Failure with cloning database structure and then restore from source db

Question

Hello, dear experts!
I am trying to duplicate a database.
First step was to use the DBCA to create a template. Because there is no downtime possible, I only created a template of the structure without user schema and without objects.
At another destination I used this template to create a new database "CLON".
This worked successful. The database is up and running. (But quite empty).
Then I transferred the latest full backup of my source database to the new database. I have the same structure at the destiantion as at the
source.
Now I want to use rman restore to build up the clone with this backup. Therefore I start RMAN and connect to the target without catalog.
"RMAN> restore database;"
leads to:
Starten restore um 27.09.07
Kanal ORA_DISK_1 wird benutzt
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of restore command at 09/27/2007 16:26:16
RMAN-06026: some targets not found - aborting restore
RMAN-06023: no backup or copy of datafile 709 found to restore
RMAN-06023: no backup or copy of datafile 708 found to restore
RMAN-06023: no backup or copy of datafile 707 found to restore
"RMAN> list backup;"
does not show any results.
"RMAN> crosscheck backup;"
Zugewiesener Kanal: ORA_DISK_1
Kanal ORA_DISK_1: SID=17 Gerõtetyp=DISK
"RMAN> list backup;"
does not show any results.
To me this looks as if RMAN does not identify the/any backup.
What has to be done to solve the problem?
Thank you in advance,
Hergen.

