from __future__ import print_function, division, absolute_import import contextlib from uuid import uuid4 import weakref from .parallel import AutoBatchingMixin, ParallelBackendBase, BatchedCalls from .parallel import parallel_backend try: import distributed except ImportError: distributed = None if distributed is not None: from distributed.client import Client, _wait from distributed.utils import funcname, itemgetter from distributed import get_client, secede, rejoin from distributed.worker import thread_state from distributed.sizeof import sizeof from tornado import gen def is_weakrefable(obj): try: weakref.ref(obj) return True except TypeError: return False try: TimeoutError = TimeoutError except NameError: # Python 2 backward compat class TimeoutError(OSError): pass class _WeakKeyDictionary: """A variant of weakref.WeakKeyDictionary for unhashable objects. This datastructure is used to store futures for broadcasted data objects such as large numpy arrays or pandas dataframes that are not hashable and therefore cannot be used as keys of traditional python dicts. Futhermore using a dict with id(array) as key is not safe because the Python is likely to reuse id of recently collected arrays. """ def __init__(self): self._data = {} def __getitem__(self, obj): ref, val = self._data[id(obj)] if ref() is not obj: # In case of a race condition with on_destroy. raise KeyError(obj) return val def __setitem__(self, obj, value): key = id(obj) try: ref, _ = self._data[key] if ref() is not obj: # In case of race condition with on_destroy. raise KeyError(obj) except KeyError: # Insert the new entry in the mapping along with a weakref # callback to automatically delete the entry from the mapping # as soon as the object used as key is garbage collected. def on_destroy(_): del self._data[key] ref = weakref.ref(obj, on_destroy) self._data[key] = ref, value def __len__(self): return len(self._data) def clear(self): self._data.clear() def _funcname(x): try: if isinstance(x, BatchedCalls): x = x.items[0][0] except Exception: pass return funcname(x) class Batch(object): def __init__(self, tasks): self.tasks = tasks def __call__(self, *data): results = [] with parallel_backend('dask'): for func, args, kwargs in self.tasks: args = [a(data) if isinstance(a, itemgetter) else a for a in args] kwargs = {k: v(data) if isinstance(v, itemgetter) else v for (k, v) in kwargs.items()} results.append(func(*args, **kwargs)) return results def __reduce__(self): return Batch, (self.tasks,) def _joblib_probe_task(): # Noop used by the joblib connector to probe when workers are ready. pass class DaskDistributedBackend(ParallelBackendBase, AutoBatchingMixin): MIN_IDEAL_BATCH_DURATION = 0.2 MAX_IDEAL_BATCH_DURATION = 1.0 def __init__(self, scheduler_host=None, scatter=None, client=None, loop=None, wait_for_workers_timeout=10, **submit_kwargs): if distributed is None: msg = ("You are trying to use 'dask' as a joblib parallel backend " "but dask is not installed. Please install dask " "to fix this error.") raise ValueError(msg) if client is None: if scheduler_host: client = Client(scheduler_host, loop=loop, set_as_default=False) else: try: client = get_client() except ValueError: msg = ("To use Joblib with Dask first create a Dask Client" "\n\n" " from dask.distributed import Client\n" " client = Client()\n" "or\n" " client = Client('scheduler-address:8786')") raise ValueError(msg) self.client = client if scatter is not None and not isinstance(scatter, (list, tuple)): raise TypeError("scatter must be a list/tuple, got " "`%s`" % type(scatter).__name__) if scatter is not None and len(scatter) > 0: # Keep a reference to the scattered data to keep the ids the same self._scatter = list(scatter) scattered = self.client.scatter(scatter, broadcast=True) self.data_futures = {id(x): f for x, f in zip(scatter, scattered)} else: self._scatter = [] self.data_futures = {} self.task_futures = set() self.wait_for_workers_timeout = wait_for_workers_timeout self.submit_kwargs = submit_kwargs def __reduce__(self): return (DaskDistributedBackend, ()) def get_nested_backend(self): return DaskDistributedBackend(client=self.client), -1 def configure(self, n_jobs=1, parallel=None, **backend_args): return self.effective_n_jobs(n_jobs) def start_call(self): self.call_data_futures = _WeakKeyDictionary() def stop_call(self): # The explicit call to clear is required to break a cycling reference # to the futures. self.call_data_futures.clear() def effective_n_jobs(self, n_jobs): effective_n_jobs = sum(self.client.ncores().values()) if effective_n_jobs != 0 or not self.wait_for_workers_timeout: return effective_n_jobs # If there is no worker, schedule a probe task to wait for the workers # to come up and be available. If the dask cluster is in adaptive mode # task might cause the cluster to provision some workers. try: self.client.submit(_joblib_probe_task).result( timeout=self.wait_for_workers_timeout) except gen.TimeoutError: error_msg = ( "DaskDistributedBackend has no worker after {} seconds. " "Make sure that workers are started and can properly connect " "to the scheduler and increase the joblib/dask connection " "timeout with:\n\n" "parallel_backend('dask', wait_for_workers_timeout={})" ).format(self.wait_for_workers_timeout, max(10, 2 * self.wait_for_workers_timeout)) raise TimeoutError(error_msg) return sum(self.client.ncores().values()) def _to_func_args(self, func): collected_futures = [] itemgetters = dict() # Futures that are dynamically generated during a single call to # Parallel.__call__. call_data_futures = getattr(self, 'call_data_futures', None) def maybe_to_futures(args): for arg in args: arg_id = id(arg) if arg_id in itemgetters: yield itemgetters[arg_id] continue f = self.data_futures.get(arg_id, None) if f is None and call_data_futures is not None: try: f = call_data_futures[arg] except KeyError: if is_weakrefable(arg) and sizeof(arg) > 1e3: # Automatically scatter large objects to some of # the workers to avoid duplicated data transfers. # Rely on automated inter-worker data stealing if # more workers need to reuse this data # concurrently. [f] = self.client.scatter([arg]) call_data_futures[arg] = f if f is not None: getter = itemgetter(len(collected_futures)) collected_futures.append(f) itemgetters[arg_id] = getter arg = getter yield arg tasks = [] for f, args, kwargs in func.items: args = list(maybe_to_futures(args)) kwargs = dict(zip(kwargs.keys(), maybe_to_futures(kwargs.values()))) tasks.append((f, args, kwargs)) if not collected_futures: return func, () return (Batch(tasks), collected_futures) def apply_async(self, func, callback=None): key = '%s-batch-%s' % (_funcname(func), uuid4().hex) func, args = self._to_func_args(func) future = self.client.submit(func, *args, key=key, **self.submit_kwargs) self.task_futures.add(future) def callback_wrapper(future): result = future.result() self.task_futures.remove(future) if callback is not None: callback(result) future.add_done_callback(callback_wrapper) ref = weakref.ref(future) # avoid reference cycle def get(): return ref().result() future.get = get # monkey patch to achieve AsyncResult API return future def abort_everything(self, ensure_ready=True): """ Tell the client to cancel any task submitted via this instance joblib.Parallel will never access those results """ self.client.cancel(self.task_futures) self.task_futures.clear() @contextlib.contextmanager def retrieval_context(self): """Override ParallelBackendBase.retrieval_context to avoid deadlocks. This removes thread from the worker's thread pool (using 'secede'). Seceding avoids deadlock in nested parallelism settings. """ # See 'joblib.Parallel.__call__' and 'joblib.Parallel.retrieve' for how # this is used. if hasattr(thread_state, 'execution_state'): # we are in a worker. Secede to avoid deadlock. secede() yield if hasattr(thread_state, 'execution_state'): rejoin()