Use this procedure to downgrade Oracle Domain Services Cluster to Oracle Grid Infrastructure 12c Release 2 (12.2) after a successful upgrade.

1. As grid user, downgrade the Management Database to Oracle Grid Infrastructure 12c Release 2 (12.2):

   1. Manually copy the most recent time zone files from 19c Grid home to 12c Release 2 (12.2) Grid home, where timezlrg_number is the name of the most recent timzlrg file and timezone_number is the name of the most recent timezone file:

      $ cp $ORACLE_HOME/oracore/zoneinfo/timezlrg_number.dat /u01/app/12.2.0/grid/oracore/zoneinfo/timezlrg_number.dat
      $ cp $ORACLE_HOME/oracore/zoneinfo/timezone_number.dat /u01/app/12.2.0/grid/oracore/zoneinfo/timezone_number.dat

   2. Downgrade application schema using the following command syntax from 19c Grid home:

      $ $ORACLE_HOME/bin/mgmtua downgrade -local -oldOracleHome /u01/app/12.2.0/grid -skipSystemSchemaDowngrade

   3. Disable and stop MGMTDB resource from 19c Grid home:

      $ cd $ORACLE_HOME/bin
      $ ./srvctl disable mgmtdb
      $ ./srvctl stop mgmtdb
      

   4. Downgrade system schema using the following procedure:

      1. Set ORACLE_SID environment variable for 19c Grid home:

         $ export ORACLE_SID=-MGMTDB
         $ cd $ORACLE_HOME/bin

      2. Start CDB and all PDBs in downgrade mode:

         $ ./sqlplus / as sysdba
         SQL> startup downgrade
         SQL> alter pluggable database all open downgrade;
         SQL> exit
         

      3. Downgrade 19c Management Database using the following command syntax, where /u01/app/grid2 is the Oracle base for Oracle Grid Infrastructure 19c:

         $ $ORACLE_HOME/perl/bin/perl $ORACLE_HOME/rdbms/admin/catcon.pl -d 
         /u01/app/grid2 -e -l /u01/app/grid2/cfgtoollogs/mgmtua -b mgmtdowngrade -r 
         $ORACLE_HOME/rdbms/admin/catdwgrd.sql

      4. Set ORACLE_HOME and ORACLE_SID environment variables for 12c Release 2 (12.2) Grid home:

         $ export ORACLE_HOME=/u01/app/12.2.0/grid/
         $ export ORACLE_SID=-MGMTDB
         $ cd $ORACLE_HOME/bin

      5. Start CDB and all PDBs in upgrade mode:

         $ ./sqlplus / as sysdba
         SQL> shutdown immediate
         SQL> startup upgrade
         SQL> alter pluggable database all open upgrade;
         SQL> exit
         

      6. Run catrelod script for 12c Release 2 (12.2) Management Database using the following command syntax, where /u01/app/grid is the Oracle base for Oracle Grid Infrastructure 12c Release 2 (12.2):

         $ $ORACLE_HOME/perl/bin/perl $ORACLE_HOME/rdbms/admin/catcon.pl -d 
         /u01/app/grid -e -l /u01/app/grid/cfgtoollogs/mgmtua -b mgmtdowngrade 
         $ORACLE_HOME/rdbms/admin/catrelod.sql

      7. Recompile all invalid objects after downgrade using the following command syntax from 12c Release 2 (12.2) Grid home:

         $ $ORACLE_HOME/perl/bin/perl $ORACLE_HOME/rdbms/admin/catcon.pl -d 
         /u01/app/grid -e -l /u01/app/grid/cfgtoollogs/mgmtua -b mgmtdowngrade 
         $ORACLE_HOME/rdbms/admin/utlrp.sql

   5. Manually stop the Management Database:

      $ ./sqlplus / as sysdba
      SQL> shutdown immediate
      SQL> exit

2. As root user, use the following command syntax rootcrs.sh -downgrade from 19c Grid home to downgrade Oracle Grid Infrastructure on all nodes, in any sequence. For example:

   # $ORACLE_HOME/crs/install/rootcrs.sh -downgrade
   

   Run this command from a directory that has write permissions for the Oracle Grid Infrastructure installation user. You can run the downgrade script in parallel on all cluster nodes, but one.

3. As root user, downgrade the last node after you downgrade all other nodes:

   # $ORACLE_HOME/crs/install/rootcrs.sh -downgrade

4. As grid user, remove Oracle Grid Infrastructure 19c Grid home as the active Oracle Clusterware home:

   1. On any of the cluster member nodes where the rootupgrade.sh script has run successfully, log in as the Oracle Grid Infrastructure installation owner.

   2. Use the following command to start the installer, where /u01/app/19.0.0/grid is the location of the new (upgraded) Grid home:

      $ cd $ORACLE_HOME/oui/bin
      ./runInstaller -nowait -waitforcompletion -ignoreSysPrereqs
      -updateNodeList -silent CRS=false 
      ORACLE_HOME=/u01/app/19.0.0/grid
      "CLUSTER_NODES=node1,node2,node3" 
       -doNotUpdateNodeList 

      Add the flag -cfs if the Grid home is a shared home.

5. As root user, start the 12c Release 2 (12.2) Oracle Clusterware stack on all nodes.

   # crsctl start crs

6. As grid user, set Oracle Grid Infrastructure 12c Release 2 (12.2) Grid home as the active Oracle Clusterware home:

   1. On any of the cluster member nodes where the rootupgrade script has run successfully, log in as the Oracle Grid Infrastructure installation owner.

   2. Use the following command to start the installer, where the path you provide for ORACLE_HOME is the location of the home directory from the earlier Oracle Clusterware installation.

      $ cd $ORACLE_HOME/oui/bin
      $ ./runInstaller -nowait -waitforcompletion -ignoreSysPrereqs 
      -updateNodeList -silent CRS=true 
      ORACLE_HOME=/u01/app/12.2.0/grid
      "CLUSTER_NODES=node1,node2,node3"

7. As grid user, downgrade CHA models from any node where the Grid Infrastructure stack is running from 12c Release 2 (12.2) Grid home and Management Database and ochad are up:

   $ ./srvctl stop cha
   $ ./chactl import model -file /u01/app/12.2.0/grid/cha/model/os_gold.svm -name DEFAULT_CLUSTER
   $ ./chactl import model -file /u01/app/12.2.0/grid/cha/model/db_gold.svm -name DEFAULT_DB
   $ ./srvctl start cha

   In the example above, DEFAULT_CLUSTER and DEFAULT_DB are function names that you must pass as values.