NetVault Plug-in for Oracle 12.2


	NOTE: You can also use the following procedure to restore the Oracle Server to a different host, one that is not the original production server, while the production server is still active. However, as stated previously, if your goal is create a copy of the target production database for ongoing use on a new host, do not use this procedure. Instead, use the Duplicate Database backup type. For more information, see Duplicate Database backup.

The same procedure should be used whether a Recovery Catalog was being used as part of the backup strategy for the RMAN Repository. RMAN always stores information concerning the backups in records in the Control File. The version of this information in the Control File is the authoritative record of RMAN’s backups of your database. RMAN can conduct all necessary backup and recovery operations using just the repository information in the Control File.

For variations on the following disaster recovery procedure, see Performing Disaster Recovery in the Oracle Database Backup and Recovery Advanced User’s Guide.

To perform a disaster recovery without a Recovery Catalog, the following requirements must be met:

•

Oracle Database software installed: Install the same version of the Oracle database software on the system as existed on the damaged server. The ORACLE_HOME should be identical on both the source and destination Oracle server. You do not have to create the database.

•

Oracle password file created: Use the orapwd utility to create the Oracle password file.

•

On Linux:

orapwd file=<Oracle_Home>/dbs/orapw<SID> password=<password>

•

On Windows:

orapwd file=<Oracle_Home>\database\orapw<SID> password=<password>

•

Oracle service on Windows created and started: Create the Oracle Service on Windows with Oracle’s oradim utility. The Oracle SID should be identical to the original or source database. This option creates an identical Oracle Service name. For example, from an OS prompt run:

oradim -new -sid <sid_name>

After the Windows service is created, start the service with the following command from an OS prompt:

net start oracleservice<sid_name>

•

NetVault Backup software and Plug‑in for Oracle installed: The same version of NetVault Backup software and the plug-in must be installed and configured as it originally existed on the damaged server. Ensure that the plug-in’s default attributes are defined as specified in Configuring default settings.

•

DBID of the source database identified: Identify the DBID of the damaged database. If the DBID was not recorded as part of the initial Plug‑in for Oracle installation, look for it in the NetVault Backup binary logs for RMAN-based backups of the source database. The DBID is also available on the Details dialog box or the Backup Contents tab.

•

Default location for SPFILE must exist: Verify that the default location for the SPFILE exists. Failure to do so results in an ORA-17502 Failed to create file error during the SPFILE restore.

•

Backups of all tablespaces made available: Verify that you have the latest backups of all the tablespaces.

To perform the recovery, complete the following steps.

Restore the Oracle Net Services Configuration Files.

If the External Configuration Files were included in a previous RMAN backup or the Plug-in for FileSystem was used to back up the Oracle Net Services configuration files, such as “listener.ora,” “sqlnet.ora,” and “tnsnames.ora,” restore the files to the destination Oracle Database Server. Otherwise, configure the Oracle Net Services on the destination Oracle server.

Start the Oracle listener.

For example, from an OS command prompt run:

lsnrctl start

Start RMAN with no catalog, and connect to the target database:

RMAN TARGET / NOCATALOG

Set the DBID to the damaged database DBID.

SET DBID <Source_DBID>;

Start the target instance without mounting the database.


	IMPORTANT: A series of Oracle errors might appear if a PFILE is not available when you run the following command. This process is acceptable, and you can ignore the errors. Oracle uses a default PFILE to ensure that the restore process for the SPFILE continues — “starting Oracle instance without parameter file for retrieval of spfile.”

STARTUP NOMOUNT;

Restore the SPFILE from Autobackup.

RUN {

ALLOCATE CHANNEL C1 TYPE 'SBT_TAPE';

RESTORE SPFILE FROM AUTOBACKUP;

}

Restart the database with the restored SPFILE.

STARTUP FORCE NOMOUNT;

Restore the Control File from Autobackup.

If you are restoring the Control File from the last Autobackup, run:

RUN {

ALLOCATE CHANNEL CH1 TYPE 'SBT_TAPE';

RESTORE CONTROLFILE FROM AUTOBACKUP;

}


	IMPORTANT: If you do not have an Autobackup of the Control File, complete the steps in Retrieving a Control File manually before continuing.

On the Create Restore Job — Choose Saveset page, find the latest available backup that included all the tablespaces, and click Next.

On the Create Selection Set page, navigate to the source NetVault Backup Client and database, select the Whole Database node, and click Edit Plugin Options.

