10.2.0.4 to 10.2.0.5 installation of Patch Set Release (Windows)

Environment: 10.2.0.4 Cluster database zlm10g (dual node, ZLM10G1,ZLM10G2)

System: Windows 2003 Server 64Bit

Memory: 2G RAM

Storage: ASM

Goal: Upgrade the cluster database from 10.2.0.4 to 10.2.0.5, including cluster software and OEM patch installation

First, pre-upgrade preparation:

1. See if an invalid object exists

Sql> Select Object_name,status from Dba_objects where

status= ' INVALID ';

The purpose is to determine whether these invalid objects were generated by the upgrade process or if they existed before the upgrade.

2. Check the time zone information in the database

Sql> select version from V$timezone_file;

2.1 Return value is 4

In this case, if you have TZ information in the database, no additional action is required in the upgrade

2.2 Return value greater than 4

Please refer to [ID 553812.1] Documentation

2.3 If the return value is less than 4

Perform the following steps

@?/rdbms/admin/utltzpv4.sql;

SELECT * from Sys.sys_tzuv2_temptab;

If no rows are returned, no additional action is required during the upgrade process

If you return specific column information that contains TZ, the upgrade operation affects the TZ information

If the SYS user has an scheduler object, it can be ignored during the upgrade process, but if there is user data or a user-created job, you need to back up the data before upgrading, remove all user-created jobs after the upgrade, and then rebuild

Two ways to upgrade:

Rolling upgrade (no down time)

Non-rolling upgrade (with down time)

Second, upgrade the cluster database software

1. First, prepare the 10.2.0.5 patch media:

P8202632_10205_mswin-x86-64

2. Stop Oracle-related services and processes for all nodes

2.1 Stop CRS Resources

Emctlstop Dbconsole

Isqlplusctlstop

Lsnrctlstop

Srvctlstop service-d zlm10g

Srvctlstop databse-d zlm10g

Srvctlstop asm-n Rac1

Srvctlstop Asm-n RAC2

Srvctlstop listener-n Rac1

Srvctlstop Listener-n RAC2

Srvctlstop nodeapps-n Rac1

Srvctlstop Nodeapps-n RAC2

CRS_STAT-T Check if cluster resources are all closed

Srvctlstop CRS

2.2 Stop Service

Run services.msc to see if all Oracle-related services have been closed, plus

A distributed Transactioncoordinator service to stop.

2.3 Stopping a process

See if there are any Oracle-related processes running in Task Manager, all closed, such as: One.exe

3. Back up the following directories (copy to other paths):

%oracle_crs_home%

%oracle_home%

%oracle_inventory%

Note: It is recommended that the database be first RMAN Backup

4. Install the cluster software patch

During installation, the path selection overrides the original path installation, which is the Crs_1 directory

After the software is installed, run Patch102.bat in Node 1 and Node 2 respectively to complete the actual installation process, after running the script, the system will automatically run CRS related services

5. Check that the cluster patch is installed correctly

Crsctlquery CRS Softwareversion

Crsctlquery CRS Activeversion

If the cluster patch is properly installed, the output of these 2 commands should be 10.2.0.5.0

Third, database software patch installation

1. Install the media and upgrade the Clusterware using the same:

P8202632_10205_mswin-x86-64

2. Turn off Oracle-related services and processes

2.1 Stop CRS Resources

Emctlstop Dbconsole

Isqlplusctlstop

Lsnrctlstop

Srvctlstop service-d zlm10g

Srvctlstop databse-d zlm10g

Srvctlstop asm-n Rac1

Srvctlstop Asm-n RAC2

Srvctlstop listener-n Rac1

Srvctlstop Listener-n RAC2

2.2 Stop Service

Run services.msc to see if all Oracle-related services have been closed, plus

A distributed Transactioncoordinator service to stop.

