def test_replay_pause(self): # Get a precompiled parameter dictionary with basic ctd fields pdict_id = self.dataset_management.read_parameter_dictionary_by_name('ctd_parsed_param_dict',id_only=True) context_ids = self.dataset_management.read_parameter_contexts(pdict_id, id_only=True) # Add a field that supports binary data input. bin_context = ParameterContext('binary', param_type=ArrayType()) context_ids.append(self.dataset_management.create_parameter_context('binary', bin_context.dump())) # Add another field that supports dictionary elements. rec_context = ParameterContext('records', param_type=RecordType()) context_ids.append(self.dataset_management.create_parameter_context('records', rec_context.dump())) pdict_id = self.dataset_management.create_parameter_dictionary('replay_pdict', parameter_context_ids=context_ids, temporal_context='time') stream_def_id = self.pubsub_management.create_stream_definition('replay_stream', parameter_dictionary_id=pdict_id) replay_stream, replay_route = self.pubsub_management.create_stream('replay', 'xp1', stream_definition_id=stream_def_id) dataset_id = self.create_dataset(pdict_id) scov = DatasetManagementService._get_simplex_coverage(dataset_id) bb = CoverageCraft(scov) bb.rdt['time'] = np.arange(100) bb.rdt['temp'] = np.random.random(100) + 30 bb.sync_with_granule() DatasetManagementService._persist_coverage(dataset_id, bb.coverage) # This invalidates it for multi-host configurations # Set up the subscriber to verify the data subscriber = StandaloneStreamSubscriber(self.exchange_space_name, self.validate_granule_subscription) xp = self.container.ex_manager.create_xp('xp1') self.queue_buffer.append(self.exchange_space_name) subscriber.start() subscriber.xn.bind(replay_route.routing_key, xp) # Set up the replay agent and the client wrapper # 1) Define the Replay (dataset and stream to publish on) self.replay_id, process_id = self.data_retriever.define_replay(dataset_id=dataset_id, stream_id=replay_stream) # 2) Make a client to the interact with the process (optionall provide it a process to bind with) replay_client = ReplayClient(process_id) # 3) Start the agent (launch the process) self.data_retriever.start_replay_agent(self.replay_id) # 4) Start replaying... replay_client.start_replay() # Wait till we get some granules self.assertTrue(self.event.wait(5)) # We got granules, pause the replay, clear the queue and allow the process to finish consuming replay_client.pause_replay() gevent.sleep(1) subscriber.xn.purge() self.event.clear() # Make sure there's no remaining messages being consumed self.assertFalse(self.event.wait(1)) # Resume the replay and wait until we start getting granules again replay_client.resume_replay() self.assertTrue(self.event.wait(5)) # Stop the replay, clear the queues replay_client.stop_replay() gevent.sleep(1) subscriber.xn.purge() self.event.clear() # Make sure that it did indeed stop self.assertFalse(self.event.wait(1)) subscriber.stop()
def test_dm_end_2_end(self): #-------------------------------------------------------------------------------- # Set up a stream and have a mock instrument (producer) send data #-------------------------------------------------------------------------------- self.event.clear() # Get a precompiled parameter dictionary with basic ctd fields pdict_id = self.dataset_management.read_parameter_dictionary_by_name('ctd_parsed_param_dict',id_only=True) context_ids = self.dataset_management.read_parameter_contexts(pdict_id, id_only=True) # Add a field that supports binary data input. bin_context = ParameterContext('binary', param_type=ArrayType()) context_ids.append(self.dataset_management.create_parameter_context('binary', bin_context.dump())) # Add another field that supports dictionary elements. rec_context = ParameterContext('records', param_type=RecordType()) context_ids.append(self.dataset_management.create_parameter_context('records', rec_context.dump())) pdict_id = self.dataset_management.create_parameter_dictionary('replay_pdict', parameter_context_ids=context_ids, temporal_context='time') stream_definition = self.pubsub_management.create_stream_definition('ctd data', parameter_dictionary_id=pdict_id) stream_id, route = self.pubsub_management.create_stream('producer', exchange_point=self.exchange_point_name, stream_definition_id=stream_definition) #-------------------------------------------------------------------------------- # Start persisting the data on the stream # - Get the ingestion configuration from the resource registry # - Create the dataset # - call persist_data_stream to setup the subscription for the ingestion workers # on the stream that you specify which causes the data to be persisted #-------------------------------------------------------------------------------- ingest_config_id = self.get_ingestion_config() dataset_id = self.create_dataset(pdict_id) self.ingestion_management.persist_data_stream(stream_id=stream_id, ingestion_configuration_id=ingest_config_id, dataset_id=dataset_id) self.addCleanup(self.stop_ingestion, stream_id) #-------------------------------------------------------------------------------- # Now the granules are ingesting and persisted #-------------------------------------------------------------------------------- self.launch_producer(stream_id) self.wait_until_we_have_enough_granules(dataset_id,40) #-------------------------------------------------------------------------------- # Now