Steps to Migrate the R12.2 instance from OEL 6 to OEL 7 server

on
STEP 1: Check pre-requisites

Once the server is ready, make sure to check the pre-requisites mentioned in the Oracle Document(1330701.1). 
  • Check the list of RPMs that need to be installed.                          
           We can check whether the rpm is installed or not using the below command. Replace libgcc with the rpm you are searching for.    

$ rpm -qa --qf "%{n}-%{v}-%{r}.%{arch}\n" | grep libgcc

  • Update the kernel parameter values in /etc/sysctl.conf. Values suggested in the document are minimum, you can set a value more than what is suggested in the document. It should not be less than the value mentioned in the document. In our case, we have installed 19C Grid for ASM which requires a few of the kernel parameter values to be more than what is suggested in this documents.  So it's always better to keep the highest value in the /etc/sysctl.conf file and after updating all the values in the file. Please use sysctl -p to restart the system to invoke the new settings.
  • Make sure host entries are added in the /etc/hosts file.
STEP 2: Check the mounts

Make sure all the required mount points are mounted in the server. 

STEP 3: Install 19c Grid ASM

Install 19c Grid ASM. If I add the Step to install 19c Grid ASM here this post would be too big. So I will publish installation steps as different blog posts.

STEP 4:Run preclone on source DB

The plan here is to copy the Oracle database and application binary from the source instance to the target instance.
Since we are copying the Oracle binaries it is important to run preclone on both database and application. Log in to the source database server and source env then run adpreclone.pl. 

cd $ORACLE_HOME/appsutil/scripts/$CONTEXT_NAME
    
perl adpreclone.pl dbTier

STEP 5: Run preclone on the source application
Login to source application server and source the run file system env then run adpreclone.pl. 

. EBSapps.env run

cd $ADMIN_SCRIPTS_HOME

perl adpreclone.pl appsTier

STEP 6: Take RMAN backup

We didn't have the privilege of doing the snap clone for the 4 TB database. We only have a shared mount between two data centers and RMAN. Some might say that taking a compressed backup will take more time cause it uses the CPU overhead to compress the backup. The time taken to compress the backup is less compared to the time taken to place the 4TB RMAN backup in the NFS mount. The compressed backup almost half the time as the null compression backup and also the backup size was reduced to 800GB.
rman target / msglog=L0_backup.log >>EOF
RUN {
  CONFIGURE CONTROLFILE AUTOBACKUP ON;
  CONFIGURE CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK TO '/backup/TEST01/rman/TEST01_autobcf_%F';
  # allocate channels here
  allocate channel D1 device type disk ;
  allocate channel D2 device type disk ;
  allocate channel D3 device type disk ;
  allocate channel D4 device type disk ;
  allocate channel D5 device type disk ;
  allocate channel D6 device type disk ;
  backup  as compressed backupset incremental level 0
      format '/backup/TEST01/rman/TEST01_db_l0_s%s_p%p_t%t' (database) PLUS ARCHIVELOG ;
       backup tag hot_spf_bkup
      format '/backup/TEST01/rman/TEST01_spfile_s%s_p%p' (spfile);
   release channel D1;
   release channel D2;
   release channel D3;
   release channel D4;
   release channel D5;
   release channel D6;
   }
EOF
Once the RMAN completes make sure to give chmod 777 permission to RMAN backup files.

STEP 6: Take TAR backup of oracle binaries in SOURCE

After starting the RMAN instead of waiting for it to complete. We can take the tar of backup of database oracle home. 

cd /backup/TEST01

nohup tar -C /wwe/test01/db/ -cvzf TEST01_DB_HOME.tar.gz 12.1.0 &

While taking the tarball of application top it good takes only EBSapps in the run file system of the source instance. This will save both time and disk space.

cd /backup/TEST01

nohup tar  -C /wwe/test01/apps/fs1/ -cvzf TEST01_APP.tar.gz EBSapps &

Once the tar completes make sure to give chmod 777 permission to tarball files.

STEP 6:  UNTAR backup in TARGET

Log in to the database server and run the untar. We had kept the directory structure the same as the source instance.

cd /wwe/test01/db/

nohup tar -xvzf /backup/TEST01/TEST01_DB_HOME.tar.gz &

Log in to the application server and run the untar.

cd /wwe/test01/apps/fs1/

nohup tar -xvzf /backup/TEST01/TEST01_APP.tar.gz &
STEP 7: Restore the RMAN backup in the Target database

Update the pfile in the target database for the RMAN restore.

db_unique_name=TEST01_NEWHOSTDB301

db_create_file_dest='+TEST01_DATA'

db_file_name_convert='+TEST01_DATA/TEST01/DATAFILE/','+ORA_DATA/TEST01_NEWHOSTDB301/DATAFILE/'

log_file_name_convert='+TEST01_DATA/TEST01/onlinelog/','+ORA_DATA/TEST01_NEWHOSTDB301/ONLINELOG/'

We have updated the environment file with the new hostname and update the database $CONTEXT_FILE with the new values. Source the updated environment file and run the RMAN restore script. 
rman target / msglog=L0_restor.log >>EOF
rman auxiliary / 
run
{
allocate auxiliary channel c1 type disk;
allocate auxiliary channel c2 type disk;
allocate auxiliary channel c3 type disk;
allocate auxiliary channel c4 type disk;
allocate auxiliary channel c5 type disk;
allocate auxiliary channel c6 type disk;
allocate auxiliary channel c7 type disk;
allocate auxiliary channel c8 type disk;
duplicate database to 'TEST01' backup location '/backup/TEST01/rman/' NOFILENAMECHECK;
  }
EOF
Once the RMAN restore completes. Create temp data files in the new instance. 

