再次和oracle bug相遇 ORA-16163
晚上加班搞一个单机的备机做物理DG，由于之前做过一次相同平台相同版本的rac下的dg，这次开始都还比较顺利，但是在最后打开mrp进程后，归档过不到备库去，随即查看主库日志文件，如下：
Mon May 20 23:36:15 2013 Creating archive destination LOG_ARCHIVE_DEST_2: 'cakmcdb_stdby' LNS0 started with pid=14 Mon May 20 23:37:18 2013 LGWR: Error 16163 creating archivelog file 'cakmcdb_stdby' Mon May 20 23:37:18 2013 Errors in file /opt/oracle/admin/cakmcdb/bdump/cakmcdb_lgwr_26431.trc: ORA-16163: LGWR network server host attach error LGWR: Completed archiving log 2 thread 1 sequence 1380 Thread 1 advanced to log sequence 1380 Current log# 2 seq# 1380 mem# 0: /opt/oracle/oradata/cakmcdb/redo02.log Mon May 20 23:37:18 2013 ARC0: Evaluating archive log 1 thread 1 sequence 1379 ARC0: Beginning to archive log 1 thread 1 sequence 1379 Creating archive destination LOG_ARCHIVE_DEST_1: '/oradata/arch/1_1379.dbf' ARC0: Completed archiving log 1 thread 1 sequence 1379
后来查看metalink
Ora-16163: Lgwr Network Server Host Attach Error [ID 377332.1]
Applies to:
Oracle Server - Enterprise Edition - Version: 10.2.0.2 and later [Release: 10.2 and later ]
Information in this document applies to any platform.
Symptoms
The physical standby was working fine and was in constant sync with the primary. After
enabling the Fast-Start Failover using Grid Control, we started getting these errors below.
ORA-16163: LGWR network server host attach error
Cause
Bug 5220576 AFTER CHANGING PROTECTION MODE LGWR: ERROR 16163 CREATING ARCHIVELOG
During EM grid control FSFO enable, net_timeout is set to 1 if the standby needs to be shutdown.
It is set back to the original value immediately after the FSFO enable completes. The bug is that
the original value is not restored if the FSFO enable operation fails, leaving net_timeout=1 and
the corresponding ORA-16163 error.
Solution
This bug is fixed in the 10.2.0.3 patchset.
The workaround is to use DGMGRL to set net_timeout back to the original value.
DGMGRL> edit database '
我很纳闷，同样的数据库版本，就是一个是rac一个是单节点，一个是sunos 5.10 一个是5.8 怎么就会触发bug呢，再说了我也没有用broker管理dg呀，再者也把net_timeout=180了还是不行呀...
郁闷....看来以后自己就少碰9i的库了....
后来开了SR问了下oracle，他们说是将主库参数 process从150调大到200尝试解决该问题，由于这个系统今年的停机时间已用，所以暂时就取消DG的配置。