D*
UNIT LOAD ACTIVE SUB DESCRIPTION
DmAPService.service loaded active running DM Assistant Plug-In Service(DmAPService).
DmServiceDMSERVER.service loaded active running DM Instance Service
● DmServicedzzzk.service loaded failed failed DM Instance Service
● DmServiceTOPICIS.service loaded failed failed DM Instance Service
DmServicexzspkp.service loaded active running DM Instance Service
启动实例
[root@db191 dmbak]# systemctl start DmServiceTOPICIS
Job for DmServiceTOPICIS.service failed because the control process exited with error code. See "systemctl status DmServiceTOPICIS.service" and "journalctl-xe" for details.
启动失败
查看启动失败日志
日志位置：安装目录/log下
cd /opt/dmdbms/log
tail -300f dm_TOPICIS_202410.log
输出如下：
2024-10-16 13:43:30.587 [INFO] database P0000001716 T0
000000000000001716 DM Database Server 64 V8 03134284044-20230703-194450-20040 startup...
2024-10-16 13:43:31.840 [INFO] database P0000001716 T0000000000000001716 fil_sys_init
2024-10-16 13:43:31.990 [INFO] database P0000001716 T0000000000000001716 Database mode = 0, oguid = 0
2024-10-16 13:43:31.991 [FATAL] database P0000001716 T0000000000000001716 /opt/dmdb/topicis/topicis/topicis01.log not exist,can not startup
2024-10-17 09:07:28.861 [INFO] database P0000020309 T0000000000000020309 INI parameter DPC_2PC changed, the original value 1, new value 0
2024-10-17 09:07:28.866 [INFO] database P0000020309 T0000000000000020309 version info: develop
2024-10-17 09:07:28.868 [INFO] database P0000020309 T0000000000000020309 os_sema2_create_low, create and inc sema success, key:125441262, sem_id:2, sem_value:1!
2024-10-17 09:07:28.874 [INFO] database P0000020309 T0000000000000020309 ini_set_secini, ini parameter PWD_POLICY changed, the original value 2, new value0
2024-10-17 09:07:29.650 [INFO] database P0000020309 T0000000000000020309 DM Database Server 64 V8 03134284044-20230703-194450-20040 startup...
2024-10-17 09:07:29.924 [INFO] database P0000020309 T0000000000000020309 fil_sys_init
2024-10-17 09:07:30.075 [INFO] database P0000020309 T0000000000000020309 Database mode = 0, oguid = 0
2024-10-17 09:07:30.075 [FATAL] database P0000020309 T0000000000000020309 /opt/dmdb/topicis/topicis/topicis01.log not exist,can not startup
解决办法
查看初始化参数
着重关注的参数有：
page size：单位字节 而初始化参数界面单位是K，注意转换
extent size：单位页
case sensitive：取值：Y、y、1 表示敏感；N、n、0 表示不敏感。默认值为 Y
charset：0 代表 GB18030，1 代表 UTF-8，2 代表韩文字符集 EUC-KR。默认为 0
length in char：VARCHAR 类型对象的长度是否以字符为单位。取值为 1 或者 Y 表示是，取值为 0 或 N 表示否。默认值为 0
cd /opt/dmdb/topicis/topicis
ls -l dminit*
输出如下：
[dmdba@db191 topicis]$ ls -l dminit*
-rw-r--r-- 1 dmdba dinstall 1009 6月 17 22:09 dminit20240617220946.log
[dmdba@db191 topicis]$ more dminit20240617220946.log
start init database: V8, 2024-06-17 22:09:46
init params:
db path: /opt/dmdb/topicis/topicis
db name: topicis
auto overwrite: 0
page size: 8192
extent size: 16
char_fix_storage: 0
sql_log_forbid: 0
secur_flag: 2
time zone: +08:00
string case sensitive: 1
charset: 1
length in char: 0
page check mode: 0
page check algorithm id: 0
priv flag: 0
env label: 0
rlog enc flag: 0
use new hash: 1
blank pad mode: 0
sec priv mode: 0
huge with delta: 1
rlog gen for huge: 1
pseg_mgr_flag: 0
log file path: /opt/dmdb/topicis/topicis/topicis01.log
log file path: /opt/dmdb/topicis/topicis/topicis02.log
create ini file /opt/dmdb/topicis/topicis/dm.ini success.
create rlog file /opt/dmdb/topicis/topicis/topicis01.log success.
create rlog file /opt/dmdb/topicis/topicis/topicis02.log success.
SYSTEM file : /opt/dmdb/topicis/topicis/SYSTEM.DBF
MAIN file : /opt/dmdb/topicis/topicis/MAIN.DBF
ROLL file : /opt/dmdb/topicis/topicis/ROLL.DBF
create dm database success. 2024-06-17 22:09:50
移除系统服务(重要)
systemctl disable DmServiceTOPICIS
cd /opt/dmdbms/bin
ls -l DmServiceTOPICIS
mv DmServiceTOPICIS DmServiceTOPICIS_bak_date +%F
，不然后面初始化实例提示如下报错：
更改故障实例的端口(重要)
将/opt/dmdb/topicis/topicis/dm.ini中的PORT_NUM更改为未占用的端口，本文档更改为9999，不然后面初始化实例提示如下报错：
创建新的数据目录并赋予权限
mkdir -p /opt/dmdb/topicis/topicisnew
chown -R dmdba:dinstall /opt/dmdb/topicis/topicisnew
初始化新实例
采用的是图形化，避免参数错。
运行Xmanager Power Suite 7下tools文件夹下的Xmanager - Passive工具
su - dmdba
export DISPLAY=192.168.19.74:0.0
cd /opt/dmdbms/tool
./dbca.sh
执行以下脚本
su - root
mv /opt/dmdbms/bin/DmServicetopicis.service /usr/lib/systemd/system/DmServicetopicis.service
systemctl enable DmServicetopicis.service
systemctl start DmServicetopicis.service
创建业务用户
create user topicis identified by "123456789"
limit failed_login_attemps unlimited, password_lock_time unlimited, password_grace_time unlimited;
grant "DBA" to topicis;
逻辑恢复
su - dmdba
nohup /opt/dmdbms/bin/dimp USERID=topicis/123456789:5238 FILE=db_20240617_TOPICIS_5238.dmp DIRECTORY=/opt/dmbak LOG=db_20241017_TOPICIS_5238_imp.log LOG_WRITE=Y PARALLEL=28 &
参考链接：达梦DM8之使用dmmdf恢复误删除的undo/redo文件_file dm.key not found, use default license!-CSDN博客
