How to upgrade Oracle with existing Harvest installation on Linux
search cancel

How to upgrade Oracle with existing Harvest installation on Linux


Article ID: 217067


Updated On:


CA Harvest Software Change Manager


Our current Harvest database is using Oracle 12r. What would I need to do if I repointed that dev server/brokers to an Oracle 18.0C database that was a copy of the Oracle 12 database back before the update? Is it as simple as running some sql file to update/create the new tables, or do we have to go through the entire process again?


Release : 14.0

Component : CA Harvest Software Change Manager


  1. Install the new Oracle Client version on the broker machine. Copy the tnsnames.ora file from the old Oracle Client’s network/admin folder to the new Oracle Client’s network/admin folder.  You’ll know the new Oracle Client is set up correctly when you can:

- Reset ORACLE_HOME to the new client’s home folder.  Example:
   export ORACLE_HOME=/app/oracle/product/19.3.0/client_1

- Login to the Harvest database.  Example:
   sqlplus [email protected]

- Select the contents of the “hartableinfo” table.  Example:
   select * from hartableinfo;

  1. Update the ORACLE_HOME setting in the CA SCM software owner’s .profile or .bash_profile.


  1. Stop all SCM-related processes before beginning the installation. To run the “bkrd” command you’ll need to be logged in as the CA SCM software owner’s userid.  Example:

su - cascm
bkrd -shutdown
pgrep rtserver | xargs kill
ps -ef | grep 'hserver\|bkrd\|rtserver'

  1. Navigate to the SCM software’s installation directory and execute the script. Example:

cd $CA_SCM_HOME/install
- Select option 3 “Change CA Harvest SCM installation configuration”
- Select option 3 “Keep Existing Authentication Settings and Continue”
- Review carefully the settings for ORACLE_HOME and ORACLE_SID.  If they are correct, select option 0 “No Change”
- Select option 4 “Accept the Settings and Continue”
- Type “n” to specify that you do NOT want to use CAcrypto

  1. Restart the broker, ensure that all the SCM processes are able to start and run a command line utility as a test. Example:

ps -ef | grep 'bkrd\|hserver\|rtserver'
./hgetusg -b $HOSTNAME -prompt -cu ; cat hgetusg.log