def __init__(
self,
control_address, # type: str
credentials=None,
worker_id=None, # type: Optional[str]
# Caching is disabled by default
state_cache_size=0,
# time-based data buffering is disabled by default
data_buffer_time_limit_ms=0,
profiler_factory=None, # type: Optional[Callable[..., Profile]]
status_address=None, # type: Optional[str]
):
self._alive = True
self._worker_index = 0
self._worker_id = worker_id
self._state_cache = StateCache(state_cache_size)
if credentials is None:
_LOGGER.info('Creating insecure control channel for %s.',
control_address)
self._control_channel = GRPCChannelFactory.insecure_channel(
control_address)
else:
_LOGGER.info('Creating secure control channel for %s.',
control_address)
self._control_channel = GRPCChannelFactory.secure_channel(
control_address, credentials)
grpc.channel_ready_future(self._control_channel).result(timeout=60)
_LOGGER.info('Control channel established.')
self._control_channel = grpc.intercept_channel(
self._control_channel, WorkerIdInterceptor(self._worker_id))
self._data_channel_factory = data_plane.GrpcClientDataChannelFactory(
credentials, self._worker_id, data_buffer_time_limit_ms)
self._state_handler_factory = GrpcStateHandlerFactory(
self._state_cache, credentials)
self._profiler_factory = profiler_factory
self._fns = KeyedDefaultDict(
lambda id: self._control_stub.GetProcessBundleDescriptor(
beam_fn_api_pb2.GetProcessBundleDescriptorRequest(
process_bundle_descriptor_id=id)
)) # type: Mapping[str, beam_fn_api_pb2.ProcessBundleDescriptor]
# BundleProcessor cache across all workers.
self._bundle_processor_cache = BundleProcessorCache(
state_handler_factory=self._state_handler_factory,
data_channel_factory=self._data_channel_factory,
fns=self._fns)
if status_address:
try:
self._status_handler = FnApiWorkerStatusHandler(
status_address, self._bundle_processor_cache
) # type: Optional[FnApiWorkerStatusHandler]
except Exception:
traceback_string = traceback.format_exc()
_LOGGER.warning(
'Error creating worker status request handler, '
'skipping status report. Trace back: %s' %
traceback_string)
else:
self._status_handler = None
# TODO(BEAM-8998) use common
# thread_pool_executor.shared_unbounded_instance() to process bundle
# progress once dataflow runner's excessive progress polling is removed.
self._report_progress_executor = futures.ThreadPoolExecutor(
max_workers=1)
self._worker_thread_pool = thread_pool_executor.shared_unbounded_instance(
)
self._responses = queue.Queue(
) # type: queue.Queue[beam_fn_api_pb2.InstructionResponse]
_LOGGER.info(
'Initializing SDKHarness with unbounded number of workers.')
def default_factory(id):
# type: (str) -> beam_fn_api_pb2.ProcessBundleDescriptor
return self._control_stub.GetProcessBundleDescriptor(
beam_fn_api_pb2.GetProcessBundleDescriptorRequest(
process_bundle_descriptor_id=id))