oracle迁移数据库后启动报错ora-600[25025]解决办法

迁移脚本的日志中报错RMAN-06571: datafile 78 does not have recoverable copy，经查看发现78号文件曾经被offline drop掉。于是重建控制文件，在控制文件中把78号文件去掉，重建控制后，数据库能够mount，mount后数据文件是一致，但是open 时会报错ora-600，异常宕掉。

SYS@mydbtst> alter database open;
alter database open
*
ERROR at line 1:
ORA-01092: ORACLE instance terminated. Disconnection forced
ORA-00600: internal error code, arguments: [25025], [78], [], [], [], [], [],
[], [], [], [], []
Process ID: 6028
Session ID: 521 Serial number: 15

经检查，这个78号文件是undo的一个文件。上述的处理方法，对于处理一般的数据文件是可行的，但是对于undo文件的问题，就不能用上述这个方法处理了。应该用下面的处理方法，具体的处理方法如下:

1. 加上管理方式为manual：

……
*.streams_pool_size=134217728
*.undo_retention=3600
*.undo_tablespace='UNDOTBS1'
*.undo_management='manual'  
~
2. 启动之后，做recovery：

SYS@mydbtst> startup pfile='/tmp/pfile_bak.ora';
ORACLE instance started.
 
Total System Global Area 2522189824 bytes
Fixed Size                  2230912 bytes
Variable Size            1157629312 bytes
Database Buffers         1342177280 bytes
Redo Buffers               20152320 bytes
Database mounted.
ORA-01113: file 1 needs media recovery
ORA-01110: data file 1: '+DG_DATA001/mydbtst/datafile/system.453.813666469'
 
 
 
 
SYS@mydbtst> recover database using backup controlfile;
ORA-00279: change 9436796441761 generated at 08/01/2014 10:07:10 needed for
thread 1
ORA-00289: suggestion : +FRA_SMALL_MDG
ORA-15173: entry 'ARCHIVELOG' does not exist in directory 'mydbtst'
ORA-00280: change 9436796441761 for thread 1 is in sequence #4 <<<<<<<需要sequence # 4的日志，查v$log和v$logfile之后，确认哪个redo log放入
 
 
Specify log: {=suggested | filename | AUTO | CANCEL}
+DATA_SMALL_MDG/mydbtst/onlinelog/group_1.1156.854401821
Log applied.
Media recovery complete.
SYS@mydbtst>
SYS@mydbtst> alter database open;
alter database open
*
ERROR at line 1:
ORA-01589: must use RESETLOGS or NORESETLOGS option for database open
 
 
SYS@mydbtst> alter database open resetlogs;
 
Database altered.
 
SYS@mydbtst>
SYS@mydbtst>

3. 新建新的undo

SYS@mydbtst> create undo tablespace undo_new datafile '+DG_DATA001/mydbtst/datafile/undo_new01.dbf' size 200m;
 
Tablespace created.
 
SYS@mydbtst>

4. 检查UNDOTBS1的undo segment，幸运的是，我没有发现needs recovery的undo segment，所以后续也不用隐含参数”_corrupted_rollback_segments”跳过need recovery的undo segment来重启了。

SYS@mydbtst> select tablespace_name,segment_name,status from dba_rollback_segs;
 
TABLESPACE_NAME                SEGMENT_NAME                   STATUS
------------------------------ ------------------------------ ----------------
SYSTEM                         SYSTEM                         ONLINE
UNDOTBS1                       _SYSSMU1_3981220827$           OFFLINE
UNDOTBS1                       _SYSSMU2_2541240231$           OFFLINE
UNDOTBS1                       _SYSSMU3_4103266798$           OFFLINE
UNDOTBS1                       _SYSSMU4_1110676785$           OFFLINE
UNDOTBS1                       _SYSSMU5_3829116805$           OFFLINE
UNDOTBS1                       _SYSSMU6_347720470$            OFFLINE
UNDOTBS1                       _SYSSMU7_3507999319$           OFFLINE
UNDOTBS1                       _SYSSMU8_3681584916$           OFFLINE
UNDOTBS1                       _SYSSMU9_630455542$            OFFLINE
UNDOTBS1                       _SYSSMU10_2221096320$          OFFLINE
 
……
 
TABLESPACE_NAME                SEGMENT_NAME                   STATUS
------------------------------ ------------------------------ ----------------
UNDO_NEW                       _SYSSMU110_852046607$          OFFLINE
UNDO_NEW                       _SYSSMU111_1868020771$         OFFLINE
 
112 rows selected.
 
SYS@mydbtst>

5. 由于不需要隐含参数跳过，可以直接drop原来的undo：

SYS@mydbtst> drop tablespace UNDOTBS1 including contents and datafiles;
 
Tablespace dropped.
6. 修改pfile为auto和新的undo_new

cnsz181007:mydbtst > vi pfile_bak.ora
 
……
 
*.streams_pool_size=134217728
*.undo_retention=3600
*.undo_tablespace='undo_new'
*.undo_management='auto'
~
7. 用该pfile重启：

cnsz181007:mydbtst > sqlplus "/ as sysdba"
 
SQL*Plus: Release 11.2.0.3.0 Production on Fri Aug 1 11:59:56 2014
 
Copyright (c) 1982, 2011, Oracle.  All rights reserved.
 
Connected to an idle instance.
 
SYS@mydbtst> startup pfile='/tmp/pfile_bak.ora';
ORACLE instance started.
 
Total System Global Area 2522189824 bytes
Fixed Size                  2230912 bytes
Variable Size            1157629312 bytes
Database Buffers         1342177280 bytes
Redo Buffers               20152320 bytes
Database mounted.
Database opened.
SYS@mydbtst>