Creating a Duplicate Database on a New Host using RMAN duplicate feature

posted Mar 16, 2011, 12:32 PM by Sachchida Ojha
Duplicate database from server A to server B (Non ASM)

Assumed database names:

Primary Database SID:       PROD
Duplicate Database SID:    AUX
RMAN Catalog SID:          RMAN

====================================================================================
Steps

1.  Backup the primary database.

2.  Determine how much disk space will be required.

3.  Ensuring you have enough space on your target server.

4.  Making the backup available for the duplicate process.

5.  Creating the init.ora & administration directories for the duplicate database.

6.  Ensuring SQL*NET connections to primary database and RMAN catalog are working.

7.  Prepare RMAN duplicate script.

8.  Execute the RMAN script.

=====================================================================================

1. Backup of the primary database.

Host A (Target)

Prior to creating a duplicate database you must have a backup of the target database. In this example we assume the backup media is disk. Find appropriate disk space on host A and take a full backup including archive logs and current controlfile (Figure 1a). If you are planning to duplicate a subset of the database refer to (Figure 1b) this illustrates the RMAN command to backing up certain tablespaces.

[oracle@linux] export ORACLE_SID=PROD

[oracle@linux] rman target=/ catalog=rman/rman@RMAN

RMAN> run {
        allocate channel d1 type disk;
        backup format '/backups/PROD/df_t%t_s%s_p%p' database;
        sql 'alter system archive log current';
        backup format '/backups/PROD/al_t%t_s%s_p%p' archivelog all;
        release channel d1;
      }

       Figure 1a - This command will perform a full database backup including archivelogs and the current controlfile.


[oracle@linux] export ORACLE_SID=PROD

[oracle@linux] rman target=/ catalog=rman/rman@RMAN

RMAN> run {
        allocate channel d1 type disk;
        backup format '/backups/PROD/df_t%t_s%s_p%p'
               tablespace SYSTEM, SYSAUX, UNDO, USERS;
        sql 'alter system archive log current';
        backup format '/backups/PROD/al_t%t_s%s_p%p' archivelog all;
        release channel d1;
     }

    Figure 1b- This command will perform a tablespace backup ( SYSTEM, SYSAUX, UNDO & USERS) including archive logs and the current controlfile.

2. Determine how much disk space will be required.

Host A(Target) -

After deciding what you will be duplicating, calculate the entire space this will require on the new host. The full database calculation (Figure 2a) will calculate the entire space required whereas (figure 2b) allows you to enter the tablespace names in order to calculate the space required. Figure 2c provides a sample output.

Note: sql valid for version 10g only.
SQL> select DF.TOTAL/1048576 "DataFile Size Mb",
            LOG.TOTAL/1048576 "Redo Log Size Mb",
            CONTROL.TOTAL/1048576 "Control File Size Mb",
            (DF.TOTAL + LOG.TOTAL + CONTROL.TOTAL)/1048576 "Total Size Mb"
     from dual,
    (select sum(a.bytes) TOTAL from dba_data_files a) DF,
    (select sum(b.bytes) TOTAL from v$log b) LOG,
    (select sum((cffsz+1)*cfbsz) TOTAL from x$kcccf c) CONTROL;
     Figure 2a - Calculate total space for all datafiles within database.

SQL> select DF.TOTAL/1048576 "DataFile Size Mb",
            LOG.TOTAL/1048576 "Redo Log Size Mb",
            CONTROL.TOTAL/1048576 "Control File Size Mb",
            (DF.TOTAL + LOG.TOTAL + CONTROL.TOTAL)/1048576 "Total Size Mb" from dual,
    (select sum(a.bytes) TOTAL from dba_data_files a
     where tablespace_name in('SYSTEM','UNDOTBS1', 'SYSAUX', 'USERS')) DF,
    (select sum(b.bytes) TOTAL from v$log b) LOG,
    (select sum((cffsz+1)*cfbsz) TOTAL from x$kcccf c) CONTROL;
     Figure 2b - Calculate space for list of tablespaces within primary database.

