ORA-00600: internal error code, arguments: [2037]

今天一家用户意外断电，导致数据库无法启动，启动的时候如下提示：

ksdpec: called for event 13740 prior to event group initialization
Starting up ORACLE RDBMS Version: 10.2.0.1.0.
System parameters with non-default values:
processes = 150
sga_target = 612368384
control_files = F:\ORADATA\ORCL\CONTROL01.CTL, F:\ORADATA\ORCL\CONTROL02.CTL, F:\ORADATA\ORCL\CONTROL03.CTL
db_block_size = 8192
compatible = 10.2.0.1.0
db_file_multiblock_read_count= 16
db_recovery_file_dest = d:\oracle\product\10.2.0/flash_recovery_area
db_recovery_file_dest_size= 2147483648
undo_management = AUTO
undo_tablespace = UNDOTBS1
remote_login_passwordfile= EXCLUSIVE
db_domain =
dispatchers = (PROTOCOL=TCP) (SERVICE=orclXDB)
job_queue_processes = 10
audit_file_dest = D:\ORACLE\PRODUCT\10.2.0\ADMIN\ORCL\ADUMP
background_dump_dest = D:\ORACLE\PRODUCT\10.2.0\ADMIN\ORCL\BDUMP
user_dump_dest = D:\ORACLE\PRODUCT\10.2.0\ADMIN\ORCL\UDUMP
core_dump_dest = D:\ORACLE\PRODUCT\10.2.0\ADMIN\ORCL\CDUMP
db_name = orcl
open_cursors = 300
pga_aggregate_target = 203423744
PMON started with pid=2, OS id=3476
PSP0 started with pid=4, OS id=1568
MMAN started with pid=6, OS id=1564
DBW0 started with pid=8, OS id=3460
DBW1 started with pid=10, OS id=3464
LGWR started with pid=12, OS id=1596
CKPT started with pid=14, OS id=4052
SMON started with pid=16, OS id=1572
RECO started with pid=18, OS id=3668
CJQ0 started with pid=20, OS id=4008
MMON started with pid=22, OS id=3992
MMNL started with pid=24, OS id=3984
Tue Dec 17 17:32:08 2013
starting up 1 dispatcher(s) for network address '(ADDRESS=(PARTIAL=YES)(PROTOCOL=TCP))'...
starting up 1 shared server(s) ...
Tue Dec 17 17:32:09 2013
ALTER DATABASE MOUNT
Tue Dec 17 17:32:14 2013
Setting recovery target incarnation to 2
Tue Dec 17 17:32:14 2013
Successful mount of redo thread 1, with mount id 1362246041
Tue Dec 17 17:32:14 2013
Database mounted in Exclusive Mode
Completed: ALTER DATABASE MOUNT
Tue Dec 17 17:32:14 2013
ALTER DATABASE OPEN
Tue Dec 17 17:32:15 2013
Beginning crash recovery of 1 threads
parallel recovery started with 15 processes
Tue Dec 17 17:32:15 2013
Started redo scan
Tue Dec 17 17:32:15 2013
Completed redo scan
65573 redo blocks read, 497 data blocks need recovery
Tue Dec 17 17:32:16 2013
Started redo application at
Thread 1: logseq 32195, block 2, scn 708440896
Tue Dec 17 17:32:16 2013
Recovery of Online Redo Log: Thread 1 Group 1 Seq 32195 Reading mem 0
Mem# 0 errs 0: F:\ORADATA\ORCL\REDO01.LOG
Tue Dec 17 17:32:16 2013
Errors in file d:\oracle\product\10.2.0\admin\orcl\bdump\orcl_p008_1112.trc:
ORA-00600: internal error code, arguments: [2037], [33735871], [3300893190], [6], [24], [0], [684246207], [100858044]

Tue Dec 17 17:32:16 2013
Errors in file d:\oracle\product\10.2.0\admin\orcl\bdump\orcl_p011_4056.trc:
ORA-00600: internal error code, arguments: [2037], [67559644], [3772555782], [6], [1], [0], [688447708], [100731143]

Tue Dec 17 17:32:16 2013
Errors in file d:\oracle\product\10.2.0\admin\orcl\bdump\orcl_p013_2884.trc:
ORA-00600: internal error code, arguments: [2037], [8391083], [162243074], [6], [9], [0], [685705643], [34694206]

Tue Dec 17 17:32:16 2013
Errors in file d:\oracle\product\10.2.0\admin\orcl\bdump\orcl_p005_416.trc:
ORA-00600: internal error code, arguments: [2037], [12645188], [4081361414], [6], [2], [0], [685503300], [100728832]

这么多错误提示，重点关注ORA-00600 [2037],Metalink上有篇文章，虽然提示有所出入，但是大体原因如下：

CAUSE

The problem is caused by Bug 4899479 Undo/redo corruption if distributed transactions used.

Details:
     Redo/undo corruption and associated dumps and/or ORA-600 errors/memory corruption can
     occur from a transaction which uses "in memory undo" pool memory if the transaction also
     includes distributed operations. e.g.: Insert using buffered inserts, distributed operation,
     subsequent insert can lead to this problem if the first insert uses in-memory undo.

SOLUTION

To resolve the corruption, you will need to do a point in time recovery from a good backup.

To prevent the bug from occurring, you may choose either of these options:

1. Apply the 10.2.0.4 patchset which was not available at the time of this writing (Feb 2008).
2. Check MetaLink for backport Patch 4899479 for your operating system and database release.

可以看到，除了BUG原因，一般情况下是redo或者undo损坏，我们这里显然不是BUG,根据在报错之前有提示

Recovery of Online Redo Log: Thread 1 Group 1 Seq 32195 Reading mem 0
Mem# 0 errs 0: F:\ORADATA\ORCL\REDO01.LOG
我判断是redo文件损坏，于是按照一般流程处理redo损坏的恢复方式尝试启动数据库，结果数据库启动成功，运气不错，看来oracle同样的错误，有时间错误提示会给人一些误导。