def _start_job(self, job, partition_list): cqm = ComponentProxy(self.COMP_QUEUE_MANAGER) try: self.logger.info("trying to start job %d on partition %r" % (job.jobid, partition_list)) cqm.run_jobs([{'tag':"job", 'jobid':job.jobid}], partition_list) except ComponentLookupError: self.logger.error("failed to connect to queue manager") return self.started_jobs[job.jobid] = self.get_current_time()
def _start_job(self, job, partition_list): cqm = ComponentProxy(self.COMP_QUEUE_MANAGER) try: self.logger.info("trying to start job %d on partition %r" % (job.jobid, partition_list)) cqm.run_jobs([{'tag': "job", 'jobid': job.jobid}], partition_list) except ComponentLookupError: self.logger.error("failed to connect to queue manager") return self.started_jobs[job.jobid] = self.get_current_time()
def _start_job(self, job, partition_list, resid=None): """Get the queue manager to start a job.""" cqm = ComponentProxy("queue-manager") try: self.logger.info("trying to start job %d on partition %r" % (job.jobid, partition_list)) cqm.run_jobs([{'tag':"job", 'jobid':job.jobid}], partition_list, None, resid) except ComponentLookupError: self.logger.error("failed to connect to queue manager") return self.started_jobs[job.jobid] = self.get_current_time()
def _start_job(self, job, partition_list, resid=None): """Get the queue manager to start a job.""" cqm = ComponentProxy("queue-manager") try: self.logger.info("trying to start job %d on partition %r" % (job.jobid, partition_list)) cqm.run_jobs([{ 'tag': "job", 'jobid': job.jobid }], partition_list, None, resid, job.walltime) except ComponentLookupError: self.logger.error("failed to connect to queue manager") return self.started_jobs[job.jobid] = self.get_current_time()
exc = JobDeleteError(*args) print >> sys.stderr, "Job %s: ERROR - %s" % (exc.jobid, exc.message) raise SystemExit, 1 else: raise elif opts['run']: location = opts['run'] part_list = ComponentProxy("system").get_partitions([{ 'name': location }]) if len(part_list) != 1: print "Error: cannot find partition named '%s'" % location raise SystemExit, 1 try: response = cqm.run_jobs(spec, location.split(':'), whoami) except xmlrpclib.Fault, flt: if flt.faultCode == JobRunError.fault_code: args = eval(flt.faultString) exc = JobRunError(*args) print >> sys.stderr, "Job %s: ERROR - %s" % (exc.jobid, exc.message) raise SystemExit, 1 else: raise elif opts['addq']: existing_queues = get_queues(cqm) if [ qname for qname in args if qname in [q.get('name') for q in existing_queues] ]:
except xmlrpclib.Fault, flt: if flt.faultCode == JobDeleteError.fault_code: args = eval(flt.faultString) exc = JobDeleteError(*args) print >>sys.stderr, "Job %s: ERROR - %s" % (exc.jobid, exc.message) raise SystemExit, 1 else: raise elif opts['run']: location = opts['run'] part_list = ComponentProxy("system").get_partitions([{'name': location}]) if len(part_list) != 1: print "Error: cannot find partition named '%s'" % location raise SystemExit, 1 try: response = cqm.run_jobs(spec, location.split(':'), whoami) except xmlrpclib.Fault, flt: if flt.faultCode == JobRunError.fault_code: args = eval(flt.faultString) exc = JobRunError(*args) print >>sys.stderr, "Job %s: ERROR - %s" % (exc.jobid, exc.message) raise SystemExit, 1 else: raise elif opts['addq']: existing_queues = get_queues(cqm) if [qname for qname in args if qname in [q.get('name') for q in existing_queues]]: print 'queue already exists' response = '' elif len(args) < 1:
