FLASHAREA_DG 磁盘组 不断的dismount 和mount

有一个小的生产库，操作系统版本：Red Hat Enterprise Linux Server release 5.5 (Tikanga)

oracle 是  10g Enterprise Edition Release 10.2.0.5.0 - 64bit   rac  的

最近发现他的alter 日志里面老是提示如下信息：

SUCCESS: diskgroup FLASHAREA_DG was mounted
Mon May 06 18:44:23 CST 2013
SUCCESS: diskgroup FLASHAREA_DG was dismounted
Mon May 06 19:31:45 CST 2013
Thread 2 advanced to log sequence 9642 (LGWR switch)
  Current log# 6 seq# 9642 mem# 0: +DATADG/gpsdb/redo06.log
Mon May 06 19:31:45 CST 2013
SUCCESS: diskgroup FLASHAREA_DG was mounted
Mon May 06 19:32:56 CST 2013
SUCCESS: diskgroup FLASHAREA_DG was dismounted
Mon May 06 20:20:53 CST 2013
Thread 2 advanced to log sequence 9643 (LGWR switch)
  Current log# 7 seq# 9643 mem# 0: +DATADG/gpsdb/redo07.log
Mon May 06 20:20:53 CST 2013
SUCCESS: diskgroup FLASHAREA_DG was mounted
Mon May 06 20:22:01 CST 2013
SUCCESS: diskgroup FLASHAREA_DG was dismounted
Mon May 06 21:06:21 CST 2013

-------

但是数据库还是能正常使用。找了一下相关资料，发现原来是一个BUG ，但是不影响数据库的使用。

This is intended behavior.

From bug:4465015 Asm Diskgroup For Flash Recovery Area Display Mount And Dismount In Db Alert Log (Not published)

This is expected behavior. The messages you see are not errors (that is why they are prefixed with "SUCCESS"). The database instance is creating and then closing archive logs one at a time in the LOG disk group.
Whenever a database closes its last file in a disk group, it dismounts the disk group.
Whenever a database accesses a disk group when it does not have any other files open in the disk group, it mounts the disk group.

If a diskgroup is used only for flashback feature and rman backups in a database, many log messages are seen in alert.log. However, the same may also occur for diskgroups keeping datafiles but in less frequency.

Solution:
This is intended behavior. No solution is required. If you do not want to see the log messages frequently, you can put a mirrored control-file, or mirrored online redo, or dummy online tablespace data file on the diskgroup. This message will not be output frequently in this case. Because CKPT keeps opening a control-file, LGWR keeps opening an online redo, DBW keeps opening an online tablespace data file. So, the diskgroup is not dismounted until the database shutdown.

 

 

这是因为一个bug：4465015导致的。不过Oracle也说这是一个正常的行为：
当关闭某个disk group上的最后一个文件时，Oracle会将这个diskgroup dismount；
当需要读取某个disk group某一个文件时，Oracle再把它mount。

在这个案例中，归档日志放在一个独立的disk group中，而归档日志是隔段时间才会产生一个，所以会频繁在alert日志中看到这个信息。
这个信息是正常的，但是如果你觉得比较烦，可以把控制文件、redo的一个副本或者放一个很小的数据文件到这个disk group上，依赖CKPT、LGWR、DWR几个进程分别需要频繁访问以上文件来避免。

----------

好吧！既然没有不良影响 ，就暂时不动吧