fork and thread and signal... oooooops

[Haeyoung Kim]

I know little about system programming, so my code has become a
monster.
My situation is : 1) used /etc/init.d script functions such as daemon,
killproc to start/stop python script.
2) python script is composed of two tasks. launching java server
program and running script that connects that server to see if it's
alive or dead.
so, I programmed python script like this.

PID = 0
ISALIVE = 1

def exitProgram():
    ISALIVE=0
    killServer()

def killServer():
    if PID:
	os.kill(PID,SIGINT)
	PID = 0

class Monitor(Thread):
    def run(self):
        # connect to server and check its heartbeat
	# and call restartServer and exit when server is dead.

def launchServer():
    pid = fork()
    if not pid:
	execv('java',args)	#launch server
    else:
	PID = pid		#start cheking server heartbeat
	Monitor().start()	#and set PID for killing server.
	p, e = os.wait()	#wait until server is killed
				#because of killServer or restartSErver.

#hope to clean up java server when shell script 'killproc' terminate
#this script
signal.signal( SIGINT, exitProgram ) 
signal.signal( SIGHUP, exitProgram ) 

if __name__==__main__:
    pid = fork()
    if not pid: os.exit
    else:
	pid = fork()
	if not pid: os.exit
	else:			#now start main program
	    while ISALIVE:
		launchServer()

Yes, it's stupid. But I didn't imagined such a picture at first. 
When shell script calls killproc(in init.d script), it sends signal,
and kill server-using os.kill(java_server_pid)- and get out of
os.wait() and died out generously - just my hope.
But it betrayed my hope. it throws TypeError on os.wait(), killed
anyway, but there are still my java program. :(
I think signal doesn't call exitProgram() - why?
Is there any clean solution to such a matter?
thanks in advance your kindness for reading my naive code.