script spawned twice

[Leonardo Gabrielli]

Dear all,
I'm sorry to bother here in the list, but I can't find upstart experts 
on other forums.

I am testing an upstart job for automatic connection of machines based 
on avahi and other stuff, similar to UPnP but focused on professional audio.

The problem I'm facing it's hawkward: a script launched by the upstart 
job appears in the process list twice, one instance lives forever (as it 
must) and another spurious one is born periodically and dies in a few 
seconds. The strange thing is: the parent of the spurious script is the 
first instance, but in my script there is no command that should give 
birth to another instance of the script!

Details:

the .conf in /etc/init launches these commands in the "script" stanza:
[CODE]script
         if [ -f /home/ccrma/on-startup/WeMUST/wemust-start.sh ]; then
                 sudo -H -E -u ccrma 
/home/ccrma/on-startup/WeMUST/wemust-start.sh > /home/ccrma/on-startup/$
                 sudo -H -E -u ccrma 
/home/ccrma/on-startup/WeMUST/afterConnect.sh > /home/ccrma/on-startup/$
         elif [ -f  /home/ccrma/on-startup/load_default_patch ]; then
                 sudo -H -E -u ccrma 
/home/ccrma/on-startup/load_default_patch
         fi
end script[/CODE]

(I know it is deprecated to have long scripts in the "script" stanza, I 
see if I can improve this later).
The abnormal behavior is with the wemust-start.sh script.

The wemust-start.sh script is quite long, and includes call to jackd, 
avahi, jacktrip and more, but there is NO reference whatsoever to 
itself, so I can't see how the script could initiate a new 
wemust-start.sh instance.

I know that the parent process to the periodically appearing instance is 
the first instance of wemust-start.sh from the following command:
ps -p `ps aux | grep [w]emust-start | awk 'NR == 2'{ print $2 }'` -o ppid=

So I read the second line (whenever it is not empty) of ps to get its 
pid and then send the pid to ps -o ppid to know what is the parent pid. 
The parent pid is the first wemust-start instance. The first instance 
has of course parent pid = 1.

Maybe the problem is with the conf file? I tried to mingle with the 
"expect" and "respawn" stanzas to no avail...
Any suggestions?

The entire .conf file:

[CODE]description     "WeMUST daemon"

# starts at runlevel 2 after avahi
start on (runlevel [2]
         and started avahi-daemon)
stop on runlevel [016]

#expect fork
#respawn limit 2 1

pre-start script
         echo "[`date +"%m/%d/%Y %H:%M:%S"`] UPSTART: Starting 
wemust.conf startup job" >> /home/ccrma/on-startup/WeMUST/.globaldump
end script

script
         if [ -f /home/ccrma/on-startup/WeMUST/wemust-start.sh ]; then
                 sudo -H -E -u ccrma 
/home/ccrma/on-startup/WeMUST/wemust-start.sh > 
/home/ccrma/on-startup/WeMUST/wemust.log &
                 sudo -H -E -u ccrma 
/home/ccrma/on-startup/WeMUST/afterConnect.sh > 
/home/ccrma/on-startup/WeMUST/afterConnect.log &
         elif [ -f  /home/ccrma/on-startup/load_default_patch ]; then
                 sudo -H -E -u ccrma 
/home/ccrma/on-startup/load_default_patch
         fi
end script

post-stop script
         killall wemust-start.sh
         killall afterConnect.sh
         echo "[`date +"%m/%d/%Y %H:%M:%S"`] UPSTART: Launched 
wemust.conf startup job" >> /home/ccrma/on-startup/WeMUST/.globaldump
end script
[/CODE]

PLATFORM:
It's a BeagleBoard xM running Ubuntu 10.10 with a patched kernel.
Linux omap 3.2.2-x4 #1 SMP PREEMPT armv7l GNU/Linux
upstart is version 2010-02-04 0.6.5