CREATE TEMPORARY TABLESPACE TEMP TEMPFILE '+ORA_DATA' SIZE 100M AUTOEXTEND ON NEXT 100M MAXSIZE UNLIMITED;

CREATE TEMPORARY TABLESPACE TEMP1 TEMPFILE '+ORA_DATA' SIZE 100M AUTOEXTEND ON NEXT 100M MAXSIZE UNLIMITED;

CREATE TEMPORARY TABLESPACE TEMP2 TEMPFILE '+ORA_DATA' SIZE 100M AUTOEXTEND ON NEXT 100M MAXSIZE UNLIMITED;

CREATE TEMPORARY TABLESPACE TEMP3 TEMPFILE '+ORA_DATA' SIZE 100M AUTOEXTEND ON NEXT 100M MAXSIZE UNLIMITED;

ALTER DATABASE DEFAULT TEMPORARY TABLESPACE TEMP;
Run the library update script in the database as sysdba 

sqlplus / as sysdba

@$ORACLE_HOME/adupdlib.sql so
STEP 8: Cleanup FND_NODES and run autoconfig

Database in the source instance will still be pointing to the old servers. We need to clean the FND_NODES value and run autoconfig to populate the values in the table.

sqlplus apps

EXEC FND_CONC_CLONE.SETUP_CLEAN;

truncate table adop_valid_nodes;

Commit;
Make sure the listener is up and running before starting autoconfig in the source database.

cd $ORACLE_HOME/appsutil/scripts/$CONTEXT_NAME

./adautocfg.sh
Since we have updated the database $CONTEXT_FILE we have run just the autoconfig to apply the updated values instead of running the adcfgclone.pl in the database.

STEP 9: Configure Application Node

Since we didn't copy the INST while taring in the source application server, we need to manually copy the $CONTEXT_FILE from the source application server to the target application server. Once the $CONTEXT_FILE is copied, update the context_file with the new hostnames and URL(if the new server has a different URL). Once the adcfgclone completes, run autoconfig in both the run file system and patch file system

cd /wwi/test01/apps/fs1/EBSapps/comn/clone/bin

perl adcfgclone.pl appstier dualfs $CONTEXT_FILE
Once the adcfgclone completes, run autoconfig in both the run file system and patch file system.

. EBSAPPS.env RUN

cd $ADMIN_SCRIPTS_HOME

./adautocfg.sh
Before running the autoconfig in the patch file system we need to disable the ebs_logon trigger otherwise the autoconfig for the patch file system would fail. 

. EBSAPPS.env PATCH

sqlplus system

alter trigger ebs_logon disable;
After disabling the ebs_logon trigger, run the autoconfig in the patch file system. 

cd $ADMIN_SCRIPTS_HOME

./adautocfg.sh
Once the autoconfig for the patch file system completes successfully make sure to enable the ebs_logon triggers back. 

sqlplus system

alter trigger ebs_logon enable;
STEP 10: Delete the Orphan value

 Run the below delete statement to delete the Orphan values in the database.
delete from fnd_profile_option_values
where
  (level_id = 10005
   and level_value > 0
   and level_value not in (select node_id from fnd_nodes))
  or
  (level_id = 10007
   and level_value2 > 0
   and level_value_application_id=-1
   and level_value2 not in (select node_id from fnd_nodes));
/
commit;
STEP 11: Start the application service

Start the application service in the target instance. 

cd $ADMIN_SCRIPTS_HOME

echo $WEBLOGIC_PWD|sh adstrtal.sh apps/$APPS_PWD
STEP 12: Make sure no profile option is pointing SOURCE server

Use the below query to find the profile options which is still having the old host values.
set lines 999 pagesize 999
col name for a25
col USER_PROFILE_OPTION_NAME for a30
col PROFILE_OPTION_VALUE for a60
SELECT po.user_profile_option_name, 
       po.profile_option_name            name, 
       profile_option_value, 
       Decode(pov.level_id, 10001, 'Site', 
                            10002, 'Application', 
                            10003, 'Responsibility', 
                            10004, 'User', 
                            10005, 'Server', 
                            10006, 'Organization', 
                            10007, 'ServResp', 
                            'Undefined') LEVEL_SET 
FROM   apps.fnd_profile_option_values pov, 
       fnd_profile_options_vl po 
WHERE  Upper(profile_option_value) LIKE '%&OLD_HOST%' 
       AND pov.profile_option_id = po.profile_option_id;
STEP 13:Final Checks

Perform Sanity checks in the target instance and run fs_clone once everything fine.

Comments

Post a Comment