Oracle® Database

5.1 Review Known Preinstallation Issues

Review the information in the following sections. If any of the issues apply to your Oracle installation, follow the instructions before installing the patch set.

• Upgrading Oracle E-Business Suite Databases

• Upgrading Installations that use Physical or Logical Standby Databases

• Upgrading Oracle Workspace Manager

• Upgrading Installations that use Java Virtual Machine or Oracle interMedia

• Upgrading Oracle Clusterware

• Notes to Oracle OLAP users

http://metalink.oracle.com

http://metalink.oracle.com

http://metalink.oracle.com

5.2 Identify the Oracle Database Installation

This is not a complete software distribution. You must install it in an existing Oracle Database 10g Oracle home. Use Oracle Universal Installer to identify Oracle home directories. From the Welcome screen, click Installed Products, then click Environment tab.

If you are installing this patch set on an existing Oracle Real Applications Cluster (Oracle RAC) 10g release 10.2.0.x installation, you must run Oracle Universal Installer from the same node from which the Oracle Database software was installed.

5.3 Check Postrelease Updates

Before installing this patch set in a production environment, review document 316900.1, ALERT: Oracle 10g release 2 (10.2) Support Status and Alerts, available on the OracleMetaLink Web site:

http://metalink.oracle.com

To locate this document:

1. Log on to OracleMetalink.

2. Click Advanced at the top of the OracleMetalink page.

3. Enter 316900.1 in the Document ID field, then click Submit.

This document is created by Oracle Support Services and provides information about the status of issues discovered after this patch set was released. If you are unable to access this site, then contact Oracle Support Services before installing this patch set in a production environment.

5.4 Download and Extract the Installation Software

Download and extract the p5337014_10203_WINNT.zip patch set installation archive to a directory that is not the Oracle home directory or under the Oracle home directory. For example, Oracle_patch.

5.5 Update Oracle Time Zone Definitions

The 10.2.0.3 patch set includes an update to the Oracle time zone definitions. The 10.2.0.3 patch set updates the standard version of the Oracle 10.2 time zone definitions version-2 to version-3. This version of the time zone definitions includes the changes to daylight saving time in the USA in 2007, and many other updates.

A list of updated time zones in version-3 of these definitions are available from document 412160.1 on the OracleMetalink Web site:

http://metalink.oracle.com

To locate document 412160.1:

1. Log on to OracleMetalink.

2. Click Advanced at the top of the OracleMetalink page.

3. Enter 412160.1 in the Document ID field, then click Submit.

Perform the following actions to ensure that time zone data affected by the changes in these time zones are updated appropriately:

1. Download the patch 5601428 from OracleMetalink.

2. Install the patch. The script utltzuv2.sql is installed into the C:\>ORACLE_BASE\ORACLE_HOME\rdbms\admin directory.

3. Run utltzuv2.sql to check the data of datatype TIMESTAMP WITH TIME ZONE (TSTZ). The output of this script is a results table which highlights the affected columns.

4. Refer to the instructions in the utltzuv2.sql or OracleMetalink note 396671.1: Usage of utltzuv2.sql before updating time zone files in Oracle 10, to evaluate results and if required save the existing TSTZ data. If no data was found, no further actions are needed.

   
   

   Note: If you saved the existing TSTZ data, you need to complete Updating Oracle Time Zone Definitions under the Postinstallation Tasks section to view the change.

   
   

   To locate document 396671.1 on the OracleMetalink Web site:

   http://metalink.oracle.com
   
   

   To locate document 396671.1:

   1. Log on to OracleMetalink.

   2. Click Advanced at the top of the OracleMetalink page.

   3. Enter 396671.1 in the Document ID field, then click Submit.

   
   

   Note: If you have previously installed a patch for bug 4689959, the database is already using version 3 of the Oracle time zone definitions. In this case, the script in step 3 will display the following message: TIMEZONE data is consistent with the latest version 3 transition rules, No need to validate TIMEZONE data. In this case, no further action is needed and step 4 can be skipped. If you have previously installed a patch for bug 5632264, the database is already using version 4 of the Oracle time zone definitions, which supersedes the version included in this patch set. The script in step 3 will display the following message: Your time zone upgrade needs a different script. Please contact Oracle support!. In this case, the actions from step 4 should be skipped and you should re-apply patch 5632264 on the 10.2.0.3 installation as soon as the patch set installation is completed.

