Пример #1
0
def RunWorker(connectionStringKey):
    InitLogging()
    while (True):
        jobManager = JobManager(connectionStringKey)
        jobCount = jobManager.count()

        # not spawning new thread for unless queue has a job
        # otherwise lot of SQS calls are made and bill is high
        if jobCount > 0:
            jobProcesses = []

            while len(jobProcesses) < int(os.environ['MAX_JOB_THREADS']):
                jobProcess = Process(target=DequeueAndStartJob,
                                     args=(connectionStringKey, ))
                jobProcess.start()
                jobProcesses.append(jobProcess)

            logging.info("Process limit reached")

            for jobProcess in jobProcesses:
                jobProcess.join(150)
                if jobProcess.is_alive():
                    logging.warning("Process timed out. Terminating it")
                    jobProcess.terminate()
        else:
            logging.info("No job found in queue. Sleeping")
            time.sleep(5)
Пример #2
0
    def __init__(self):
        JobManager.__init__(self)
        pickledGraphs = "graphs.pickle"
        self.pickledGraphs = join(self.jobManagerDir, pickledGraphs)

        if not isfile(self.pickledGraphs):
            self.initGraphs()

        self.graphs = {}
        self.loadGraphs()
Пример #3
0
    def __init__(self):
        self.daemon = True

        vmms = None
        if Config.VMMS_NAME == "tashiSSH":
            from vmms.tashiSSH import TashiSSH
            vmms = TashiSSH()
        elif Config.VMMS_NAME == "ec2SSH":
            from vmms.ec2SSH import Ec2SSH
            vmms = Ec2SSH()
        elif Config.VMMS_NAME == "localDocker":
            from vmms.localDocker import LocalDocker
            vmms = LocalDocker()
        elif Config.VMMS_NAME == "distDocker":
            from vmms.distDocker import DistDocker
            vmms = DistDocker()

        self.preallocator = Preallocator({Config.VMMS_NAME: vmms})
        self.jobQueue = JobQueue(self.preallocator)
        if not Config.USE_REDIS:
            # creates a local Job Manager if there is no persistent
            # memory between processes. Otherwise, JobManager will
            # be initiated separately
            JobManager(self.jobQueue).start()

        logging.basicConfig(
            filename=Config.LOGFILE,
            format="%(levelname)s|%(asctime)s|%(name)s|%(message)s",
            level=Config.LOGLEVEL,
        )
        self.start_time = time.time()
        self.log = logging.getLogger("TangoServer")
        self.log.info("Starting Tango server")
Пример #4
0
def DequeueAndStartJob(connectionStringKey):
    """
    Dequeue a job from the queue and start executing it.
    """

    logging.info("Dequeing a job.")
    jobManager = JobManager(connectionStringKey)
    job = jobManager.dequeueJob()

    if job is None:
        logging.info("No job found.")
        return

    logging.info("Job found. Starting it now." + "Job id: %s. Job Name: %s.",
                 job.jobId, job.jobName)
    RunJob(job)
Пример #5
0
def main():
    global server_socket, clients
    server_socket, clients = None, []
    global last_auto_resched

    try:
        defs.db_m = DatabaseManager()
        defs.job_m = defs.db_m.get_jobManager()
        if not defs.job_m:
            defs.job_m = JobManager()
            defs.db_m.register_jobManager(defs.job_m)
            transaction.commit()

        last_auto_resched = time()

        server_socket = listen_on(SERVER_BIND, SERVER_PORT)
        socket2addr = {}

        print(WELCOME_MESSAGE)
        print("Enter 'Help' or 'H' for more details.\n")
        show_prompt()

        while True:
            r, _, _ = select([server_socket, stdin] + clients, [], [])
            for c in r:
                if c == server_socket:
                    c, addr = server_socket.accept()
                    socket2addr[c] = addr
                    clients.append(c)
                elif c == stdin:
                    handle_request(input())
                    show_prompt()
                else:
                    handle_client(c, socket2addr[c])

            auto_resched()
            transaction.commit()

    except Exception as e:
        ndprint('Error occurred: \n\t' + repr(e))

    finally:
        print("Start closing.")
        if defs.db_m:
            defs.db_m.close()
        if server_socket:
            server_socket.close()
        for c in clients:
            c.close()
        print("Done closing.")
Пример #6
0
    def __init__(self):

        logging.basicConfig(
                filename = self.LOGFILE,
                format = "%(levelname)s|%(asctime)s|%(name)s|%(message)s",
                level = Config.LOGLEVEL
                )

        vmms = None

        if Config.VMMS_NAME == "localSSH":
            from vmms.localSSH import LocalSSH
            vmms = LocalSSH()
        elif Config.VMMS_NAME == "tashiSSH":
            from vmms.tashiSSH import TashiSSH
            vmms = TashiSSH()
        elif Config.VMMS_NAME == "ec2SSH":
            from vmms.ec2SSH import Ec2SSH
            vmms = Ec2SSH()
        elif Config.VMMS_NAME == "localDocker":
            from vmms.localDocker import LocalDocker
            vmms = LocalDocker()
            

        self.vmms = {Config.VMMS_NAME: vmms}
        self.preallocator = Preallocator(self.vmms)
        self.queue = JobQueue(self.preallocator)

        if not Config.USE_REDIS:
            # creates a local Job Manager if there is no persistent
            # memory between processes. Otherwise, JobManager will
            # be initiated separately
            JobManager(self.queue, self.vmms, self.preallocator)

        self.tango = TangoServer(self.queue, self.preallocator, self.vmms)

        logging.basicConfig(
            filename=self.LOGFILE,
            format="%(levelname)s|%(asctime)s|%(name)s|%(message)s",
            level=Config.LOGLEVEL
        )

        logging.getLogger('boto').setLevel(logging.INFO)
        self.log = logging.getLogger("TangoREST")
        self.log.info("Starting RESTful Tango server")
        self.status = Status()
Пример #7
0
 def __init__(self):
     JobManager.__init__(self, 'CLUSTER_JOBSQUEUE_CONNECTIONSTRING')