Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: ORA-00600 problem with restored backups

Re: ORA-00600 problem with restored backups

From: Jesper <gollumNOSPAM_at_image.dk>
Date: Mon, 24 Jan 2000 20:19:40 GMT
Message-ID: <388cac20.127312195@news.image.dk>


Basically your problem is that you are backing up the controlfiles while still in backup mode. This way you will recover to a state where the datafiles are not necessarily intact. What you should do is:

for each tablespace loop
  alter tablespace <ts> begin backup;
  copy all datafiles from tablespace
  alter tablespace <ts> end backup;
end loop;
alter database backup controlfile to '<filename>'; alter system switch logfile;

then restore the datafiles and the controlfile copy on the backup server, mount the database and apply all requested logfiles during recovery.

ORA-600 [4146] is to be considered as a corruption. It signals that there is a mismatch between the undo block and the itl. Possible causes are a lost write or a forced open of the database. For the first you should check your hardware, the second is the one that applies in your case, since your database is not consistent when opened (you shouldn't by any chance have a line in your init.ora that reads: _allow_resetlogs_corruption=true ???).

The ORA-81 is probably caused by the corruption.

Hope this helps,
Jesper

On 24 Jan 2000 15:41:20 GMT, greg_at_nmia.com (Gregory Riggs) wrote:

>Hello,
>
>We are trying to verify that we can restore our production database from our
>nightly hot backup tapes. We have Oracle7 Server Release 7.3.2.3.1 running on
>OpenVMS VAX V7.1, which is nolonger supported by Oracle, but due to the legacy
>software we are using, we can not upgrade Oracle.
>
>When we load the database from tape onto a different, but identically
>configured VAX (the same version Oracle and OpenVMS), the database mounts,
>recovers, and opens without any errors. However, when we try and select from
>some of the tables we get the following Oracle error:
>
>SELECT COUNT(*) FROM <some_table>;
>
>ERROR:
>ORA-00081: address range [1024, 5120) is not readable
>ORA-00600: internal error code, arguments: [4146], [225], [222], [], [], [],
>[], []
>
>Does anyone have any ideas what this error might indicate or suggestions of
>something we could try?
>
>The database is needed 24 hours/day, so each night we make a hot backup of the
>database to tape. We put all the table spaces into backup mode with "ALTER
>TABLESPACE <tablespace_name> BEGIN BACKUP;" before backing up the data files
>'*.DBS' , redo log files '*.RDO', and control files '*.CON'. After this, we
>then take all table spaces out of backup mode with "ALTER TABLESPACE
><tablespace_name> END BACKUP;", issue "ALTER SYSTEM SWITCH LOGFILE;", and then
>backup all the arhcive files '*.ARC' since the previous night's hot backup. To
>restore the database, we restore the backed up files to the other server, and
>issue "STARTUP MOUNT <name>;", "RECOVER DATABASE;", and "ALTER DATABASE OPEN;"
>
>We've tried restoring the database from about a weeks worth of nightly backup
>tapes, but always get the same internal error code problem.
>
>Thanks in Advance!
>Greg Riggs
>greg_at_nmia.com
>
Received on Mon Jan 24 2000 - 14:19:40 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US