   
   

5.6 Stop All Services

Go to the following sections based on your Oracle Database installation type:

• Stopping All Services for a Single Instance Installation

• Stopping All Services for an Oracle Clusterware Installation

• Stopping All Services for an Oracle RAC Installation

5.7 Back Up the System

Oracle recommends that you create a backup of the Oracle Database 10g installation before you install the patch set.

6.1 Installing the Oracle Database 10g Patch Set Interactively

To install the Oracle Database 10g patch set interactively:

OUI-10091: There are no patches that need to be applied from the patch set Oracle Database 10g Release 2 Patch Set 2
10.2.0.3

7.1 Upgrading Earlier Oracle Database to Oracle Database 10g Release 2 (10.2.0.3)

Review the following sections before upgrading an earlier Database to Oracle Database 10g release 2 (10.2.0.3):

• Upgrading Earlier Oracle Database Releases to Oracle Database 10g Release 2 (10.2.0.3)

• Upgrading a Release 9.2 Database Not Using Oracle Label Security

7.2 Upgrading Oracle Database 10g Release 10.2.0.x to Oracle Database 10g Release 10.2.0.3

See one of the following sections for upgrading an Oracle Database 10g release 10.2.0.x to Oracle Database 10g release 10.2.0.3:

• Upgrading a Release 10.2 Database using Oracle Database Upgrade Assistant

• Manually Upgrading a Release 10.2 Database

7.3 Oracle Database Extensions for .NET

If you use Oracle Database Extensions for .NET, then complete the following steps and start the OracleORACLE_HOMEClrAgent service:

1. Use SQL*Plus to deploy the DBMSClr.plb package:

   C:\> sqlplus /NOLOG
   SQL> CONNECT SYS/SYS_Password AS SYSDBA
   SQL> @ORACLE_BASE\ORACLE_HOME\rdbms\admin\DBMSClr.plb
   
   

2. Start the OracleORACLE_HOMEClrAgent service:

   From the Start menu, select Programs, then Administrative Tools and then Services to open the Services control panel then select OracleORACLE_HOMEClrAgent and click Start.

7.4 Updating Oracle Time Zone Definitions

If you saved TSTZ data as mentioned in Update Oracle Time Zone Definitions section under Preinstallation Tasks, you need to restore this data to the TSTZ columns where they originated from.

Refer to the instructions in the utltzuv2.sql script in step 3 of Update Oracle Time Zone Definitions or OracleMetalink note 396671.1: Usage of utltzuv2.sql before updating time zone files in Oracle 10.

To locate document 396671.1 on the OracleMetalink Web site:

http://metalink.oracle.com

To locate document 396671.1:

1. Log on to OracleMetalink.

2. Click Advanced at the top of the OracleMetalink page.

3. Enter 396671.1 in the Document ID field, then click Submit.

8.1 Removing the Patch Set Software for Single Instance Installation

Perform the following steps for backing out from the 10.2.0.3 patch release:

1. Use SQL*Plus to log in to the database as the SYS user with SYSDBA privileges:

   C:\> sqlplus /NOLOG
   SQL> CONNECT SYS/SYS_Password AS SYSDBA
   SQL> SHUTDOWN IMMEDIATE
   SQL> EXIT
   
   

2. Take a backup of the following file in the 10.2.0.3 Oracle home directory:

   C:\> ORACLE_BASE\ORACLE_HOME\rdbms\admin\catrelod.sql
   C:\> ORACLE_BASE\ORACLE_HOME\network\admin\tnsnames.ora
   
   

3. Enter the following SQL*Plus commands:

   C:\> cd ORACLE_BASE\ORACLE_HOME
   C:\> sqlplus /NOLOG
   SQL> CONNECT SYS/SYS_Password AS SYSDBA
   SQL> STARTUP DOWNGRADE
   SQL> SPOOL downgrade.out
   SQL> @ORACLE_BASE\ORACLE_HOME\rdbms\admin\catdwgrd.sql
   SQL> SPOOL OFF
   SQL> SHUTDOWN IMMEDIATE
   SQL> EXIT
   
   