Answer 1

Hi, I assume that by clone you mean a duplicate of the DB including ALL tablespaces and not just a few right? YOu want to also rename the DB according to what I read.
As a note, your new database does not see the backups because you do not connect to the catalog, and you have no backup info as a result.
Correct me if I am wrong if I misundestand your situation:
First if you want to duplicate, you do not need to create an empty DB. Just create an new instance without any controlfiles or datafiles. Then you restore your rman backup
You could perhaps adapt this procedure I have used recently to clone databases with RMAN backups: You just need RMAN backups, taken as hot backups is fine. You also need the controlfile autobackup from RMAN. If you do not want to use the catalog, you can use the controlfile as catalog, and register the backup info to the controlfile, then restore.
Here is the procedure:
RMAN RESTORE DB FROM COMPLETE LOSS (without use of catalog)
1-     Install oracle db home and patch.
Note: As much as possible, patch to the same version of the DB to be restored. If you do not, you should patch to a later release (do not do this if at all possible, as problems may occur). After opening with reset logs, you will have to startup with the upgrade option and run the cat upgrade script to update the DB to the current patchset.
2-     Install or configure an instance on the desired node. It is not necessary to create a database.
a.     Run oradim to create the instance.
Oradim –new –SID sid –SYSPWD sysdbapass –STARTMODE manual|automatic –PFILE ‘….path to pfile….\initsid.ora’
NOTE: The pfile is optional, required if you do not use a spfile
b.     Create and configure a listener, as rman requires oracle net connectivity.
i.     Use oracle utilities to create listener.
ii.     Use oracle utilities to configure listener to register the instance just created. (or manually edit the listener.ora file).
c.     Configure a tnsnames.ora file so that rman can reach the database through the listener.
d.     Create all directories for datafiles, admin, redologs, archive, flash_recovery_area etc if required.
3-     Create a pfile or get one (ideally the backups should include it).
a.     Change paths in the pfile to reflect new file structure. Also change any settings that are relevant.
Note: if using an spfile, then the rman controlfile autobackup will have it. In that case startup in nomount, start RMAN after setting ORACLE_SID=sid, with ‘RMAN target / NOCATALOG.
Run this in rman
RUN {
SET CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK TO ‘………………..\CONTROL_%F’;
RESTORE SPFILE TO PFILE ‘………\initsid.ora’ FROM AUTOBACKUP;
SHUTDOWN ABORT;
NOTE:
-     If you do not have a specific archive file location (archivelog mode only) the flash recovery area size may need to be increased, depending on the number of archive files to recover.
-     If restoring to a different instance (want to restore an old backup to a different db name, set the db_name in the pfile the same as the db name in the backups. You can rename the db later.
4-     Get the backups desired to restore the DB, and make them available to the node where the restore is to be done.
5-     Startup the database in nomount mode. Startup the service is necessary, then startup the database in nomount mode: specify the pfile location if not in a standard location.
‘STARTUP NOMOUNT PFILE=’……\initsid.ora’’;
6-     Get DB id from controlfile autobackup. Get the most recent controlfile autobackups from the backup files, and find the database id. Below is an example
# CONTROL_c-XXXXXXXXXX-YYYYMMDD-ZZ
# XXXXXXXXXX IS THE DBID
7-     START RMAN after setting ORACLE_SID=sid, with ‘RMAN target / NOCATALOG’.
8-     SET DBID. Get the dbid from above. Run ‘SET DBID XXXXXXXXXXXX;’
9-     Restore controlfile
RUN {
SET CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK TO '… \CONTROL_%F';
RESTORE CONTROLFILE FROM AUTOBACKUP;
ALTER DATABASE MOUNT;
NOTE: If restoring old files, then you must add MAXDAYS=xxx to the restore controlfile command, so that it will find the desired file. Default is 7 days. (explaining why it can’t find older then 7 days).
10-     Catalog backup files. It is probable that the backup files restored will be in a different location once restored (applicable to disk backups). You need to catalog them in RMAN so that RMAN knows about them. List all backup files and run these commands for each backup and controlfile autobackup:
‘CATALOG BACKUPPIECE ….\CONTROL_C-310452700-20070918-01';’
‘CATALOG BACKUPPIECE …\BACKUP_HEIS8FIK_1_1';
NOTES: If the backups are in the same location as they have always been, then a recatalog is not necessary, as the control file has the right backup files information.
If the backups were restored to a different location, then a recatalog must be done so that RMAN can find them. If the regular location of backups is not available, then rman needs to be updated to reflect that. In that case run these commands:
     CROSSCHECK BACKUPS;
     DELETE EXPIRED BACKUP;
     CROSSCHECK ARCHIVELOG ALL;
     DELETE EXPIRED ARCHIVELOG ALL;
If this is test restore (to test disaster recovery for example), likely all the backups are still there, but it is desired to use the restored backup files only. RMAN needs to be updated to reflect that the regular backups do not exist. Make the regular backup files unavailable to RMAN (such are renaming a directory, or preventing access through security), and then run the above commands.
11-     Find the datafile names and numbers. This is only required if datafiles are in a different location on the database to be restored.
Using SQLPlus, set the oracle_sid, then run sqlplus / ;
Run the following command
     Select file# as “file/grp#’, name from v$datafile; (for datafiles).
     Select group#, member from v$logfile. (for redologs)
12-     Restore the database
(use only if all datafile locations are the same!)
RUN
# allocate a channel to disk
ALLOCATE CHANNEL c1 DEVICE TYPE DISK MAXPIECESIZE 2 G FORMAT '…path…\BACKUP_%U';
RESTORE DATABASE;
RECOVER DATABASE;
(use if datafiles are not in the same location)
RUN
ALLOCATE CHANNEL C1 DEVICE TYPE DISK FORMAT ‘…\BACKUP_%U';
SET NEWNAME FOR DATAFILE 1 TO ‘….\SYSTEM01.DBF';
SET NEWNAME FOR DATAFILE 2 TO '…\UNDOTBS01.DBF';
SET NEWNAME FOR DATAFILE 3 TO '…\SYSAUX01.DBF';
SET NEWNAME FOR DATAFILE 4 TO '…\USERS01.DBF';
SET NEWNAME FOR DATAFILE 5 TO '…\TOOLS_01.DBF ';
# add other tablespace locations here.
# We need to tell the controlfile the new location for the redologs
SQL "ALTER DATABASE RENAME FILE ''…\REDO01.LOG'' TO      ''…\REDO01.LOG''";
SQL "ALTER DATABASE RENAME FILE ''…\REDO02.LOG'' TO      ''…\REDO02.LOG''";
SQL "ALTER DATABASE RENAME FILE ''…\REDO03.LOG'' TO      ''…\REDO03.LOG''";
RESTORE DATABASE;
# we need to tell the controlfile the location of the new datafiles.
SWITCH DATAFILE ALL;
RECOVER DATABASE;
13-     Reset the logs. ALTER DATABASE OPEN RESETLOGS;
NOTE: If the oracle home is of a newer revision then the database, the open resetlogs will generate an error. In that case open the database with the upgrade option and then run the catalog upgrade as per patch instructions.
14-     Create a temporary tablespace. The system thinks it has a temp tablespace, but it needs to be recreated
a.     Create another temp tablespace (modify as needed).
     CREATE TEMPORARY TABLESPACE temp2
          TEMPFILE '......\temp02.dbf' SIZE 128M REUSE
          AUTOEXTEND ON NEXT 16M MAXSIZE unlimited
          EXTENT MANAGEMENT LOCAL UNIFORM SIZE 1M;
b.     Point the server to the new file
ALTER DATABASE DEFAULT TEMPORARY TABLESPACE temp2;
c.     Delete the missing one:
          DROP TABLESPACE temp INCLUDING CONTENTS AND DATAFILES;
d.     RECREATE THE temp tablespace;
          CREATE TEMPORARY TABLESPACE temp
               TEMPFILE 'd:\......\temp_01.dbf' SIZE 128M REUSE
          AUTOEXTEND ON NEXT 16M MAXSIZE unlimited
          EXTENT MANAGEMENT LOCAL UNIFORM SIZE 1M;
e.     Change system default tablespace again to temp:
          ALTER DATABASE DEFAULT TEMPORARY TABLESPACE temp;
f.     Delete the transition temp tablespace temp2:
DROP TABLESPACE temp2 INCLUDING CONTENTS AND DATAFILES;
15-     Rename the database if the restored database name needs to change.
Use the nid utility to rename the db. IMPORTANT: the temporary tablespace needs to exist for this to work, or the db rename will fail and a restore will have to be done.
Use :
-     Shutdown database in immediate mode.
-     Startup in nomount.
-     Open command prompt.
-     set oracle_sid = sid
-     nid sys/pass dbname=newname logfile=….\logfile.txt
-     shutdown the db (if required)
-     Change the dbname in the pfile or spfile to the new dbname
-     Create a new password file:
o     Orapwd file=….. password=…
IMPORTANT NOTE: if you rename a DB, your existing backups will not be valid, do backups of the newly renamed db!!!
16-     Open the database and test.
NOTE: Whenever you open with resetlogs, YOU MUST BACKUP YOUR DATABASE. Your old backups can only be used to restore the old DB.
I hope this helps.