Database Media Recovery Failure

DBA Notes: 2011/12/27

Cheng Li

(Database: Oracle 9i, OS: AIX)

 

Database Media Recovery Failure

 

Problem Description: We are using EMC BCV technique to perform. our recovery the same data in another machine (reporting server) and startup. It’s running so smooth for a few years. But currently it failed again and again. EMC suggested suspending database before running BCV split. That can ensure Source devices have the full capability to startup Oracle DB. We don’t know what the impact is for that. As this is production system and need to be available for 7x24 hours. May you help us to investigate the issue and give me the suggestion? The following is one of the error message found in DB log. 

Logs of media recovery failure (Failed to perform. media recovery)
Sat Dec 17 02:45:39 2011
Successful mount of redo thread 1, with mount id 3306275218
Sat Dec 17 02:45:39 2011
Completed: CREATE CONTROLFILE SET DATABASE "RPTCS1" RESETLOGS
Sat Dec 17 02:45:39 2011
ALTER DATABASE RECOVER DATABASE USING BACKUP CONTROLFILE UNTIL CANCEL
Media Recovery Start
ORA-279 signalled during: ALTER DATABASE RECOVER DATABASE USING BACKUP CONT...
Sat Dec 17 02:45:39 2011
ALTER DATABASE RECOVER CONTINUE DEFAULT
Sat Dec 17 02:45:39 2011
Media Recovery Log /archdestcs1/CTMCS1/ctmcs1_1_174294.arc
Recovery interrupted.
Sat Dec 17 02:45:39 2011
Media Recovery failed with error 368
ORA-283 signalled during: ALTER DATABASE RECOVER CONTINUE DEFAULT ...
Sat Dec 17 02:45:39 2011
ALTER DATABASE RECOVER CANCEL
ORA-1112 signalled during: ALTER DATABASE RECOVER CANCEL ...
Sat Dec 17 02:45:39 2011
ALTER DATABASE OPEN RESETLOGS
ORA-1195 signalled during: ALTER DATABASE OPEN RESETLOGS...
Sat Dec 17 02:45:39 2011

 

Solution:

According to your reference: Successful BCV splitting Guarantees Successful Oracle Recovery [ID 221779.1], we are applying the similar processes with third party media EMC BCV synchronization and SPLIT combined. But, we still meet “Media Recovery failed with error 368” issues. As suggested from EMC, we plan to add “ALTER SYSTEM SUSPEND” and “ALTER SYSTEM RESUME”. As we discussed with EMC support so far, EMC BCV synchronization has limitation to completely synchronize the source with “high transactions” to target. So, as suggested from EMC, we expect to know what the database impact will be if we change the Oracle DB to “suspend” status. And how long the off peak time we will be expected to change the database back to “resume” status as routine operation. 

The ability to suspend io operations on a database during a split mirrror backup may be of benefit
depending on the software/hardware in used for the mirrors.
The only case where ALTER SYSTEM SUSPEND is needed is when the mirror split will take forever
without it. It was implemented to get around some of the problems with some mirroring
implementations (where, if we kept doing writes the splitting of mirrors would take a long
time/forever to complete).

Bottom line is: this feature is not supplied for Oracle's benefit - only use it if your mirroring
software/hardware vendor requires it.

And what kind of processes are still running under database suspended? 
Even is suspend on the following process are still writings : 
DBWR -- flushing blocks.
LGWR -- flushing redo buffers (every 3 seconds)
CKPT -- updating datafile headers

 

 

Reference:

http://web.njit.edu/info/oracle/DOC/backup.102/b14191/osbackup007.htm#BRADV204

http://docs.oracle.com/cd/B28359_01/server.111/b28310/start005.htm

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/26136400/viewspace-715136/，如需转载，请注明出处，否则将追究法律责任。

转载于:http://blog.itpub.net/26136400/viewspace-715136/