Shutdown of cache database not being completed

[wayne_104]

Hi

I have the shutdown script of my cache database in rc2.d as Kcache950

This is a link to a script in /sbin/init.d

#!/bin/sh
set -x
#
# /etc/rc.d/init.d/cache
# Cashier Control Script
#
# chkconfig: 2345 55 01
# description: Cache DB service
#


RETVAL=0
CACHE_USR="cacheusr"

case "$1" in
start)
echo -n ""
echo "Starting cache:"
echo $PATH
/usr/bin/su - $CACHE_USR -c "/home/cacheusr/cache_start.sh"
RETVAL=0
echo
[ $RETVAL = 0 ] && touch /var/lock/subsys/cache
;;
stop)
echo -n "Stopping cache: "
/usr/bin/su - $CACHE_USR -c "/home/cacheusr/cache_stop.sh &"
CHILD_PID=${!}
wait ${CHILD_PID} # parent will wait until child finishes

RETVAL=0
echo
[ $RETVAL = 0 ] && rm -f /var/lock/subsys/cache
;;
restart)
stop
start
;;
*)
echo "usage: $0 (start|stop|restart|help)"
esac
cache: END

however hp ux does not wait for this script to complete before shuting down so the database crashes.

Why???

[Earl_Crowder]

Well, the "su - $CACHE_USER" spawns a new shell, which in turn backgrounds the cache_stop.sh, and then exits. The parent shell doesn't know that the child shell spawned a process in the background.

You could change it to not background the cache_stop.sh, and remove the CHILD_PID assignment and the call to wait.

Also, you can remove the "-n" from the echo commands.

[Zinky]

Looks like the above is for Linux sir.

It should still technically work... the /var/lock scheme I don;t think is available in HP-UX.

So drop & in :

/usr/bin/su - $CACHE_USR -c "/home/cacheusr/cache_stop.sh &"

[wayne_104]

Hi Guys thanks for that.

The /usr/bin/su - $CACHE_USR -c "/home/cacheusr/cache_stop.sh &"
CHILD_PID=${!}
wait ${CHILD_PID} # parent will wait until child finishes
was suggested as my shutdown of the db does not complete before the system goes down.

it gets as far as the messages below.

01/13/11-07:58:24:862 (14349) 0 Receive daemon exited due to SIGTERM received
01/13/11-07:58:24:862 (14369) 0 Application Monitor exited due to SIGTERM receiv
ed


During a shutdown it should end with cahe shutdown complete message.

which if i do it manualy is the case.

And when i bring up the db it says it is fine.

However when i do a reboot it does not and starting the db states it is recovering from a crash.

[Zinky]

Have you tried removing the & and the wait?

[wayne_104]

yip i have with the same reult

[Earl_Crowder]

Well then sounds like the cache_stop.sh script isn't waiting for the database to be shut down cleanly before it finishes.

Might be best to modify the cache_stop.sh to actually wait for the shutdown to complete before exiting?

[Zinky]

Or check if cache_stop.sh has a background spec in whatever binary or script it calls?

Or even add a looper there to check if the software has actually stopped?

[RickT_1]

How are you rebooting, what command? Reboot?

According to the manual:

WARNINGS
reboot does not invoke the shutdown scripts associated with subsystems to bring them down in a cautious
manner. See shutdown(1M).

If you want your shutdown scripts to work you need to use shutdown.


Rick

[wayne_104]

Rick you are the man thanks.