OEM – Grid on RAC – Configuration
Install 11.2 RAC / ASM software on both nodes
Install 10.2.0.1 db software
Upgrade db software to 10.2.0.4
Create a RAC 10.2.0.4 database called emrep
CONFIGURE Red Hat 5.3 for Oracle on both nodes
Install required rpm’s:
 libXmu-1.0.2-5.i386.rpmÂ
 libXmu-1.0.2-5.x86_64.rpmÂ
 openmotif-2.3.1-2.el5.i386.rpmÂ
 openmotif-2.3.1-2.el5.x86_64.rpm
Configure library symbolic links:
 ln -s /usr/lib/libgdbm.so /usr/lib/libdb.so.2
 ln -s /usr/lib/libgdbm.so.2 /usr/lib/libdb.so.2
PREPARE FORÂ OEM GC ON NODE ONE
SQL>
Check NLS and invalid procedures
 connect sysman/pwd
 select value from nls_database_parameters where parameter = ‘NLS_LENGTH_SEMANTICS’
 exec emd_maintenance.analyze_emd_schema(‘SYSMAN’);
 connect / as sysdba
 select object_name, object_type from all_objects where owner = ‘SYSMAN’ and status ‘VALID’;
 select object_name, object_type from all_objects where status ‘VALID’ and object_name like ‘DBMS%’;
Check URL’s
 set serveroutput on size 4000
 set lines 255
 spool run_after_upgrade
 begin
 for rec in (select target_type, property_value from mgmt_type_properties where target_type in (
 ‘RHreposURL’,’SusereposURL’,’SolarisreposURL’,’WindowsreposURL’,’PatchCrossRef’))
 loop
 begin
 dbms_output.put_line(‘INSERT INTO MGMT_PATCHING_PROPERTIES(PROPERTY_TYPE, PROPERTY_NAME,Â
 PROPERTY_VALUE) values
  (”os_patching_repos_url”,”||REC.TARGET_TYPE||”,”||rec.PROPERTY_VALUE||”);’);
 end;
 end loop;
 dbms_output.put_line(‘COMMIT;’);
 end;
 /
 spool off
Clean up jobs
 connect sysman/pwd
 execute emd_maintenance.remove_em_dbms_jobs;
 commit;
 exit
srvctl stop database -d emrep
srvctl start database -d emrep
 select a.instance_name as sid, b.value as jobqueue from gv$instance a, gv$parameter b where a.inst_id = b.inst_id
 and b.name = ‘job_queue_processes’;
 (returns 10)
 alter system set job_queue_processes = 0 sid=’*’ scope=both;
 select count(*) from dba_jobs_running;
  exec dbms_scheduler.set_scheduler_attributes(‘SCHEDULER_DISABLED’,’TRUE’);
 select count(*) from dba_scheduler_running_jobs where slave_process_id is not null;
STOP EMREP DATABASE ON NODE 2
INSTALLÂ OEM GC 10.2.0.3 ON NODE 1
UPGRADEÂ OEM GC to 10.2.0.5 ON NODE 1
 set oracle environment variables to OMS10G
 cd 3731593/Disk1
 ./runInstaller
 answer all questions and wait about an hour.
 set oracle environment variables to AGENT10G
 ./runInstaller
 answer all questions and wait about an hour.
 startup and test grid control.
ON NODE TWO
Install 10.2.0.3 GC as software only install “Additional Management Service” from the installer screen.
Upgrade to 10.2.0.5
change $OMS_HOME/sysman/config/emoms.properties on both hosts
oracle.sysman.eml.mntr.emdRepConnectDescriptor=
(DESCRIPTION\=(LOAD_BALANCE\=on)(FAILOVER\=on)
 (ADDRESS_LIST\=
 (ADDRESS\=(PROTOCOL\=TCP)(HOST\=node1_vip.fcdev.local)(PORT\=1521))
 (ADDRESS\=(PROTOCOL\=TCP)(HOST\=node2_vip.fcdev.local)(PORT\=1521))
 )
 (CONNECT_DATA\=(SERVICE_NAME\=emrep))(FAILOVER_MODE\=(TYPE\=select)(METHOD\=basic))
)
*****************************Then run the following to change the monitoring configuration
*****************************used for the Management services and repository target.
 emctl config emrep -conn_desc “emrep”
Issue the following command from the management server home
 emctl secure oms
Modify $ORACLE_HOME/Apache/Apache/conf/ssl.conf to use port 443
export the configuration file from the first OMS
emctl exportconfig oms
ON NODE TWO
copy exportconfig to 2’nd node
emctl importconfig oms -file
emctl start oms
emcli setup -url=https://dev1orc01.fcdev.local:4444/em -username=sysman -password=pwd -nocertvalidate
emctl secure agent -emdWalletSrcUrl https://node2.fcdev.local:1159/em
ON NODE 1
emctl secure agent -emdWalletSrcUrl https://node1.fcdev.local:1159/em
ON BOTH NODES:
emctl config oms loader -shared yes -dir /u01/app/oracle/acfsmounts/fra_oms2/oms10g/SHAREDLOADER
where the path above is available and shared between both nodes; this is why we have used 11.2 and Oracle ACFS to created a shared directory.
OEM GC SHOULD NOW BE WORKING FROM BOTH NODES IN THE CLUSTER.
Discussion ¬