First of all, before applying the 10.2.0.4 patchset to the Oracle database you should verify if the DSTv4 (USA 2007) update included in the 10.2.0.4 patchset has implications on your database. Following the NoteId 553812.1 provided by Metalink the goal is to see if you have stored information that is affected and if you are already using a DST version higher then DSTv4 and what then to do.
Because I'm patching from a lower 10.2.0.X release to 10.2.0.4 I hav to execute the following query to check the current version of the Oracle time zone definitions:
SQL> SELECT version FROM v$timezone_file;
Becuse my current timezone version is lower then 4, I need to execute this other query to check if I'm storing *user* TZ (TSTZ and TSLTZ) data:
SQL> select c.owner || '.' || c.table_name || '(' || c.column_name || ') -' || c.data_type || ' ' col
from dba_tab_cols c, dba_objects o
where c.data_type like '%TIME ZONE'
and c.owner=o.owner
and c.table_name = o.object_name
and o.object_type = 'TABLE'
order by col
My output shows there is nothing outside the Data Dictionary (= other then SYS objects).
The following queries
SQL> SELECT object_name FROM dba_objects WHERE object_id IN (SELECT obj# FROM scheduler$_window);
SQL> SELECT object_name FROM dba_objects WHERE object_id IN (SELECT obj# FROM scheduler$_job);
show I have only 6 jobs defined and I have also no user TZ (TSTZ and TSLTZ) data, so there is no action to take for the Oracle time zone definitions, and I can simply upgrade from 10.2.0.1 to 10.2.0.4.
Now I have to stop one node because I need to patch first the clusterware software and it can be done using rolling updates.
Run the following commands from the first node of your RAC installation:
emctl stop dbconsole
isqlplusctl stop
srvctl stop service -d DWHDB -s DWHDB_TAF -i DWHDB1
srvctl stop instance -d DWHDB -i DWHDB1 -o immediate
srvctl stop asm -n bl3306
srvctl stop nodeapps -n bl3306
Go into the directory Disk1 and run ./runInstaller &
A Welcome screen appears.
Specify the home details for the clusterware home.
Click next on the cluster nodes screen.
It checks for product-specific prerequisites.
Click Install
And the Clusterware patching is starting.
Before click the EXIT button and complete the installation,login as root on the first node and execute the following command as suggested:
Run
/u01/app/oracle/product/10.2.0/crs_1/crsctl stop crs
/u01/app/oracle/product/10.2.0/crs_1/install/root102.sh
They stop and start the crs services on the first node.
Execute the same command on every nodes, I ru them only on my second node:
/u01/app/oracle/product/10.2.0/crs_1/crsctl stop crs
/u01/app/oracle/product/10.2.0/crs_1/install/root102.sh
Then execute as oracle user from one node:
crsctl query crs activeversion
Execute as oracle user from second node:
crsctl query crs activeversion
The clusterware software has been updated.
Click EXIT and then YES from the OUI interface.
Now execute the following command to completely stop the database (this patchset doesn't support rolling updates) from your first node:
emctl stop dbconsole
isqlplusctl stop
srvctl stop service -d database DWHDB -s DWHDB_TAF -i DWHDB1
srvctl stop database -d DWHDB -o immediate
srvctl stop asm -n bl3306
srvctl stop nodeapps -n bl3306
Execute the following command to completely stop the asm instance from your second node:
emctl stop dbconsole
isqlplusctl stop
srvctl stop asm -n bl3305
srvctl stop nodeapps -n bl3305
As oracle user run again the runInstaller
Click NEXT on Welcome screen.
Choose the Oracle home.
Click NEXT on cluster installation mode screen.
Click NEXT on prerequisite checks screen.
Enable and register your Oracle Configuration Manager.
Accept the license agreement
and test your registration
Click INSTALL on summary screen.
The installation is proceeding.
When asked execute as root user the following command on your first node
/u01/app/oracle/product/10.2.0/db_1/root.sh
and then again from your second node.
Clic EXIT on End of Installation screen
and the YES
Only from your first node, login as oracle user and execute:
srvctl start asm -n bl3306
srvctl start listener -n bl3306
Then run dbua
Click NEXT on Welcome screen.
Select Upgrade a Database and click NEXT.
Select your database and click NEXT, providing a user with sysdba privileges.
The upgrade of the database starts.
Choose the degree of parallelism to recompile any invalid objects, turn off archiving and click NEXT.
Specify your Flash Recovery Area and click NEXT.
Click FINISH on Summary screen.
Your database is now 10.2.0.4 patched.