get the data in one chunk using an RPC Call to start_retreive #-------------------------------------------------------------------------------- replay_data = self.data_retriever.retrieve(dataset_id) self.assertIsInstance(replay_data, Granule) rdt = RecordDictionaryTool.load_from_granule(replay_data) self.assertTrue((rdt['time'][:10] == np.arange(10)).all(),'%s' % rdt['time'][:]) self.assertTrue((rdt['binary'][:10] == np.array(['hi']*10, dtype='object')).all()) #-------------------------------------------------------------------------------- # Now to try the streamed approach #-------------------------------------------------------------------------------- replay_stream_id, replay_route = self.pubsub_management.create_stream('replay_out', exchange_point=self.exchange_point_name, stream_definition_id=stream_definition) self.replay_id, process_id = self.data_retriever.define_replay(dataset_id=dataset_id, stream_id=replay_stream_id) log.info('Process ID: %s', process_id) replay_client = ReplayClient(process_id) #-------------------------------------------------------------------------------- # Create the listening endpoint for the the retriever to talk to #-------------------------------------------------------------------------------- sub_id = self.pubsub_management.create_subscription(self.exchange_space_name,stream_ids=[replay_stream_id]) self.addCleanup(self.pubsub_management.delete_subscription, sub_id) self.pubsub_management.activate_subscription(sub_id) self.addCleanup(self.pubsub_management.deactivate_subscription, sub_id) subscriber = StandaloneStreamSubscriber(self.exchange_space_name, self.validate_granule_subscription) subscriber.start() self.addCleanup(subscriber.stop) self.data_retriever.start_replay_agent(self.replay_id) self.assertTrue(replay_client.await_agent_ready(5), 'The process never launched') replay_client.start_replay() self.assertTrue(self.event.wait(10)) self.data_retriever.cancel_replay_agent(self.replay_id) #-------------------------------------------------------------------------------- # Test the slicing capabilities #-------------------------------------------------------------------------------- granule = self.data_retriever.retrieve(dataset_id=dataset_id, query={'tdoa':slice(0,5)}) rdt = RecordDictionaryTool.load_from_granule(granule) b = rdt['time'] == np.arange(5) self.assertTrue(b.all() if not isinstance(b,bool) else b)
def test_replay_pause(self): # Get a precompiled parameter dictionary with basic ctd fields pdict_id = self.dataset_management.read_parameter_dictionary_by_name( 'ctd_parsed_param_dict', id_only=True) context_ids = self.dataset_management.read_parameter_contexts( pdict_id, id_only=True) # Add a field that supports binary data input. bin_context = ParameterContext('binary', param_type=ArrayType()) context_ids.append( self.dataset_management.create_parameter_context( 'binary', bin_context.dump())) # Add another field that supports dictionary elements. rec_context = ParameterContext('records', param_type=RecordType()) context_ids.append( self.dataset_management.create_parameter_context( 'records', rec_context.dump())) pdict_id = self.dataset_management.create_parameter_dictionary( 'replay_pdict', parameter_context_ids=context_ids, temporal_context='time') stream_def_id = self.pubsub_management.create_stream_definition( 'replay_stream', parameter_dictionary_id=pdict_id) replay_stream, replay_route = self.pubsub_management.create_stream( 'replay', 'xp1', stream_definition_id=stream_def_id) dataset_id = self.create_dataset(pdict_id) scov = DatasetManagementService._get_coverage(dataset_id) bb = CoverageCraft(scov) bb.rdt['time'] = np.arange(100) bb.rdt['temp'] = np.random.random(100) + 30 bb.sync_with_granule() DatasetManagementService._persist_coverage( dataset_id, bb.coverage) # This invalidates it for multi-host configurations # Set up the subscriber to verify the data subscriber = StandaloneStreamSubscriber( self.exchange_space_name, self.validate_granule_subscription) xp = self.container.ex_manager.create_xp('xp1') self.queue_buffer.append(self.exchange_space_name) subscriber.start() subscriber.xn.bind(replay_route.routing_key, xp) # Set up the replay agent and the client wrapper # 1) Define the Replay (dataset and stream to publish on) self.replay_id, process_id = self.data_retriever.define_replay( dataset_id=dataset_id, stream_id=replay_stream) # 2) Make a client to the interact with the process (optionall provide it a process to bind with) replay_client = ReplayClient(process_id) # 3) Start the agent (launch the process) self.data_retriever.start_replay_agent(self.replay_id) # 4) Start replaying... replay_client.start_replay() # Wait till we get some granules self.assertTrue(self.event.wait(5)) # We got granules, pause the replay, clear the queue and allow the process to finish consuming replay_client.pause_replay() gevent.sleep(1) subscriber.xn.purge() self.event.clear() # Make sure there's no remaining messages being consumed self.assertFalse(self.event.wait(1)) # Resume the replay and wait until we start getting granules again replay_client.resume_replay() self.assertTrue(self.event.wait(5)) # Stop the replay, clear the queues replay_client.stop_replay() gevent.sleep(1) subscriber.xn.purge() self.event.clear() # Make sure that it did indeed stop self.assertFalse(self.event.wait(1)) subscriber.stop()