墨墨导读：客户的一套开发环境，清理空间时redo被运维人员当作log误删除，一线同事先接手处理，过程中遇到问题升级本文做出分析。

接手后，数据库处于mount状态，之前恢复过程中已经做过resetlogs的操作，也设置了"_allow_resetlogs_corruption"隐藏参数为true，目前直接开库会提示需要恢复，重新进行resetlogs时报错ORA-600 [2662]，起初看到这个错误心中略有些放松，根据经验，推下SCN就好了：

SQL> alter database open;alter database open*ERROR at line 1:ORA-01113: file 1 needs media recoveryORA-01110: data file 1:'/oradata/ODSDB/datafile/data_D-ODSDB_I-3532120983_TS-SYSTEM_FNO-1_1bujd4ob.dbf'
SQL> recover database ;ORA-00283: recovery session canceled due to errorsORA-00399: corrupt change description in redo logORA-00353: log corruption near block 3 change 32581427074 time 08/25/202011:39:04ORA-00312: online log 5 thread 1: '/oradata/ODSDB/datafile/redo05.log'SQL> recover datafile 1;ORA-00283: recovery session canceled due to errorsORA-00399: corrupt change description in redo logORA-00353: log corruption near block 3 change 32581427074 time 08/25/202011:39:04ORA-00312: online log 5 thread 1: '/oradata/ODSDB/datafile/redo05.log'
SQL> alter database open resetlogs;alter database open resetlogs*ERROR at line 1:ORA-01139: RESETLOGS option only valid after an incomplete database recoverySQL> recover database using backup controlfile until cancel;ORA-00279: change 32581427072 generated at  needed for thread 1Specify log: {<RET>=suggested | filename | AUTO | CANCEL}autoORA-00266: name of archived log file neededORA-01547: warning: RECOVER succeeded but OPEN RESETLOGS would get error belowORA-01194: file 1 needs more recovery to be consistentORA-01110: data file 1:'/oradata/ODSDB/datafile/data_D-ODSDB_I-3532120983_TS-SYSTEM_FNO-1_1bujd4ob.dbf'SQL>  alter database open resetlogs; alter database open resetlogs*ERROR at line 1:ORA-00603: ORACLE server session terminated by fatal errorORA-00600: internal error code, arguments: [2662], [7], [2516656013], [7],[2516660859], [327174016], [], [], [], [], [], []ORA-00600: internal error code, arguments: [2662], [7], [2516656012], [7],[2516660859], [327174016], [], [], [], [], [], []ORA-01092: ORACLE instance terminated. Disconnection forcedORA-00600: internal error code, arguments: [2662], [7], [2516656008], [7],[2516660859], [327174016], [], [], [], [], [], []Process ID: 3997864Session ID: 580 Serial number: 5

重新启动到mount状态，根据上面报错计算要推进的SCN值：

SQL> select open_mode from v$database;OPEN_MODE
--------------------
MOUNTED
SQL> select current_scn from v$database;
SQL> set num 30
30217546463
select 7*power(2,32) + 2516660859 from dual;select 7*power(2,32) + 2516661000 from dual;
32581432072

使用oradebug poke进行SCN推进：

oradebug setmypid
oradebug dumpvar sga kcsgscn_
oradebug poke 0x700000000019B70 8 32581432072
oradebug dumpvar sga kcsgscn_

推进SCN后，尝试直接开库报错ORA-00399、ORA-00353、ORA-00312：

SQL> oradebug setmypidStatement processed.SQL> oradebug dumpvar sga kcsgscn_kcslf kcsgscn_ [700000000019B70, 700000000019BA0) = 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 07000000 00019850SQL> oradebug poke 0x700000000019B70 8 32581432072BEFORE: [700000000019B70, 700000000019B78) = 00000000 00000000AFTER:  [700000000019B70, 700000000019B78) = 00000007 96013308SQL> oradebug dumpvar sga kcsgscn_kcslf kcsgscn_ [700000000019B70, 700000000019BA0) = 00000007 96013308 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 07000000 00019850SQL> alter database open;alter database open*ERROR at line 1:ORA-00399: corrupt change description in redo logORA-00353: log corruption near block 3 change 32581427074 time 08/25/202011:39:04ORA-00312: online log 5 thread 1: '/oradata/ODSDB/datafile/redo05.log'

看起来有些麻烦，根据报错推断之前的resetlogs其实都没有完全成功。
此时查询各个checkpoint_change#都是一致的，正常应该可以open成功：

select distinct checkpoint_change# from v$database;select distinct checkpoint_change# from v$datafile;select distinct checkpoint_change# from v$datafile_header;

可事实上，即使尝试read only开库都会报错ORA-16433：

SQL> alter database open read only;alter database open read only*ERROR at line 1:ORA-16433: The database must be opened in read/write mode.

