Changed sighandler in autoserv to call SIGKILL on its children instead of SIGTERM.
Changed monitor_db to call SIGTERM only once. We call SIGCONT before SIGTERM in order to make the stopped process continue.

Signed-off-by: Svitlana Tumanova <stumanova@google.com>




git-svn-id: http://test.kernel.org/svn/autotest/trunk@1312 592f7852-d20e-0410-864c-8624ca9c26a4
diff --git a/scheduler/monitor_db b/scheduler/monitor_db
index d88ebeb..48fa107 100755
--- a/scheduler/monitor_db
+++ b/scheduler/monitor_db
@@ -411,16 +411,9 @@
 
 
 	def kill(self):
-		for i in range(0, 4):
-			if self.proc.poll() == None:
-				os.kill(self.proc.pid, signal.SIGTERM)
-				time.sleep(5)
-			# Check that the process was terminated
-			if self.proc.poll() != None:
-				return
-			
-		print ("""Error: process %d has not terminated""" % 
-								self.proc.pid)
+		if self.proc.poll() == None:
+			os.kill(self.proc.pid, signal.SIGCONT)
+			os.kill(self.proc.pid, signal.SIGTERM)
 		
 
 	def exit_code(self):