4. Review the downgrade.out file for errors.

5. Restore the Oracle10g installation and Central Inventory that you backed up before applying the patch set.

6. After restoring the original Oracle home, copy the saved version of catrelod.sql script into the restored Oracle home rdbms\admin directory. Also copy the saved version of 10.2.0.3\network\admin\tnsnames.ora directory into the restored Oracle home \network\admin\tnsnames.ora directory, and then perform the following steps:

   1. Start the OracleServiceSID service from the Control Panel by selecting Administrative Tools and then Services.

   2. Enter the following SQL*Plus commands:

      C:\> sqlplus /NOLOG
      SQL> CONNECT SYS/SYS_Password AS SYSDBA
      SQL> STARTUP DOWNGRADE
      SQL> SPOOL catrelod.out
      SQL> @ORACLE_BASE\ORACLE_HOME\rdbms\admin\catrelod.sql
      SQL> SPOOL OFF
      SQL> SHUTDOWN IMMEDIATE
      
      

7. Review the catrelod.out file for errors.

   
   

   Note: Ignore the ORA-29844 and ORA-31085 errors found in the reload.out file. These errors are tracked with Oracle bugs 4758112 and 4425495 respectively. After reloading the 10.2.0.2 release, the version numbers for Oracle Data Mining and Oracle OLAP Analytic Workspace remain at 10.2.0.3. These issues are tracked with Oracle bugs 4758695 and 4751917.

   
   

8. Start the database and recompile the remaining invalid objects:

   SQL> STARTUP
   SQL> @ORACLE_BASE\ORACLE_HOME\rdbms\admin\utlrp.sql
   
   

8.2 Removing the Patch Set Software for Oracle RAC Installation

Perform the following steps for backing out from the 10.2.0.3 patch release:

1. Use SQL*Plus to log in to the database as the SYS user with SYSDBA privileges:

    C:\> sqlplus /NOLOG
    SQL> CONNECT SYS/SYS_Password AS SYSDBA
   
   

2. Set the CLUSTER_DATABASE initialization parameter to FALSE:

   
   

   Note: If the system uses an initialization parameter file, change the value of the CLUSTER_DATABASE initialization parameters to TRUE in the initialization parameter file (initsid.ora).

   
   

   SQL> ALTER SYSTEM SET CLUSTER_DATABASE=FALSE SCOPE=spfile; 
   
   

3. Exit SQL*Plus and shut down the database:

   C:\> ORACLE_BASE\ORACLE_HOME\bin\srvctl stop database -d db_name
   
   

4. Take a backup of the following file in 10.2.0.3 Oracle home directory:

   C:\> ORACLE_BASE\ORACLE_HOME\rdbms\admin\catrelod.sql
   C:\> ORACLE_BASE\ORACLE_HOME\network\admin\tnsnames.ora
   
   

5. Enter the following SQL*Plus commands:

   C:\> sqlplus /NOLOG
   SQL> CONNECT SYS/SYS_Password AS SYSDBA
   SQL> STARTUP DOWNGRADE
   SQL> SPOOL catdwgrd.out
   SQL> @ORACLE_BASE\ORACLE_HOME\rdbms\admin\catdwgrd.sql
   SQL> SPOOL OFF
   SQL> SHUTDOWN IMMEDIATE
   SQL> EXIT
   
   

6. Complete the following steps to stop all Oracle Database Services:

   1. Stop all node applications running on the node. Enter the following command, where node is the name of the node where the applications are running:

      C:\> ORACLE_BASE\ORACLE_HOME\bin\srvctl stop nodeapps -n node
      
      

      For Automatic Storage Management instances, perform the following step:

      C:\> ORACLE_BASE\ORACLE_HOME\bin\srvctl stop asm -n node
      
      

   2. Stop the services OracleServiceinst_name and OracleDBConsoleinst_name from the Control Panel by selecting Administrative Tools and then Services.

7. Restore the Oracle10g installation and Central Inventory that you backed up before applying the patch set on both the nodes.