Click the Restore Options tab, and select the applicable options:

•

To mount the database before the performing the restore, select Mount Database Before Restore.

•

To include read-only datafiles in the restore, select Check Read-Only Datafiles and Restore if Necessary.

Because recovery must be performed in a disaster recovery scenario, on the Perform Recovery tab, select Perform Database Point-in-Time Recovery.

PIT Recovery recovers the database to a specific time, log sequence number, or SCN. If the specific time or log sequence of the failure is not known, Quest recommends that you use the SCN option.

On the Perform Recovery tab, select Check Read-Only Datafiles and Recover if Not Current to include read-only datafiles in the recovery process.

The read-only files are included only if they are needed to complete recovery. If they are not damaged, RMAN omits them.

On the Recovery Catalog tab, clear the Use Catalog option to ensure that the Recovery Catalog is not used for this restore.

Complete the Schedule and Advanced Options tabs, and submit the job.

RMAN automatically chooses the best available Full, Incremental, or Archive Log backups and restore and recover the database.

In RMAN, update the RMAN metadata.

CROSSCHECK ARCHIVELOG ALL;

Exit RMAN.


	IMPORTANT: After completing a disaster recovery, Quest recommends that you perform a Full or Incremental Level 0 backup with the database selected on the Create Selection Set page.

Retrieving a Control File manually

When Control File Autobackups are not available, the RMAN Catalog is not used, and only Control File backups were performed by selecting the Control File node on the NetVault Backup Selections page, you might be able to use the following procedure to extract the Control File from a backup piece.

On the Create Restore Job — Choose Saveset page, find the latest available backup that is marked with “Controlfile Included.”

Recover the Control File.

SQL> STARTUP NOMOUNT;

DECLARE

devtype varchar2(256);

done boolean;

BEGIN

devtype := dbms_backup_restore.deviceallocate('sbt_tape', params=>'');

dbms_backup_restore.restoresetdatafile;

dbms_backup_restore.restorecontrolfileto

('<destination PathAndNameOfControlFile>');

dbms_backup_restore.restorebackuppiece

('<backupSegment NameIdentifiedEarlier>',done=>done);

devtype varchar2(256);

done boolean;

BEGIN

devtype := dbms_backup_restore.deviceallocate('sbt_tape', params=>'');

dbms_backup_restore.restoresetdatafile;

dbms_backup_restore.restorecontrolfileto

('C:\oracle\product\10.1.0\oradata\o10gg\control01.ctl');

dbms_backup_restore.restorebackuppiece

('ZETK2003:#3293:ZETK2003:274_578391288:O10GG_1135958971',done=>done);

END;

Use the REPLICATE CONTROLFILE command to copy the restored Control File.

RMAN TARGET / NOCATALOG

SET DBID <Source_DBID>;

connect TARGET /

RUN {

ALLOCATE CHANNEL C1 TYPE 'SBT_TAPE';

replicate controlfile from

'<sourcePathAndNameOfControlFile>';

}

Using advanced procedures with RMAN restores

This topic contains subtopics that describe other restore operations that can be performed with Plug‑in for Oracle.

•

Renaming or relocating files

•

Duplicating a database in a non-RAC environment

•

Performing table-level recovery

Renaming or relocating files

With the RMAN backup method, the plug-in lets you rename the Control File and datafiles during restore. Renaming files can be useful if you do not want to overwrite the existing version and want to create a copy of the file. You can also relocate the files to a different directory while restoring it with the plug-in.

Complete the steps outlined in Preparing the database for recovery.

In the Navigation pane, click Create Restore Job, select Plug‑in for Oracle from the Plugin Type list.

In the saveset table, select the applicable item, and click Next.

For more information, see Selecting data for a restore.

On the Create Selection Set page, select the datafile or Control File to restore.

From the Actions list, select Rename.

In the Rename/Relocate dialog box, enter the applicable information.

•

Rename: Enter the new name.

•

Relocate: Enter the new path.

The plug-in uses the RMAN SET NEWNAME command to restore or rename — or both — RMAN backups. The SET NEWNAME command has precedence over the DB_CREATE_FILE_DEST parameter. If the full path is not specified for a restore or rename, the plug-in behaves as follows:

•

SPFILE: The file is restored to the “$ORACLE_HOME/dbs” directory with the new name.

•

Datafile:

•

If you specify a new location when the file is created, the file is restored or renamed to this location with the new name.

•

If you do not specify a new location when the file is created, the file is restored or renamed to “$ORACLE_HOME/dbs” with the new name.