def test_something(self): logging.basicConfig() try: cqm = ComponentProxy("queue-manager") except ComponentLookupError: assert not "failed to connect to queue manager" # add a queue queues = cqm.add_queues([{"tag": "queue", "name": "default"}]) assert len(queues) == 1 # try adding a job to a queue that doesn't exist try: jobs = cqm.add_jobs([{"tag": "job", "queue": "jonx"}]) except xmlrpclib.Fault: # trying to add a job to a queue that doesn't exist results in an xmlrpc Fault pass else: assert not "Adding job to non-existent queue should raise xmlrpclib.Fault" # get the list of available partitions and add them to the pool of managed partitions try: simulator = ComponentProxy("system") except ComponentLookupError: assert not "failed to connect to simulator" for part_name in self.system._partitions: partitions = simulator.add_partitions([{"tag": "partition", "name": part_name, "queue": "default"}]) assert len(partitions) == 1 partitions = simulator.set_partitions( [{"tag": "partition", "name": part_name}], {"functional": True, "scheduled": True} ) assert len(partitions) == 1 partitions = simulator.get_partitions([{"name": "*", "size": "*", "queue": "*"}]) assert len(partitions) > 0 # now run a real job # # 1. add the job to the default queue # 2. obtain a partition for it to run on # 3. start running it on that paritition # 4. check that it started running # 5. sleep for a bit, and then check that it's still running # 6. sleep some more and then check to see if it actually finished running nodes = partitions[0]["size"] jobs = cqm.add_jobs( [ { "queue": "default", "mode": "co", "command": "/bin/ls", "outputdir": os.getcwd(), "walltime": 4, "nodes": nodes, "procs": nodes, "args": [], "user": "******", "jobid": "*", } ] ) assert len(jobs) == 1 job = jobs[0] jobid = job["jobid"] job_location_args = [ { "jobid": jobid, "nodes": job["nodes"], "queue": job["queue"], "utility_score": 1, "threshold": 1, "walltime": job["walltime"], "attrs": {}, } ] locations = simulator.find_job_location(job_location_args, []) assert locations.has_key(jobid) location = locations[jobid] cqm.run_jobs([{"jobid": jobid}], location) r = cqm.get_jobs([{"jobid": jobid, "state": "*", "is_active": True}]) if not r: assert not "the job didn't start" time.sleep(20) r = cqm.get_jobs([{"jobid": jobid, "state": "*", "is_active": True}]) if len(r) != 1: assert not "the job has stopped running prematurely" start_time = time.time() while True: r = cqm.get_jobs([{"jobid": jobid, "state": "*", "is_active": True}]) if r: if time.time() - start_time > 240: assert not "the job seems to have run overtime" else: time.sleep(5) else: break # this time, we'll add a job to the queue, start the job, sleep for a bit # and then try to kill the job before it has finished nodes = partitions[0]["size"] jobs = cqm.add_jobs( [ { "queue": "default", "mode": "co", "command": "/bin/ls", "outputdir": os.getcwd(), "walltime": 4, "nodes": nodes, "procs": nodes, "args": [], "user": "******", "jobid": "*", } ] ) assert len(jobs) == 1 job = jobs[0] jobid = job["jobid"] job_location_args = [ { "jobid": jobid, "nodes": job["nodes"], "queue": job["queue"], "utility_score": 1, "threshold": 1, "walltime": job["walltime"], "attrs": {}, } ] locations = simulator.find_job_location(job_location_args, []) assert locations.has_key(jobid) location = locations[jobid] cqm.run_jobs([{"jobid": jobid}], location) r = cqm.get_jobs([{"jobid": jobid, "state": "*", "is_active": True}]) if not r: assert not "the job didn't start" time.sleep(20) r = cqm.get_jobs([{"jobid": jobid, "is_active": True}]) if len(r) != 1: assert not "the job has stopped running prematurely" cqm.del_jobs([{"jobid": jobid}]) start_time = time.time() while True: r = cqm.get_jobs([{"jobid": jobid, "is_active": True, "state": "*"}]) if r: if time.time() - start_time > 30: assert not "the job didn't die when asked to" else: time.sleep(1) else: break