DataFile Size Mb Redo Log Size Mb Control File Size Mb Total Size Mb
---------------- ---------------- -------------------- -------------
             900              150             20.34375    1070.34375
     Figure 2c - Sample output of space calculation.

3. Ensuring you have enough space on your target server.

Host B (Aux)

Prior to starting the duplicate process you must ensure you have enough disk space within file system to hold the database.

[oracle@linux] df -kh
Filesystem         Size     Used     Avail     Use%     Mounted on
/dev/vg01/root     9.9G     2.8G      6.6G      30%              /
/dev/sda1          145M      16M      122M      12%          /boot
none               999M        0      999M       0%       /dev/shm
/dev/vg01/tmp      2.0G     383M      1.5G      20%           /tmp
/dev/vg01/u01       20G      12G      7.0G      62%           /u01
/dev/vg01/u02      4.9G    1010M      3.6G      22%           /u02
/dev/vg01/backups  5.9G     1.2G      4.4G      22%       /backups
/dev/vg01/oradata   15G      13G      2.0G      87%       /oradata
     Figure 3a - Sample output showing the space available on your filesystem.

Compare the results received from this query with the output from 2a or 2b and ensure you have enough diskspace for your duplicate database.


4. Making the backup available for the duplicate process.

If your backup resides on disk you will need to copy this back up from host A to host B. Ensure you place it in the same directory as where it was created. In the example below (figure 4a) the backup piece resides in ‘/backups/PROD’ these files need to be copied into the same directory on host B.

Furthermore also ensure than all archive log backups have also been moved and any archive logs in the archive log directory that may be require for the recovery. The archive logs required will depend on the point in time of the recovery.

RMAN> list backup;

BS Key Type LV Size Device Type Elapsed Time Completion Time
------- ---- -- ---------- ----------- ------------ -------------------
22 Full 529M DISK 00:00:51 2006/05/16 11:12:54
BP Key: 22 Status: AVAILABLE Compressed: NO Tag: TAG20060516T111203
Piece Name: /backups/PROD/df_t590584323_s23_p1
List of Datafiles in backup set 22
File LV Type Ckp SCN Ckp Time Name
---- -- ---- ---------- ------------------- ----
1 Full 1393845 2006/05/16 11:12:03 /oradata/PROD/system01.dbf
2 Full 1393845 2006/05/16 11:12:03 /oradata/PROD/undotbs01.dbf
3 Full 1393845 2006/05/16 11:12:03 /oradata/PROD/sysaux01.dbf
4 Full 1393845 2006/05/16 11:12:03 /oradata/PROD/users01.dbf


BS Key Size Device Type Elapsed Time Completion Time
------- ---------- ----------- ------------ -------------------
24 48M DISK 00:00:06 2006/05/16 11:13:07
BP Key: 24 Status: AVAILABLE Compressed: NO Tag: TAG20060516T111301
Piece Name: /backups/PROD/al_t590584381_s25_p1

List of Archived Logs in backup set 24
Thrd Seq Low SCN Low Time Next SCN Next Time
---- ------- ---------- ------------------- ---------- ---------
1 78 1344750 2006/05/15 04:55:34 1353208 2006/05/15 10:00:19
1 79 1353208 2006/05/15 10:00:19 1353213 2006/05/15 10:00:20
1 80 1353213 2006/05/15 10:00:20 1371564 2006/05/15 22:00:11
1 81 1371564 2006/05/15 22:00:11 1373291 2006/05/15 22:00:59
1 82 1373291 2006/05/15 22:00:59 1381066 2006/05/16 03:00:05
1 83 1381066 2006/05/16 03:00:05 1390685 2006/05/16 09:03:00
1 84 1390685 2006/05/16 09:03:00 1393870 2006/05/16 11:13:00
1 85 1393870 2006/05/16 11:13:00 1393929 2006/05/16 11:13:00
     Figure 4a - A list backup showing the backup pieces that need to be copied across Host B

