Clusterware 10g: what kicks off 3rd-party failover script?

Question

I am trying to use Oracle Clusterware 10g (installed on two identical Dell Linux boxes,
OEL 5.2) to failover a 3rd-party application, specifically websphere MQ. I have a ksh
script that is supposed to start/stop/restart the application.
Clusterware will fail over the IP address fine to the other box if I yank the primary box's
cable, but the app itself doesn't fail over. If I knew what (in Clusterware) did the
"kicking off" of the ksh script I could troubleshoot it more. I used the instructions in:
http://www.oracle.com/technology/products/database/clusterware/pdf/TWP-Oracle-Clusterware-3rd-party.pdf
and I'm pretty sure I did it right.
Any ideas?

Answer 1

Here is the action script:
#!/bin/ksh
LOG="/san1/var/mqm/config/Controlmq.log"
echo "script started at `date` from `hostname`" >> $LOG
############################### func_stopMQ ######################################
func_stopMQ () {
QUEUENAME=$1
while [[ `ps -ef | grep $QUEUENAME | grep -v grep | grep -v Control| awk {'print $2'} | wc -l` -gt 1 ]]
do
echo "DEBUG : `ps -ef | grep $QUEUENAME | grep -v grep | grep -v Control` " >> $LOG
echo "there are `ps -ef | grep $QUEUENAME | grep -v grep | grep -v Control| awk {'print $2'} | wc -l` processes running associated with $QUEUENAME on `hostname`" >> $LOG
endmqm $QUEUENAME
endmqlsr -m $QUEUENAME
## echo "sleeping for 10 seconds before checking again"
sleep 10
done
## echo "All set, there are no processes associated with $QUEUENAME "
echo "stopped $QUEUENAME at `date` from `hostname`" >> $LOG
return 0
################################ func_startMQ ######################################
func_startMQ () {
QUEUENAME=$1
case $QUEUENAME in
MECM|MEQM)
PORT=1414
MEQA|MEQB)
PORT=5077
esac
## echo "starting $QUEUENAME"
strmqm $QUEUENAME
sleep 5
## echo "STARTING listener for $QUEUENAME"
runmqlsr -t tcp -p $PORT -m $QUEUENAME &
sleep 5
runmqsc $QUEUENAME < /san1/var/mqm/config/startCH_$QUEUENAME.DEF
echo "started $QUEUENAME at `date` from `hostname`" >> $LOG
return 0
################################ func_checkMQ ######################################
func_checkMQ () {
QUEUENAME=$1
PROCESSCOUNT=`ps -ef | grep $QUEUENAME | grep -v grep| grep -v Control| awk {'print $2'} | wc -l`
if [ $PROCESSCOUNT > 1  ]
then
echo "there are $PROCESSCOUNT processes associated with $QUEUENAME" >> $LOG
echo ":::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::" >> $LOG
echo "Status of $QUEUENAME : " >> $LOG
runmqsc $QUEUENAME < /san1/var/mqm/config/ChannelStatus.DEFS
return 0
else
echo "there are $PROCESSCOUNT processes associated with $QUEUENAME" >> $LOG
echo ":::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::" >> $LOG
echo "Status of $QUEUENAME : " >> $LOG
runmqsc $QUEUENAME < /san1/var/mqm/config/ChannelStatus.DEFS
return 1
fi
ACTION=$1
# QUEUENAME=$2
case $ACTION in
start)
## func_startMQ MECM
## func_startMQ MEQA
func_startMQ MEQM
func_startMQ MEQB
stop)
## func_stopMQ MECM
## func_stopMQ MEQA
func_stopMQ MEQM
func_stopMQ MEQB
check)
## func_checkMQ MECM
## func_checkMQ MEQA
func_checkMQ MEQB
echo "Sorry, the proper usage for this command is ControlMQ.sh start|stop|check"
exit
esac