查了下这个错误，看起来控制文件有些问题，于是想重建下控制文件，结果发现命令同样报错ORA-16433：

--recreate controlfilealter database backup controlfile to trace;SQL> alter database backup controlfile to trace;alter database backup controlfile to trace*ERROR at line 1:ORA-16433: The database must be opened in read/write mode.

而正常mount下，是可以执行此命令的。

没办法，只能根据控制文件的语法，查询必要信息，然后手工构建控制文件的创建脚本：

--dbnameODSDB--select userenv('language') from dual;AMERICAN_AMERICA.AL32UTF8--nomountSTARTUP NOMOUNT--create_controlfile.sqlCREATE CONTROLFILE REUSE DATABASE "ODSDB" NORESETLOGS  NOARCHIVELOG    MAXLOGFILES 16    MAXLOGMEMBERS 3    MAXDATAFILES 1024    MAXINSTANCES 8    MAXLOGHISTORY 292LOGFILE  GROUP 1 '/oradata/ODSDB/datafile/redo01.log' size 1024M blocksize 512,  GROUP 2 '/oradata/ODSDB/datafile/redo02.log' size 1024M blocksize 512,  GROUP 3 '/oradata/ODSDB/datafile/redo03.log' size 1024M blocksize 512,  GROUP 4 '/oradata/ODSDB/datafile/redo04.log' size 1024M blocksize 512,  GROUP 5 '/oradata/ODSDB/datafile/redo05.log' size 1024M blocksize 512,  GROUP 6 '/oradata/ODSDB/datafile/redo06.log' size 1024M blocksize 512-- STANDBY LOGFILEDATAFILE  '/oradata/ODSDB/datafile/data_D-ODSDB_I-3532120983_TS-SYSTEM_FNO-1_1bujd4ob.dbf',  '/oradata/ODSDB/datafile/data_D-ODSDB_I-3532120983_TS-SYSAUX_FNO-2_1aujd4oa.dbf',  '/oradata/ODSDB/datafile/data_D-ODSDB_I-3532120983_TS-ODS_FDM_P11_FNO-3_0fujd0al.dbf',  '/oradata/ODSDB/datafile/data_D-ODSDB_I-3532120983_TS-USERS_FNO-4_30ujd66a.dbf',..此处省略大量数据文件输出..  '/oradata/ODSDB/datafile/data_D-ODSDB_I-3532120983_TS-ODS_FDM_FNO-89_08ujcris1.dbf',  '/oradata/ODSDB/datafile/data_D-ODSDB_I-3532120983_TS-ODS_FDM_FNO-00_08ujcg7n.dbf'CHARACTER SET AL32UTF8;

重建控制文件后,再次推SCN，最终恢复成功：

select 7*power(2,32)+2616661492 from dual;32681432564
--recover again..conn /as sysdbastartup nomount;@create_controlfile.sqloradebug setmypidoradebug dumpvar sga kcsgscn_oradebug poke 0x700000000019B70 8 32681432564oradebug dumpvar sga kcsgscn_
recover database using backup controlfile until cancel;autoalter database open resetlogs;

最后将隐藏参数"_allow_resetlogs_corruption"设置为false，正常重启数据库正常：

alter system set "_allow_resetlogs_corruption" = false scope=spfile;shut immediatestartup

总结：主要还是推进SCN，只是这次遇到报错ORA-16433需要重建控制文件。

墨天轮原文链接：https://www.modb.pro/db/31928（复制到浏览器中打开或者点击“阅读原文”）

推荐阅读：144页！分享珍藏已久的数据库技术年刊

视频号，新的分享时代，关注我们，看看有什么新发现？

数据和云

ID：OraNews

如有收获，请划至底部，点击“在看”，谢谢！

点击下图查看更多 ↓

云和恩墨大讲堂 | 一个分享交流的地方

长按，识别二维码，加入万人交流社群

请备注：云和恩墨大讲堂

  点个“在看”

你的喜欢会被看到❤