5. Creating the init.ora & administration directories for the duplicate database.

Host B(AUX)

Create the pfile [initAUX.ora] parameter file in the $ORACLE_HOME/dbs directory for the auxiliary database. An example is shown below (figure 5a) with bare minimum settings. Note you may require to set parameters which are the same as your production database (refer to you primary init.ora)

# +----------------------------------------+
# | FILE : initAUX.ora                     |
# | DATABASE NAME : AUX                    |
# +----------------------------------------+
# Set the below to location of the clone Duplicate database / name of
# clone database.

audit_file_dest =/oradata/AUX/adump
background_dump_dest =/oradata/AUX/bdump
core_dump_dest =/oradata/AUX/cdump
user_dump_dest =/oradata/AUX/udump
db_name ="AUX"
instance_name =AUX

cluster_database=false (in case the production is a rac environment)



# Set the below to the location of the duplicate clone control file.

control_files =('/oradata/AUX/control01.ctl','/oradata/AUX/control02.ctl','/oradata/AUX/control03.ctl')



# Set the below for the from and to location for all data files / redo
# logs to be cloned.

db_file_name_convert =("/oradata/PROD/", "/oradata/AUX/")
log_file_name_convert =("/oradata/PROD/", "/oradata/AUX/")
#Set the below to the same as the production target

undo_management =AUTO
undo_retention =10800
undo_tablespace =UNDOTBS1
db_block_size = 8192
compatible = 10.2.0.1.0
     Figure 5a - Sample initAUX.ora with minimal settings

Following the creation of the initAUX.ora startup nomount the auxiliary instance.

[oracle@linux]export ORACLE_SID=AUX

[oracle@linux] sqlplus '/as sysdba'

SQLPLUS> startup nomount;
Figure 5b - startup nomount the AUX instance.


6. Ensuring SQL*NET connections to primary database and RMAN catalog are working.

Host B(AUX)

Ensure the production target database is open or at least mounted. If using a catalog database this must also be open. Test your SQL*NET connections: From Host B you must be able to connect to the primary DB as sysdba and make an rman connection to RMAN. Perform a basic test using sqlplus. When you have successfully connected exit sqlplus and move onto the next step.

[oracle@linux]% sqlplus ‘sys/oracle@PROD as sysdba’

[oracle@linux]% sqlplus rman/rman@RMAN   (not mandatory)
     Figure 6a - SQL*NET connections

7. Prepare RMAN duplicate script.

In a working directory on Host B create an RMAN script file duplicate.rcv. The example below (figure 7a) shows the command for a complete duplicate (figure 7b) skips the tablespaces which are not required in the duplicate and (figure 7b) provide the syntax required for a point in time duplicate.

run {
  allocate auxiliary channel C1 device type disk;
  duplicate target database to AUX;
}
     Figure 7a - Sample duplicate command.

run {
  allocate auxiliary channel C1 device type disk;
  duplicate target database to AUX skip tablespace ABC, XYZ;
}
     Figure 7b- Sample duplicate script omitting optional tablespaces;

run {
  set until time "to_date('Jan 01 2000 12:00:00','Mon DD YYYY HH24:MI:SS')";
  allocate auxiliary channel C1 device type disk;
  duplicate target database to AUX;
}
      Figure 7c- Sample duplicate script to a point in time.

8. Execute the RMAN script.

Start RMAN, connect to the production target, the catalog instance and also the auxiliary clone. Run the RMAN duplicate script as shown below. Before doing this ensure that the Oracle SID environment variable is set to the duplicate clone database.

[oracle@linux] export ORACLE_SID=AUX

[oracle@linux] rman target sys/sys@PROD catalog rman/rman@rman auxiliary /

RMAN> @duplicate.rcv
Comments