def create_ingestion_configuration(self,
exchange_point_id='',
couch_storage=None,
hdf_storage=None,
number_of_workers=0):
"""
@brief Setup ingestion workers to ingest all the data from a single exchange point.
@param exchange_point_id is the resource id for the exchagne point to ingest from
@param couch_storage is the specification of the couch database to use
@param hdf_storage is the specification of the filesystem to use for hdf data files
@param number_of_workers is the number of ingestion workers to create
"""
if self.process_definition_id is None:
process_definition = ProcessDefinition(
name='ingestion_worker_process',
description='Worker transform process for ingestion of datasets'
)
process_definition.executable[
'module'] = 'ion.processes.data.ingestion.ingestion_worker'
process_definition.executable['class'] = 'IngestionWorker'
self.process_definition_id = self.clients.process_dispatcher.create_process_definition(
process_definition=process_definition)
# Give each ingestion configuration its own queue name to receive data on
exchange_name = 'ingestion_queue'
##------------------------------------------------------------------------------------
## declare our intent to subscribe to all messages on the exchange point
query = ExchangeQuery()
subscription_id = self.clients.pubsub_management.create_subscription(query=query,\
exchange_name=exchange_name, name='Ingestion subscription', description='Subscription for ingestion workers')
##------------------------------------------------------------------------------------------
# create an ingestion_configuration instance and update the registry
# @todo: right now sending in the exchange_point_id as the name...
ingestion_configuration = IngestionConfiguration(name=self.XP)
ingestion_configuration.description = '%s exchange point ingestion configuration' % self.XP
ingestion_configuration.number_of_workers = number_of_workers
if hdf_storage is not None:
ingestion_configuration.hdf_storage.update(hdf_storage)
if couch_storage is not None:
ingestion_configuration.couch_storage.update(couch_storage)
ingestion_configuration_id, _ = self.clients.resource_registry.create(
ingestion_configuration)
self._launch_transforms(ingestion_configuration.number_of_workers,
subscription_id, ingestion_configuration_id,
ingestion_configuration,
self.process_definition_id)
return ingestion_configuration_id
def create_ingestion_configuration(self, exchange_point_id='', couch_storage=None, hdf_storage=None,number_of_workers=0):
"""
@brief Setup ingestion workers to ingest all the data from a single exchange point.
@param exchange_point_id is the resource id for the exchagne point to ingest from
@param couch_storage is the specification of the couch database to use
@param hdf_storage is the specification of the filesystem to use for hdf data files
@param number_of_workers is the number of ingestion workers to create
"""
if self.process_definition_id is None:
res, _ = self.clients.resource_registry.find_resources(restype=RT.ProcessDefinition,name='ingestion_worker_process', id_only=True)
if not len(res):
raise BadRequest('No ingestion work process definition found')
self.process_definition_id = res[0]
# Give each ingestion configuration its own queue name to receive data on
#----------------------------- TODO ---------------------------------------------
# Add support right here for user specified queue based on this set of ingestion
# workers
#----------------------------- TODO ---------------------------------------------
exchange_name = 'ingestion_queue'
##------------------------------------------------------------------------------------
## declare our intent to subscribe to all messages on the exchange point
query = ExchangeQuery()
subscription_id = self.clients.pubsub_management.create_subscription(query=query,
exchange_name=exchange_name, name='Ingestion subscription', description='Subscription for ingestion workers')
##------------------------------------------------------------------------------------------
# create an ingestion_configuration instance and update the registry
# @todo: right now sending in the exchange_point_id as the name...
ingestion_configuration = IngestionConfiguration( name = self.XP)
ingestion_configuration.description = '%s exchange point ingestion configuration' % self.XP
ingestion_configuration.number_of_workers = number_of_workers
if hdf_storage is not None:
ingestion_configuration.hdf_storage.update(hdf_storage)
if couch_storage is not None:
ingestion_configuration.couch_storage.update(couch_storage)
ingestion_configuration_id, _ = self.clients.resource_registry.create(ingestion_configuration)
self._launch_transforms(
ingestion_configuration.number_of_workers,
subscription_id,
ingestion_configuration_id,
ingestion_configuration,
self.process_definition_id
)
return ingestion_configuration_id
def create_ingestion_configuration(self, exchange_point_id='', couch_storage=None, hdf_storage=None,number_of_workers=0):
"""
@brief Setup ingestion workers to ingest all the data from a single exchange point.
