Re: Startup Error

From: <fitzjarrell_at_cox.net>
Date: 17 May 2005 06:53:28 -0700
Message-ID: <1116338008.879646.248910@g44g2000cwa.googlegroups.com>

Deke wrote:
> I started up a Oracle 10g database running on a WinXP machine and
found
> this:
>
> ORACLE instance started.
>
> Total System Global Area 171966464 bytes
> Fixed Size 787988 bytes
> Variable Size 145750508 bytes
> Database Buffers 25165824 bytes
> Redo Buffers 262144 bytes
> Database mounted.
> ORA-00316: log 3 of thread 1, type 0 in header is not log file
> ORA-00312: online log 3 thread 1:
> 'C:\ORACLE\PRODUCT\10.1.0\ORADATA\LAP2\REDO03.LOG'
>
>
> Can any one help in resolving this matter?
>
>
> Thanks

You could have hellped yourself with this and a quick trip to http:/tahiti.oracle.com. The redo log reported is corrupt. Since you haven't stated whether you've multiplexed your redo logs you may not have the option to replace it with one of the multiplexed copies. You can, however, do this:

SQL> startup nomount
ORACLE instance started.

Total System Global Area 171966464 bytes

Fixed Size                   787988 bytes
Variable Size             145750508 bytes
Database Buffers           25165824 bytes
Redo Buffers                 262144 bytes

SQL> alter database mount exclusive;
Database mounted.
SQL> alter database open resetlogs;
Database opened. <=== Hopefully this is all Oracle will report

If the above doesn't resolve the issue then you'll need to recreate your control files to eliminate the errant redo log:

SQL> startup nomount
ORACLE instance started.

Total System Global Area 171966464 bytes

Fixed Size                   787988 bytes
Variable Size             145750508 bytes
Database Buffers           25165824 bytes
Redo Buffers                 262144 bytes

SQL> alter database mount exclusive;
Database mounted.
SQL> alter database backup controlfile to trace;

Database altered.

SQL> in %ORACLE_BASE%\admin\<sid>\udump will be a trace file containing the commands necesary to retuild a controlfile. Edit this text to remove the corrupt redo log and the information above the actual commands and then execute the script through SQL*Plus after shutting the database down. The text below will be similar to the script you'll generate:

STARTUP NOMOUNT
CREATE CONTROLFILE REUSE DATABASE "DEV9" NORESETLOGS ARCHIVELOG -- SET STANDBY TO MAXIMIZE PERFORMANCE
    MAXLOGFILES 32
    MAXLOGMEMBERS 3
    MAXDATAFILES 1024
    MAXINSTANCES 1
    MAXLOGHISTORY 113
LOGFILE

  GROUP 1 '/oraunix/oradata/dev9/redo01.log'  SIZE 100M,
  GROUP 2 '/oraunix/oradata/dev9/redo02.log'  SIZE 100M,
  GROUP 3 '/oraunix/oradata/dev9/redo03.log'  SIZE 100M

-- STANDBY LOGFILE
DATAFILE
.....

There will be two cases listed; removing the errant redo log should allow you to use the first case (since you've already tried to open the database resetlogs and the corrupt log would not clear). After the database is open you can recreate the third redo log; I'd rename the old file and keep it in place if the issue is due to a bad disk block or sector.

I hope this helps.

David Fitzjarrell Received on Tue May 17 2005 - 08:53:28 CDT