Restore and Recover a Database to a New Host with a Different Directory Structure (Oracle and Oracle RAC)

The following procedure describes the steps involved in restoring and recovering a database (either in ARCHIVELOG or NOARCHIVELOG) mode to a new host, with a different directory structure.

Before You Begin:

• Review the general and agent-specific restore requirements accessed from Restore Backup Data prior to performing any restore.
• When you perform a database restore to a new host the source and destination machines must have the following features:
  • The same operating systems
  • The same Oracle version
  • Different database datafile directory structures
  • The same ORACLE_SID
  • Must be either a 32-bit or 64-bit system

  In the destination machine

  • Install the Oracle iDataAgent or create the Oracle RAC client and configure as client in the same CommServe in which the source machine resides
  • Create and configure a new Oracle instance, similar to the one existing in the source machine

  In addition the destination machine must have sufficient disk space to accommodate the restored database.

Required Capability: See Capabilities and Permitted Actions

To restore and recover a database to a new host with a different directory structure:

1. Copy the init<SID>.ora from the old host to the new host.
2. Edit the init<SID>.ora file on the new host to reflect all the directory structure changes (i.e., change the path for control files, archivelog destination and *dump destinations).
3. Create the directory structures as defined in init<SID>.ora file for all paths.
4. Configure the destination machine's $TNS_ADMIN/tnsnames.ora file to include the Recovery Catalog database's service entry. The easiest way to configure this file is to copy the Recovery Catalog database's service entry from the source machine's tnsnames.ora file and append it to the destination machine's tnsnames.ora file. The Recovery Catalog database's service entry should look similar to the following:
   <service_name> =
   (DESCRIPTION =
   (ADDRESS = (PROTOCOL = <protocol>)(HOST = <host>) (PORT = <##>))
   (CONNECT_DATA = (SID = <Recovery Catalog database>)))
5. Make sure that ORACLE_SID is set to the database that you are restoring and that ORACLE_HOME is set appropriately. Note the following example for Solaris:
   #export ORACLE_SID= <target database SID>
   #export ORACLE_HOME= <Oracle home directory>
   
   Note the following example for Windows 2000 machines:
   C:\set ORACLE_SID= <target database SID>
   C:\set ORACLE_HOME= <Oracle home directory>
6. From Server Manager or SQL Plus on the source machine, create a new user account within the Recovery Catalog for the destination machine by executing the commands in the following display:
   SVRMGR>create user <username> identified by <password>
   2>temporary tablespace <temp_tablespace_name>
   3>default tablespace <default_tablespace_name>
   4>quota unlimited on <default_tablespace_name>;
   Statement processed.
   SVRMGR>grant connect, resource, recovery_catalog_owner to <username>;
   Statement processed.

   Make sure that the <username>you supply for the destination machine is different than the username for the source machine. When you are granting the user database permissions, make sure to include the recovery_catalog_owner option shown above.

7. Transfer the Oracle password file "orapw<Oracle SID name>" from the source machine to the destination machine. Usually this file resides in ORACLE_HOME/dbs.
8. From the source machine, use the Oracle exp command to export the Recovery Catalog data. You are exporting the data by user. For example, if the user ID for the Recovery Catalog owner is rman, you would export the database backup information for the user rman.
9. From the source machine, use the Oracle imp command to import the contents of the file you created in the previous step into the new user account for the destination machine. Depending on the version of Oracle that you are running, you may receive the following error when you run the imp command:
   IMP-00044: unable to allocate enough memory for statement
10. Perform these steps if you receive an error while running the imp command:
    • Drop the user.
    • Recreate the user.
    • Rerun the imp command.
    • When you rerun the imp command, enter 1000000 when prompted for the insert buffer file.
    • If you receive the error message again, repeat the preceding steps and increase the buffer by a million each time you run imp until the error message is no longer displayed.
11. Ensure that the database in NOMOUNT mode in the destination machine.
12. If you want to perform the cross-machine restore from the CommCell Console then continue on with this procedure. Otherwise, if you want to perform the cross-machine restore from the RMAN command line using your own custom script, then exit this help and run the script now (refer to Oracle's Recovery Manager documentation for more information).

    When performing cross-machine restore, do the following:

    • Restore the control file and put the database in MOUNT mode.
    • Login to SQL database in the destination computer and provide the following syntax:

      SQL> alter database rename file '/oracle/oradata/rman10g1/redo01.log' to '/space/oracle/oracle10g2/oradata/redo01.log';

      SQL> alter database rename file '/oracle/oradata/rman10g1/redo02.log' to '/space/oracle/oracle10g2/oradata/redo02.log';

      SQL> alter database rename file '/oracle/oradata/rman10g1/redo03.log' to '/space/oracle/oracle10g2/oradata/redo03.log';

      Here, '/oracle/oradata/rman10g1/ is the source redo log location and '/space/oracle/oracle10g2/oradata/ is the destination redo log location.