8. After restoring the original Oracle home, copy the saved version of catrelod.sql script into the restored Oracle home rdbms\admin directory. Also copy the saved version of 10.2.0.3\network\admin\tnsnames.ora directory into the restored Oracle home \network\admin\tnsnames.ora directory, and then perform the following steps:

   1. Start the OracleServiceSID service from the Control Panel by selecting Administrative Tools and then Services.

   2. Enter the following SQL*Plus commands:

      C:\> sqlplus /NOLOG
      SQL> CONNECT SYS/SYS_Password AS SYSDBA
      SQL> STARTUP DOWNGRADE
      SQL> SPOOL catrelod.out
      SQL> @ORACLE_BASE\ORACLE_HOME\rdbms\admin\catrelod.sql
      SQL> SPOOL OFF
      SQL> SHUTDOWN IMMEDIATE
      
      

9. Start the listener, gsd and ons processes:

   1. Start the node applications by entering the following command, where node is the name of the node where the applications are running:

      C:\> ORACLE_BASE\ORACLE_HOME\bin\srvctl start nodeapps -n node
      
      

      For Automatic Storage Management instances, perform the following step:

      C:\> ORACLE_BASE\ORACLE_HOME\bin\srvctl start asm -n node
      
      

   2. Start the service OracleServiceinst_name from the Control Panel by selecting Administrative Tools and then Services.

10. Review the catrelod.out file for errors.

    
    

    Note: Ignore the ORA-29844 and ORA-31085 errors found in the catrelod.out file. These errors are tracked with Oracle bugs 4758112 and 4425495 respectively. After reloading the 10.2.0.2 release, the version numbers for Oracle Data Mining and Oracle OLAP Analytic Workspace remain at 10.2.0.3. These issues are tracked with Oracle bugs 4758695 and 4751917.

    
    

11. Start the database and recompile the remaining invalid objects:

    SQL> STARTUP
    SQL> @ORACLE_BASE\ORACLE_HOME\rdbms\admin\utlrp.sql
    
    

12. Set the CLUSTER_DATABASE initialization parameter to TRUE:

    
    

    Note: If the system uses an initialization parameter file, change the value of the CLUSTER_DATABASE initialization parameters to TRUE in the initialization parameter file (initsid.ora).

    
    

    SQL> ALTER SYSTEM SET CLUSTER_DATABASE=TRUE SCOPE=spfile;
    SQL> SHUTDOWN
    SQL> EXIT
    
    

13. Start the listener, gsd and ons processes:

    1. Start the node applications by entering the following command, where node is the name of the node where the applications are running:

       C:\> ORACLE_BASE\ORACLE_HOME\bin\srvctl start nodeapps -n node
       
       

       For Automatic Storage Management instances, perform the following step:

       C:\> ORACLE_BASE\ORACLE_HOME\bin\srvctl start asm -n node
       
       

    2. Start the service OracleServiceinst_name from the Control Panel by selecting Administrative Tools and then Services.

    
    

    Note: Option for removing Oracle Clusterware patch set is not available in 10.2.0.3. This issue is tracked with Oracle bug 5598010.

    
    

10.1 Oracle Clusterware Shutdown Stops and Then Restarts EVMD

While issuing crsctl stop crs sequentially on each node, the EVMD service may restart after Oracle Cluster Synchronization Services was stopped.

This isssue is tracked with Oracle bug 4951464.

Workaround:

Perform either of the following:

• Wait for 10 minutes

• Give the following command:

  net stop oracleevmservice
  
  

10.2 Error While Deleting a Cluster Database Instance on a Shared Oracle Home on Windows

When you delete a cluster database instance on a shared Oracle Home on Windows, there may be an error message saying that EMCA has failed. This occurs when Database Control has been configured to manage the associated cluster database.

This issue is tracked with Oracle bug 4757173.

Workaround:

If this error occurs, you must determine if Database Control was running from the node from which the instance was deleted.

For example, assume a two-node cluster with nodes, node1 and node2, where orcl1 is an instance running on node1, while orcl2 is running on node2. If Database Control is running on node2, and orcl2 is the instance deleted, then you must do the following:

1. Run ORACLE_BASE\ORACLE_HOME\bin\emca using the -reconfig dbcontrol db options to reconfigure the Database Control deployment.

2. Manually delete the state directories ORACLE_BASE\ORACLE_HOME\node2_orcl2 and ORACLE_BASE\ORACLE_HOME\oc4j\j2ee\OC4J_DBConsole_node2_orcl2

If Database Control was running on node1 instead, and orcl2 was the instance deleted, then:

• You can safely ignore the error.

• You can manually delete the state directories ORACLE_BASE\ORACLE_HOME\node2_orcl2 and ORACLE_BASE\ORACLE_HOME\oc4j\j2ee\OC4J_DBConsole_node2_orcl2, if possible.

