Hi all ,
Here the steps that I follow “when I have a big window of unavailability”:
1) Stop all stuff:
$ isqlplusctl stop
$ emctl stop dbconsole
$ srvctl stop database -d orcl -o immediate
$ srvctl stop nodeapps -n RAC1
$ srvctl stop nodeapps -n RAC2
# /u01/crs/oracle/product/10.2.0/bin/crsctl stop crs (as root user)
2) The most importante BACKUP:
Backup of all databases running on your RAC;
Backup of all Oracle Homes;
Backup of all CRS Homes;
Backup of all ASM Homes;
3) Open a X and proceed the binaries upgrade on $CRS_HOME on a node of your preference (The OUI copy the binaries to other nodes) :
./runInstaller and follow the OUI steps;
Before close the OUI, execute the script /u01/crs/oracle/product/10.2.0/install/root102.sh on all nodes
4) Proceed the binaries upgrade on $ORACLE_HOME
./runInstaller and follow the OUI steps;
Obs.: Until this moment the database was not upgraded, only binaries
5) Start the listeners on all nodes
$ srvctl start nodeapps -n RAC1
$ srvctl start nodeapps -n RAC2
6) Start DBUA to upgrade your database
dbua and and follow steps;
Obs.: If your are running more than one database repeat the step 6 (Upgrade ASM first)
7) Read logs from DBUA
8) Post upgrade verifications (s
8.1) show version of clusterware
$ crsctl query crs activeversion
8.2) Verify version of options on your database
set lin 200
col comp_name form a40
select comp_name, version, status
from sys.dba_registry;
If the catproc stays in invalid state follow the http://forums.oracle.com/forums/thread.jspa?messageID=2589633&tstart=0
After follow the steps above your version now is totally upgraded!!!
Cheers