Ai remis le cluster controller en route. Il n'est pas si simple que ça à démarrer, alors j'en ai profité pour écrire l'installeur, qui tient également lieu de documentation. (partie 5)

ClusterController/ClusterController.py

@@ -181,7 +181,7 @@ class ClusterController:
 		self.wakeUpMachinesForPendingJobs()
 		
 	def storeSessionInDatabase( self ):
-		conn = MySQLdb.connect('simpatix10', 'root', '', 'clustercontroller')
+		conn = MySQLdb.connect('simpatix10', 'clusterctrl', '', 'clustercontroller')
 		assert(conn)
 		
 		# retrieve the session id, as it's an auto_increment field
@@ -226,10 +226,11 @@ class ClusterController:
 		"""
 		"""
 		self.m_iSessionId = self.storeSessionInDatabase()
+		log("storeSessionInDatabase completed")
 		DELAY_BETWEEN_MEASURES = 10 # in seconds
 		self.m_clusterStatus.startReadingThreads()
 		while not self.m_clusterStatus.isReady():
-			#log('waiting for system to be ready')
+			log('waiting for system to be ready')
 			time.sleep(1)
 			None
 		logInfo('ClusterController::run : cluster initial readings have completed')

ClusterController/ClusterStatus.py

@@ -17,10 +17,9 @@ class ClusterStatus:
 		self.m_lock = threading.Lock() # to prevent concurrent access to this instance
 		self.m_jobsStateUpdater = JobsStateUpdater( self )
 		self.m_jobsState = None
-		#self.m_controlledMachineNames = [ 'simpatix26', 'simpatix27', 'simpatix38', 'simpatix10' ]
-		#self.m_controlledMachineNames = [ 'simpatix15' ]
-		self.m_controlledMachineNames = [] # [ 'simpatix10' ]
-		if True:
+		self.m_controlledMachineNames = [ 'simpatix30' ]
+		#self.m_controlledMachineNames = [] # [ 'simpatix30' ]
+		if False:
 			for iMachine in range(11, 40):
 				if (iMachine == 31) or (iMachine == 32):
 					continue # these machines don't seem to be able to go to sleep properly (bug 00000010)

ClusterController/Log.py

@@ -4,7 +4,6 @@ import threading
 gLogFilePath = '/var/log/ClusterController.log'
 
 def log( message ):
-	return
 	threadName = threading.currentThread().getName()
 	logMessage = time.asctime(time.localtime())+' : '+ threadName + ' : ' + message
 	print logMessage

ClusterController/QstatParser.py

@@ -20,7 +20,7 @@ class QstatParser:
 			elif c == 't':
 				jobState += JobStateFlags.TRANSFERING
 			else:
-				assert( False, 'unhandled job state flag :"' + c + '"' )
+				assert False, 'unhandled job state flag :"' + c + '"'
 		return jobState
 	def parseQstatOutput( self, qstatOutput ):
 		jobsState = JobsState()

ClusterController/SunGridEngine.py

@@ -26,7 +26,7 @@ class SunGridEngine:
 		if False: # no need for job details at the moment and since it's very slow, it's been disabled
 			for jobId, job in jobsState.getPendingJobs().iteritems():
 				(returnCode, stdout, stderr) = executeProgram( ['qstat', '-j', job.getId().asStr()] )
-				assert( returnCode != 0, 'prout' )
+				assert returnCode != 0, 'prout'
 				QstatParser().parseJobDetails( stdout, job )
 		
 		return jobsState

ClusterController/Util.py

@@ -207,6 +207,7 @@ def onException(exception):
 	strMessage += f.getvalue()
 	f.close()
 	logError(strMessage)
+	print(strMessage)
 	Lib.Util.sendTextMail( 'ClusterController <guillaume.raffy@univ-rennes1.fr>', 'guillaume.raffy@univ-rennes1.fr', 'ClusterController has stopped because of an exception', strMessage)
 	executeCommand('kill -9 %d' % os.getpid()) # stop other threads immediately
 	exit()