13. From the CommCell Browser, right-click the instance that contains the database you want to restore and recover to a new host, click All Tasks and then click Restore.
14. From the Oracle Restore Options (General) dialog box, select or enter the following:
    • Click the name of the client computer to which you want to restore in the Destination Client box.

      If your destination client is not listed, check and ensure that all the requirements stated in the Before you Begin section is satisfied.

    • Use the Catalog Connect fields to enter the connect string for the catalog database of the destination machine.
    • Select the Restore Control File option.
    • Select the Restore Data and Recover options.

      It is critical that the Recover option is selected, as Recover applies any incremental backups that are needed to bring the database back to a consistent state.

    • Select the entire database in The Latest Database View pane (if it is not already selected).
15. Click the Advanced button and verify the following options are selected in the identified tab for the Oracle Advanced Restore Options dialog box:
    • Restore Control File - To Point in Time in Ctrl & SP Files

      The Control file must be restored to a point in time that is later than or equal to time set for the Restore Data option in the Restore tab.

    • Restore Data in Restore

      The Restore Data option must be earlier than or equal to the point in time that you specified for the Restore Control File option in the Ctrl & SP Files tab.

    • Recover - Point In Time in Recover

      The point in time for the recover option must be later than or equal to the point in time specified for the Control file in the Ctrl & SP Files tab.

    • Redirect All Table Spaces to, along with the path, in Redirect

      The path for datafiles should be created manually because Oracle or Oracle iDataAgent will not create that path.

16. Click OK on the Oracle Advanced Restore Options dialog box.
17. When restoring encrypted data, refer to Data Encryption.
18. Optionally for Oracle RAC, from the Oracle Restore Options (Stream) dialog box, set the instance restore order per stream allotment.
19. Begin or schedule the restore.

    Before the restore job is run, the system will check whether the database is in the correct mode for this type of restore. If it is not, a message dialog will inform you of this and offer to automatically switch the database into the valid mode. When this happens, either click Yes on the message dialog to have the system automatically switch the database to the valid mode and continue the restore, or click No to manually switch the database to the valid mode before proceeding.

20. View the progress of the restore in the Job Controller window.
    • In ARCHIVELOG mode - When the data is restored, click OK on the confirmation window. If desired, click Detail to review information on the restore operation.
    • In NOARCHIVELOG mode - The job will report the restore as failed. This is because no archive redo log files exist. RMAN will report the following message, in reference to this error:
      
      RMAN-03022: compiling command: recover(4)
      RMAN-03026: error recovery releasing channel resources
      RMAN-08031: released channel: ch1
      RMAN-00571:
      ======================================================
      RMAN-00569:====== ERROR MESSAGE STACK FOLLOWS=========
      RMAN-00571:
      ======================================================
      RMAN-03002: failure during compilation of command
      RMAN-03013: command type: recover
      RMAN-06054: media recovery requesting unknown log: thread 1 scn 72077
      RMAN>
      Recovery Manager Complete.
    Perform the following steps if your database is in the NOARCHIVELOG mode.
21. Manually recover and open the database using the following steps.
22. From the Server Manager window, execute the command displayed in the following example:
    SVRMGR>recover database using backup control file until cancel;
    
    Server Manager displays several messages, including the following:
    Specify log: {<RET>=suggested | filename | AUTO | CANCEL}
23. Type cancel
    
    The following message is displayed:
    Media recovery cancelled.
24. Execute the commands displayed in the following example:
    SVRMGR>alter database open resetlogs;
    SVRMGR>exit
25. Reset the Recovery Catalog by executing the commands displayed in the following example:
    #rman rcvcat <catalog connect string>
    Recovery Manager: Release 8.1.5.00 - Production
    RMAN-06008: connected to recovery catalog database
    RMAN>connect target;
    RMAN-06005: connected to target database: <database_name> DBID=<##>)
    RMAN> reset database;
    RMAN-03022: compiling command: reset
    RMAN-03023: executing command: reset
    RMAN-08006: database registered in recovery catalog
    RMAN-03023: executing command: full resync
    RMAN-08002: starting full resync of recovery catalog
    RMAN-08004: full resync complete
    RMAN>exit
    
    Resetting the Recovery Catalog creates a new incarnation of the database. Before you can restore and recover the database to the current time, you must perform another full backup and wait for it to complete successfully.
26. Perform an immediate offline full backup.