Incomplete recovery using RMAN

From: Dan <dan.tudor_at_usa.net>
Date: Tue, 19 Jun 2001 10:38:15 -0400
Message-ID: <tpJX6.5738$rD3.1386885_at_news.total.net>


Hello,

Is there anybody to give me a hint with the folowing:

I make online backups of an Oracle8 database. I have some problems trying to perform DBPITR. What I did is : At backup time:
-I backup all the tablespaces one by one, using a script like the following
one:

run{
allocate channel ch123 type 'sbt_tape' format 'path\filename' parms='BLKSIZE=32768';
backup tag 'tagname' tablespace 'tablespacename' check logical; }

The backup taken with this script allows me to restore and to recover the tablespace at the last moment in time.
-I backup the archived logs
-I backup the control file using a script like the following:

run{
allocate channel ch123 type 'sbt_tape' format 'path\filename' parms='BLKSIZE=32768';
backup tag 'tagname' current controlfile; }

At restore time:
-I restore all the tablespaces but I do not recover them. I use a script
like this one:

run{
allocate channel ch123 type 'sbt_tape' format 'path\filename'; set until time "TO_DATE('6/18/2001 11/40/58','MM/DD/YYYY HH24/MI/SS')"; restore from tag='tagname' tablespace 'tablespacename'; }

-I have all the archived logs on disk
-I restore the controlfile using the following procedure:

run{
allocate channel ch123 type 'sbt_tape' format 'path\filename' parms='BLKSIZE=32768';
restore controlfile to 'Temp_path\tempfile.tmp' from tag='tagname'; shutdown abort;
startup nomount;
replicate controlfile from 'Temp_path\tempfile.tmp' ; alter database mount;
}

Now I have the datafiles as they were at the moment the backup was taken and the controlfile as well. Because the control file was the last item backed up, I have to recover all the tablespaces in the database to reach the SCN that is recorded in the control file. (During the backup process the SCN can change).
I had no errors during the execution of these scripts and I'm sure that I restored the correct files.
At this moment I can mount the database. If I try to open the database I'll get these errors:

ORA-01113: file 1 needs media recovery
ORA-01110: data file 1: "D:\Oracle\Oradata\Tavidb\System01.dbf"

The message is obvious, I have to perform database recovery. So, I use the SQL WorkSheet utility (I tried with svrmgrl and with RMAN as well) to perform the database recovery:
recover database using backup controlfile; The output is bellow:
*
ERROR at line 1:

ORA-00279: change 52847 generated at 06/18/2001 11:40:47 needed for thread 1
ORA-00289: suggestion : D:\ORACLE\ORA81\RDBMS\ARC00225.001
ORA-00280: change 52847 for thread 1 is in sequence #225

I have checked and all the archived logs are at their location on the disk and were never deleted, including the one specified by the error message. Even if I try to recover until the SCN 52847 the error is the same. I backed up all the tablespaces one by one so I cannot use "restore database" and after that "recover database" as Oracle says in the documentation because I didn't backup with "backup database". I choosed to backup the tablespaces one by one because there can be a situation when only one or more tablespaces becomes unavailable and is not necessary to restore the entire database. In this case the DBA can restore only the broken tablespace. In case of a big database this can be really useful. I expect Oracle to recover all the tablespaces using the information in the restored controlfile, but it seems that does not happen like this.

Any help would be appreciated

Thank you,
Octavian Received on Tue Jun 19 2001 - 16:38:15 CEST

Original text of this message