mjhjmtu

I have the service script below, which calls a shell script. We changed something in the script When I run the shell script directly, everything works fine. When I restart the servER, everything works fine. If I call sudo service my-service stop then sudo service my-service start, it will stop, but never start again. It says systemctl OK, but a process never kicks off, and no logs are ever written from the shell script. Even if I call sudo /etc/init.d/my-service it won't start. Again, mind you, this will start fine automatically on restart of the server. Any thoughts on what could be the problem, or how to get better logging from this failed start?

#!/bin/bash
#
# my-service
#
# chkconfig: 345 84 15
# description: Start up the My Service process.
# config: /etc/sysconfig/my-service
# processname: java

# Source function library.
. /etc/init.d/functions

# Loading the configuration parameters.
if [ -f /etc/sysconfig/my-service ]; then
 source /etc/sysconfig/my-service
fi

RETVAL=0

case "$1" in
 start)
 if [ -f /opt/my-service/latest/bin/my-service-start.sh ];
 then
 logger -s "Starting My Service"
 /bin/su -p -s /bin/sh myserviceuser /opt/my-service/latest/bin/my-service-start.sh
 RETVAL=$?
 [ $RETVAL = 0 ] && touch /var/lock/subsys/my-service
 fi
 ;;
 stop)
 if [ -f /opt/my-service/latest/bin/my-service-stop.sh ];
 then
 logger -s "Stopping My Service"
 /bin/su -p -s /bin/sh myserviceuser /opt/my-service/latest/bin/my-service-stop.sh
 RETVAL=$?
 [ $RETVAL = 0 ] && rm -f /var/lock/subsys/my-service
 fi
 ;;
 restart)
 $0 stop
 $0 start
 ;;
 *)
 echo $"Usage: $0 start"
 exit 1
 ;;
esac

exit $RETVAL

"We changed something in the script"

There is not enough information to authoritatively answer this question, since the contents of my-service-start.sh and my-service-stop.sh are a mystery.

Assuming everything you have said is true, then the case is that my-service-start.sh does not function as intended, once my-service-start.sh and my-service-stop.sh have both already been run. Even though it works fine at the first startup. This means something is altered after the scripts have been run. Things are not the same as they were when the start script ran successfully. Maybe a port your start script needs is still tied up.

I suggest looking into PID file handling to stop your script from potentially stepping on itself, and analyse your start/stop scripts for what is being altered. Try adding some error handling to your start script and see if that helps you isolate the problem.