def test_something(self): logging.basicConfig() try: cqm = ComponentProxy("queue-manager") except ComponentLookupError: assert not "failed to connect to queue manager" # add a queue queues = cqm.add_queues([{'tag': "queue", 'name': "default"}]) assert len(queues) == 1 # try adding a job to a queue that doesn't exist try: jobs = cqm.add_jobs([{'tag': "job", 'queue': "jonx"}]) except xmlrpclib.Fault: # trying to add a job to a queue that doesn't exist results in an xmlrpc Fault pass else: assert not "Adding job to non-existent queue should raise xmlrpclib.Fault" # get the list of available partitions and add them to the pool of managed partitions try: simulator = ComponentProxy("system") except ComponentLookupError: assert not "failed to connect to simulator" for part_name in self.system._partitions: partitions = simulator.add_partitions([{ 'tag': "partition", 'name': part_name, 'queue': "default" }]) assert len(partitions) == 1 partitions = simulator.set_partitions([{ 'tag': "partition", 'name': part_name }], { 'functional': True, 'scheduled': True }) assert len(partitions) == 1 partitions = simulator.get_partitions([{ 'name': "*", 'size': "*", 'queue': "*" }]) assert len(partitions) > 0 # now run a real job # # 1. add the job to the default queue # 2. obtain a partition for it to run on # 3. start running it on that paritition # 4. check that it started running # 5. sleep for a bit, and then check that it's still running # 6. sleep some more and then check to see if it actually finished running nodes = partitions[0]['size'] jobs = cqm.add_jobs([{ 'queue': "default", 'mode': "co", 'command': "/bin/ls", 'outputdir': os.getcwd(), 'walltime': 4, 'nodes': nodes, 'procs': nodes, 'args': [], 'user': "******", 'jobid': "*" }]) assert len(jobs) == 1 job = jobs[0] jobid = job['jobid'] job_location_args = [{ 'jobid': jobid, 'nodes': job['nodes'], 'queue': job['queue'], 'utility_score': 1, 'threshold': 1, 'walltime': job['walltime'], 'attrs': {} }] locations = simulator.find_job_location(job_location_args, []) assert locations.has_key(jobid) location = locations[jobid] cqm.run_jobs([{'jobid': jobid}], location) r = cqm.get_jobs([{'jobid': jobid, 'state': "*", 'is_active': True}]) if not r: assert not "the job didn't start" time.sleep(20) r = cqm.get_jobs([{'jobid': jobid, 'state': "*", 'is_active': True}]) if len(r) != 1: assert not "the job has stopped running prematurely" start_time = time.time() while True: r = cqm.get_jobs([{ 'jobid': jobid, 'state': "*", 'is_active': True }]) if r: if time.time() - start_time > 240: assert not "the job seems to have run overtime" else: time.sleep(5) else: break # this time, we'll add a job to the queue, start the job, sleep for a bit # and then try to kill the job before it has finished nodes = partitions[0]['size'] jobs = cqm.add_jobs([{ 'queue': "default", 'mode': "co", 'command': "/bin/ls", 'outputdir': os.getcwd(), 'walltime': 4, 'nodes': nodes, 'procs': nodes, 'args': [], 'user': "******", 'jobid': "*" }]) assert len(jobs) == 1 job = jobs[0] jobid = job['jobid'] job_location_args = [{ 'jobid': jobid, 'nodes': job['nodes'], 'queue': job['queue'], 'utility_score': 1, 'threshold': 1, 'walltime': job['walltime'], 'attrs': {} }] locations = simulator.find_job_location(job_location_args, []) assert locations.has_key(jobid) location = locations[jobid] cqm.run_jobs([{'jobid': jobid}], location) r = cqm.get_jobs([{'jobid': jobid, 'state': "*", 'is_active': True}]) if not r: assert not "the job didn't start" time.sleep(20) r = cqm.get_jobs([{'jobid': jobid, 'is_active': True}]) if len(r) != 1: assert not "the job has stopped running prematurely" cqm.del_jobs([{'jobid': jobid}]) start_time = time.time() while True: r = cqm.get_jobs([{ 'jobid': jobid, 'is_active': True, 'state': "*" }]) if r: if time.time() - start_time > 30: assert not "the job didn't die when asked to" else: time.sleep(1) else: break