今天现场遇见的一个问题：

运行存储过程时出现如下错误：

ORA-04031: 无法分配 4200 字节的共享内存 ("shared pool"，"insert into p_bel_gprs_pcu(p..."，"library cache"，"kkslpkp - literal info.")

上面看来是共享池问题，现场采取直接重启数据库的方法暂时解决了问题，多次扩大了shared_pool_size大小。

ORA-07445: 出现异常: 核心转储 [0000000101DEADAC] [SIGSEGV] [Address not mapped to object] [0x3301000000000000] [] []
Sat Aug 28 03:04:12 2010
Errors in file /u02/app/oracle/admin/unicom/udump/unicom_ora_6951.trc:
ORA-00600: 内部错误代码，参数: [kghpih:ds], [0x386CA1268], [], [], [], [], [], []
Sat Aug 28 03:04:20 2010
Errors in file /u02/app/oracle/admin/unicom/udump/unicom_ora_10125.trc:
ORA-07445: 出现异常: 核心转储 [0000000101DEADAC] [SIGSEGV] [Address not mapped to object] [0x3401000000000000] [] []
Sat Aug 28 03:06:46 2010
Errors in file /u02/app/oracle/admin/unicom/udump/unicom_ora_10607.trc:
ORA-00600: 内部错误代码，参数: [kghpih:ds], [0x664E0EA68], [], [], [], [], [], []
 There are 14474 memory allocation errors for object-level stat
 in the last 15 minutes

一直忙一个迁移，今天算是对这个问题有进展：

metalink [ID 310939.1]上的信息如下：

Patches
~~~~~~~
  The problem is fixed in the 9.2.0.7 and higher patchsets and patches for 9.2.0.6
  exist for several platforms. 
See Patch 4157713 for a complete list of patches for 9.2.0.6.
References
~~~~~~~~~~
 Note:4157713.8 External description of bug:4157713
看来是个bug需要打patch.