Oracle Database 11g Release 2 cluster-enabled dbconsole

You may have a small RAC database that is managed with the dbconsole rather than the grid. The documentation states that the console must be started manually on the applicable cluster node(s) whenever it is required. Couldn't the CRS cluster start and stop the dbconsole automatically?

The following setup is assumed.

• A three node RAC cluster.
• Grid infrastructure installed as user oragrid under /opt/oracle/app/11.2.0/grid.
• Oracle Database installed as user oracle under /opt/oracle/app/oracle/product/11.2.0/dbhome_1.
• The database Oracle Home is mounted at /opt/oracle/app/oracle using an ASM clustered filesystem.
• The unique name of the database to be monitored is racd3.
• The dbconsole (or agent) will be available to be started on all nodes.

As oracle (the user to run the dbconsole), create a new script /opt/oracle/app/11.2.0/grid/crs/public/local.agent.racd3_actionScript.sh.

#!/bin/sh
set -e
PATH=/usr/local/bin:/usr/bin:/bin
ORACLE_UNQNAME=racd3; export ORACLE_UNQNAME
ORACLE_SID=$ORACLE_UNQNAME
ORAENV_ASK=NO
. oraenv
unset ORACLE_SID

case "$1" in
  start)
    $ORACLE_HOME/bin/emctl start dbconsole
    ;;
  stop)
    $ORACLE_HOME/bin/emctl stop dbconsole
    ;;
  check)
    $ORACLE_HOME/bin/emctl status dbconsole
    ;;
  *)
    exit 1
    ;;
esac

exit 0

Now use this 'action script' as a dbconsole wrapper for a new cluster managed resource named local.agent.racd3.

crsctl add resource local.agent.racd3 -type local_resource -attr \
" ACTION_SCRIPT=/opt/oracle/app/11.2.0/grid/crs/public/local.agent.racd3_actionScript.sh, \
DESCRIPTION=Oracle agent for racd3 database, DEGREE=1, ENABLED=1, \
AUTO_START=restore, START_TIMEOUT=0, UPTIME_THRESHOLD=1h, CHECK_INTERVAL=60, \
STOP_TIMEOUT=0, SCRIPT_TIMEOUT=300, RESTART_ATTEMPTS=3, OFFLINE_CHECK_INTERVAL=60, \
START_DEPENDENCIES=hard(ora.c03vol1.vorabase.acfs), \
STOP_DEPENDENCIES=hard(ora.c03vol1.vorabase.acfs)"

The above will copy the action script to all cluster nodes. Being a local resource, it will start on each node. It has a start and stop hard dependency on the Oracle Home's filesystem (the executables in this filesystem are not available immediately upon reboot, but appear later once the cluster has initialised and the cluster filesystem has been mounted). Upon reboot, the dbconsole will be restarted if it was running when the cluster was shut down.

Check the status of this new resource.

[root@h01-c03-test ~]# /opt/oracle/app/11.2.0/grid/bin/crsctl status resource local.agent.racd3
NAME=local.agent.racd3
TYPE=local_resource
TARGET=OFFLINE, OFFLINE, OFFLINE
STATE=OFFLINE, OFFLINE, OFFLINE

Start the resource and recheck the status.

[root@h01-c03-test ~]# /opt/oracle/app/11.2.0/grid/bin/crsctl start resource local.agent.racd3
CRS-2672: Attempting to start 'local.agent.racd3' on 'h02-c03-test'
CRS-2672: Attempting to start 'local.agent.racd3' on 'h01-c03-test'
CRS-2672: Attempting to start 'local.agent.racd3' on 'h03-c03-test'
CRS-2676: Start of 'local.agent.racd3' on 'h02-c03-test' succeeded
CRS-2676: Start of 'local.agent.racd3' on 'h03-c03-test' succeeded
CRS-2676: Start of 'local.agent.racd3' on 'h01-c03-test' succeeded
[root@h01-c03-test ~]# /opt/oracle/app/11.2.0/grid/bin/crsctl status resource local.agent.racd3
NAME=local.agent.racd3
TYPE=local_resource
TARGET=ONLINE                , ONLINE                , ONLINE
STATE=ONLINE on h01-c03-test, ONLINE on h02-c03-test, ONLINE on h03-c03-test

[root@h01-c03-test ~]# netstat -ant|grep 1158
tcp        0      0 :::1158                     :::*                        LISTEN