def initial_startup():
"""
Calls the init() method in the model and prepares the model to receive predictions. The init
task may take a long time to complete, so the settings field ready_to_predict will be updated
asynchronously when init() completes. This will also begin the background registration task
to the server.
:return: {"status": "success"} upon startup completion. No guarantee that init() is done processing.
"""
# Run startup task async
load_dotenv()
# Register the model to the server in a separate thread to avoid meddling with
# initializing the service which might be used directly by other client later on
# We will only run the registration once the model init is complete.
def init_model_helper():
logger.debug('Beginning Model Initialization Process.')
init()
model_settings.ready_to_predict = True
logger.debug('Finishing Model Initialization Process.')
pool.submit(register_model_to_server, os.getenv('SERVER_PORT'),
os.getenv('PORT'), os.getenv('NAME'))
pool.submit(init_model_helper)
return {"status": "success", 'detail': 'server startup in progress'}
def init_model_helper():
logger.debug('Beginning Model Initialization Process.')
init()
model_settings.ready_to_predict = True
logger.debug('Finishing Model Initialization Process.')
pool.submit(register_model_to_server, os.getenv('SERVER_PORT'),
os.getenv('PORT'), os.getenv('NAME'))
def init_dataset_helper():
logger.debug('Beginning Dataset Initialization Process.')
setup_dataset()
dataset_settings.ready_to_train = True
logger.debug('Finishing Dataset Initialization Process.')
pool.submit(register_model_to_server, os.getenv('SERVER_PORT'),
os.getenv('PORT'), os.getenv('DATASET_NAME'))