在逻辑STANDBY上碰到这个错误。

启动逻辑STANDBY：

SQL> alter database start logical standby apply initial 5624048054;

Database altered.

SQL> alter database register logical logfile '/data1/archivelog/1_71917_649168938.dbf';

Database altered.

SQL> alter database register logical logfile '/data1/archivelog/1_71918_649168938.dbf';

Database altered.

.
.
.

这时检查数据库的alert文件发现错误：

Thu Jan 20 13:59:54 2011
LOGMINER: Begin mining logfile: /data1/archivelog/1_72016_649168938.dbf
Thu Jan 20 13:59:54 2011
LOGMINER: Turning ON Log Auto Delete
Thu Jan 20 13:59:54 2011
Errors in file /opt/oracle/admin/mobiledb/bdump/mobiledb_p001_23044.trc:
ORA-00600: internal error code, arguments: [krvxbpx20], [1], [72016], [695], [100], [], [], []
LOGSTDBY Analyzer process P003 started with pid=31 OS id=23057
LOGSTDBY Apply process P005 started with pid=33 OS id=23061
LOGSTDBY Apply process P004 started with pid=32 OS id=23059
LOGSTDBY Apply process P006 started with pid=34 OS id=23063
LOGSTDBY Apply process P008 started with pid=36 OS id=23067
LOGSTDBY Apply process P007 started with pid=35 OS id=23065
Thu Jan 20 13:59:55 2011
LOGSTDBY status: ORA-16211: unsupported record found in the archived redo log
LOGSTDBY Apply process P005 pid=33 OS id=23061 stopped
Thu Jan 20 13:59:56 2011
Errors in file /opt/oracle/admin/mobiledb/bdump/mobiledb_lsp0_23040.trc:
ORA-12801: error signaled in parallel query server P005
ORA-16211: unsupported record found in the archived redo log
LOGSTDBY Analyzer process P003 pid=31 OS id=23057 stopped
LOGSTDBY Apply process P004 pid=32 OS id=23059 stopped
LOGSTDBY Apply process P008 pid=36 OS id=23067 stopped
LOGSTDBY Apply process P007 pid=35 OS id=23065 stopped
LOGSTDBY Apply process P006 pid=34 OS id=23063 stopped
Thu Jan 20 13:59:56 2011
krvxerpt: Errors detected in process 28, role builder.
Thu Jan 20 13:59:56 2011
krvxmrs: Leaving by exception: 600
Thu Jan 20 13:59:56 2011
Errors in file /opt/oracle/admin/mobiledb/bdump/mobiledb_p001_23044.trc:
ORA-00600: internal error code, arguments: [krvxbpx20], [1], [72016], [695], [100], [], [], []
LOGSTDBY status: ORA-00600: internal error code, arguments: [krvxbpx20], [1], [72016], [695], [100], [], [], []

在DBA_LOGSTDBY_EVENTS视图中同样可以看到这个错误，另外还可以看到ORA-16211: unsupported record found in the archived redo log的信息，报错的对象是"SYS"."SEQ$": unsupported。这应该是SYS中的一个序列。

查询metalink，文档ID 761661.1描述了这个问题，主要导致的原因是主库的SUPPLEMENTAL_LOG_DATA_MIN、SUPPLEMENTAL_LOG_DATA_PK和SUPPLEMENTAL_LOG_DATA_UI没有都设置为TRUE。从而导致一些IMU（In Memory Undo）出现在redo记录中，使得逻辑STANDBY无法应用日志。

检查了主库的设置：

SQL> select name, supplemental_log_data_min, supplemental_log_data_pk, supplemental_log_data_ui
  2  from v$database;

NAME      SUPPLEME SUP SUP
--------- -------- --- ---
MOBILEDB  IMPLICIT YES YES

SUPPLEMENTAL_LOG_DATA_MIN的值不是YES而是IMPLICIT，看来多半就是这个值导致了问题的产生。

解决问题的方法很痛苦，将主库的三个参数设置正确后，重建逻辑STANDBY数据库。