10.3 Upgrading Preconfigured Standard Edition Databases

If you are using an Oracle Database release 10.2.0.2 preconfigured Standard Edition database, then the database contains the following components that are not supported by the Standard Edition:

• Oracle Data Mining

• Oracle OLAP Catalog

• Oracle OLAP Analytic Workspace

• Oracle OLAP API

The catupgrd.sql script does not run the patch scripts for these components. It sets the component STATUS to OPTION OFF in the DBA_REGISTRY view. The original versions of the dictionary objects for the components remain in the database but the catupgrd.sql script does not apply the patch to them.

10.4 Database Upgrade Error While Running Oracle Database Upgrade Assistant

Oracle Database Upgrade Assistant fails to upgrade 10.2.0.1 or 10.2.0.2 Oracle RAC database to 10.2.0.3 patch set if the Oracle RAC database is down.

Workaround

Enter the following command to start the database and retry the upgrade process:

C:\> ORACLE_BASE\ORACLE_HOME\bin\srvctl start database -d db_name -o mount

This issue is tracked with Oracle bug 5533745.

10.5 Upgrade of Oracle Enterprise Manager Not Supported

When you start Oracle Database Upgrade Assistant to upgrade 10.2.0.1 or 10.2.0.2 Oracle RAC database to 10.2.0.3 patch set, the following error occurs:

Inplace upgrade of Enterprise Manager is not supported for RAC databases is 
not supported in this release

Workaround:

Ignore warning.

This issue is tracked with Oracle bug 5534298.

10.6 Running Oracle Database Upgrade Assistant on a Single Node Oracle RAC Installation

Oracle Database Upgrade Assistant fails during postinstallation upgrade on a single node and displays the following error message:

ORA-32001: write SPFILE requested but no SPFILE specified at startup

Workaround:

1. Start the database with the NOMOUNT option:

   C:\> sqlplus /NOLOG
   SQL> CONNECT SYS/SYS_password AS SYSDBA
   SQL> STARTUP NOMOUNT
   
   

2. Create SPFILE from PFILE by entering the following command where pfile_location is usually ORACLE_BASE\ORACLE_HOME\admin\db_name\pfile for Optimal Felxible Architecture compliant database:

   SQL> CREATE SPFILE=SHARED_LOCATION\SPFILE.ORA FROM PFILE=pfile_location\init.ora
   
   

3. Shut down the database:

   SQL> SHUTDOWN IMMEDIATE
   SQL> EXIT
   
   

4. Clear the contents of PFILE located at ORACLE_BASE\ORACLE_HOME\dbs\initsid.ora and set the value of SPFILE as SHARED_LOCATION\SPFILE.ORA.

5. Run Oracle Database Upgrade Assistant:

   C:\> ORACLE_BASE\ORACLE_HOME\bin\dbua
   

This issue is tracked with Oracle bug 5531643.

10.7 Oracle OLAP Users Unable to Connect to Oracle Warehouse Builder

Oracle OLAP users will not be able to connect to Oracle Warehouse Builder immediately after upgrading their database to 10.2.0.3.0.

Workaround:

Download patch 5548282 from OracleMetalink or Oracle Technology Network.

This issue is tracked with Oracle bug 5548282.

10.8 Oracle Database Upgrade Assistant Summary Page Shows Incorrect Version Information

When you use Oracle Database Upgrade Assistant to upgrade the Automatic Storage Management instance of the 10.2.0.1 or 10.2.0.2 database to 10.2.0.3 patch set, the pre-upgrade summary page and the upgrade results page display the earlier 10.2 version of the Automatic Storage Management instance even if the upgrade is successful.

This issue is tracked with Oracle bug 5510050.

10.9 Memory Access Mode not Supported in Oracle RAC

In an Oracle RAC set up, if database instances are not running on the same node as Oracle Enterprise Manager Management Service, then monitoring in memory access mode does not display performance chart data.

This issue is tracked with Oracle bug 5559618.

10.10 VIP May Relocate to the Last Node While Upgrading Oracle Clusterware

Rolling upgrade or upgrade of Oracle Clusterware to 10.2.0.3 patch set may cause VIP to move to the last node.

Workaround:

Enter the following command to relocate the VIP to the preferred node:

crs_relocate VIP Resource

This issue is tracked with Oracle bug 5673067.