2.3 Stopping a process

See if there are any Oracle-related processes running in Task Manager, all closed, such as: One.exe

3. Install patches for Oracle database software

Installation path Select the same directory where the database software was installed before the Db_1 directory

4. Modifying cluster database parameters

After loading, log into the system with SYSDBA, modify the Cluster_database parameter to False

Sqlplus/nolog

Sql>alter system set cluster_database=falsescope=spfile;

Sql>shutdown Immediate

5. Run the upgrade script on Node 1 to upgrade the database

5.1 Enter upgrade mode:

Sql>startup Upgrade

5.2 Pre-check if the upgrade meets the criteria:

Sql>spool Patch_info.log

Sql> @?\rdbms\admin\utlu102i.sql

Sql>spool off

About preflight This script, the official website for each version, there are more supplementary scripts, will provide updated, more detailed preflight content, details can be consulted [ID 884522.1]

5.3 Start the formal upgrade script:

Sql>spool Patch.log

Sql> @?\rdbms\admin\utlupgrd.sql

10GR2 the process lasts about 30 minutes and 11gr2 takes about 50 minutes

5.4 Invalid objects resulting from compilation of the upgrade

Sql>spool Rp.log

Sql> @?\rdbms\admin\utlrp.sql

Sql>spool off

5.5 Check that the database components are valid after upgrade

Sql> select Comp_id,comp_name,status from Sys.dba_registry;

If the upgrade is complete properly, the value of the Status column for all components should be valid

Iv. installing with Opatch 10.2.0.5 Patch After

1. Media Preparation and description

p8350262_10205_generic (OEM patch )

P12429524_10205_mswin-x86-64

Here to illustrate, hit the main purpose of the patch is to be able to use the normal OEM database management, due to the OEM on 10.2.0.4 and 10.2.0.5 on the normal operation there is a bug, see: [ID1222603.1]

If you do not hit the OEM patch, then when you perform DBCA,DBUA,EMCA configuration OEM, you will be prompted to fail to configure EM, start database control failed because the root certificate CA fails in 31-dec-2010 and cannot communicate using the SSL protocol. resulting in a failure to start, the bug is limited to the 2 versions of PSR above

Here the 2nd media,p12429524_10205_mswin-x86-64 contains more fixed bugs, so if you install 12429524 of this PSU, then you can also do not install 8350262, choose one can

2. Opatch Installation

For all of Oracle's interim patches, which we say are often referred to as small patches (PSUs), can be installed with Opatch, which is a patch installation script provided by Oracle, which is simple to use, and the catalogue is placed in oracle_home\. Opatch

Below, the Windows platform executable file for Opatch.bat, the need to install the patch extracted to a directory, and then perform "Opatch Apply + Unzip directory path" on it, you can also enter the extracted from the list of numbers, directly execute Opatch Apply to install the patch, if you want to see the database before the patches, you can execute Opatch lsinventory[-detail], specific commands can be added-help lookup

3. Reconfigure the OEM

After the patch is installed, if you go directly to start dbconsole or you will encounter an inability to start, it is necessary to reconfigure it, mainly 3 steps:

3.1 Node 1 execution

Emctlsecure Dbconsole-reset

3.2 Node 2 execution

Emctlsecure Dbconsole

Note that the command is not with the-reset parameter

3.3 Node 1 (or Node 2) Start Dbconsole

Emctlstart Dbconsole

V. Closure

At this point, all upgrade work is completed successfully, check whether there are missing Oracle services and processes are not open, if any, all open

Services.msc

Crs_stat-t

Emctlstart Dbconsole

Isqlplusctlstart

Lsnrctlstart

Srvctlstart service-d zlm10g

Srvctlstart database-d zlm10g

Srvctlstart instance-d zlm10g-i ZLM10G1

Srvctlstart instance-d zlm10g-i ZLM10G2

Srvctlstart listener-n Rac1

Srvctlstart Listener-n RAC2

Srvctlstart asm-n Rac1

Srvctlstart Asm-n RAC2

Srvctlstart nodeapps-n Rac1

Srvctlstart Nodeapps-n RAC2

The above command is optional, not all need to be executed again, if the CRS resources are online, then there is no need to execute the