@param exchange_point_id is the resource id for the exchagne point to ingest from
@param couch_storage is the specification of the couch database to use
@param hdf_storage is the specification of the filesystem to use for hdf data files
@param number_of_workers is the number of ingestion workers to create
"""
if self.process_definition_id is None:
process_definition = ProcessDefinition(name='ingestion_worker_process', description='Worker transform process for ingestion of datasets')
process_definition.executable['module']='ion.processes.data.ingestion.ingestion_worker'
process_definition.executable['class'] = 'IngestionWorker'
self.process_definition_id = self.clients.process_dispatcher.create_process_definition(process_definition=process_definition)
# Give each ingestion configuration its own queue name to receive data on
exchange_name = 'ingestion_queue'
##------------------------------------------------------------------------------------
## declare our intent to subscribe to all messages on the exchange point
query = ExchangeQuery()
subscription_id = self.clients.pubsub_management.create_subscription(query=query,\
exchange_name=exchange_name, name='Ingestion subscription', description='Subscription for ingestion workers')
##------------------------------------------------------------------------------------------
# create an ingestion_configuration instance and update the registry
# @todo: right now sending in the exchange_point_id as the name...
ingestion_configuration = IngestionConfiguration( name = self.XP)
ingestion_configuration.description = '%s exchange point ingestion configuration' % self.XP
ingestion_configuration.number_of_workers = number_of_workers
if hdf_storage is not None:
ingestion_configuration.hdf_storage.update(hdf_storage)
if couch_storage is not None:
ingestion_configuration.couch_storage.update(couch_storage)
ingestion_configuration_id, _ = self.clients.resource_registry.create(ingestion_configuration)
self._launch_transforms(
ingestion_configuration.number_of_workers,
subscription_id,
ingestion_configuration_id,
ingestion_configuration,
self.process_definition_id
)
return ingestion_configuration_id
def test_setup_queues(self):
ingestion_config = IngestionConfiguration()
ingestion_config.queues = [IngestionQueue()]
setattr(ingestion_config,'_id','config_id')
self.pubsub_create_sub.return_value = 'subscription_id'
self.ingestion_management._existing_dataset = Mock()
retval = self.ingestion_management.setup_queues(ingestion_config, 'stream_id', 'dataset_id')
self.assertTrue(retval)
self.ingestion_management._existing_dataset.assert_called_once_with('stream_id','dataset_id')
def create_ingestion_configuration(self,name='', exchange_point_id='', queues=None):
validate_is_instance(queues,list,'The queues parameter is not a proper list.')
validate_true(len(queues)>0, 'Ingestion needs at least one queue to ingest from')
for queue in queues:
validate_is_instance(queue, IngestionQueue)
ingestion_config = IngestionConfiguration()
ingestion_config.name = name
ingestion_config.exchange_point = exchange_point_id
ingestion_config.queues = queues
config_id, rev = self.clients.resource_registry.create(ingestion_config)
return config_id
def test_setup_queues(self):
ingestion_config = IngestionConfiguration()
ingestion_config.queues = [IngestionQueue()]
setattr(ingestion_config, '_id', 'config_id')
self.pubsub_create_sub.return_value = 'subscription_id'
self.ingestion_management._existing_dataset = Mock()
retval = self.ingestion_management.setup_queues(
ingestion_config, 'stream_id', 'dataset_id')
self.assertTrue(retval)
self.ingestion_management._existing_dataset.assert_called_once_with(
'stream_id', 'dataset_id')
def create_ingestion_configuration(self,
name='',
exchange_point_id='',
queues=None):
validate_is_instance(queues, list,
'The queues parameter is not a proper list.')
validate_true(
len(queues) > 0,
'Ingestion needs at least one queue to ingest from')
for queue in queues:
validate_is_instance(queue, IngestionQueue)
ingestion_config = IngestionConfiguration()
ingestion_config.name = name
ingestion_config.exchange_point = exchange_point_id
ingestion_config.queues = queues
config_id, rev = self.clients.resource_registry.create(
ingestion_config)
return config_id
def test_persist_data_stream(self):
config = IngestionConfiguration()
self.ingestion_management.read_ingestion_configuration = Mock()
self.ingestion_management.read_ingestion_configuration.return_value = config
self.ingestion_management.is_persisted = Mock()
self.ingestion_management.is_persisted.return_value = False
self.ingestion_management.setup_queues = Mock()
self.ingestion_management.setup_queues.return_value = True
self.pubsub_read.return_value = DotDict(persisted=False)
retval = self.ingestion_management.persist_data_stream(
'stream_id', 'config_id', 'dataset_id')
self.assertEquals(retval, 'dataset_id')
