process.py

# Copyright 2009 Brian Quinlan. All Rights Reserved.
# Licensed to PSF under a Contributor Agreement.

"""Implements ProcessPoolExecutor.

The follow diagram and text describe the data-flow through the system:

|======================= In-process =====================|== Out-of-process ==|

+----------+ +----------+ +--------+ +-----------+ +---------+
| | => | Work Ids | | | | Call Q | | Process |
| | +----------+ | | +-----------+ | Pool |
| | | ... | | | | ... | +---------+
| | | 6 | => | | => | 5, call() | => | |
| | | 7 | | | | ... | | |
| Process | | ... | | Local | +-----------+ | Process |
| Pool | +----------+ | Worker | | #1..n |
| Executor | | Thread | | |
| | +----------- + | | +-----------+ | |
| | <=> | Work Items | <=> | | <= | Result Q | <= | |
| | +------------+ | | +-----------+ | |
| | | 6: call() | | | | ... | | |
| | | future | | | | 4, result | | |
| | | ... | | | | 3, except | | |
+----------+ +------------+ +--------+ +-----------+ +---------+

Executor.submit() called:
- creates a uniquely numbered _WorkItem and adds it to the "Work Items" dict
- adds the id of the _WorkItem to the "Work Ids" queue

Local worker thread:
- reads work ids from the "Work Ids" queue and looks up the corresponding
 WorkItem from the "Work Items" dict: if the work item has been cancelled then
 it is simply removed from the dict, otherwise it is repackaged as a
 _CallItem and put in the "Call Q". New _CallItems are put in the "Call Q"
 until "Call Q" is full. NOTE: the size of the "Call Q" is kept small because
 calls placed in the "Call Q" can no longer be cancelled with Future.cancel().
- reads _ResultItems from "Result Q", updates the future stored in the
 "Work Items" dict and deletes the dict entry

Process #1..n:
- reads _CallItems from "Call Q", executes the calls, and puts the resulting
 _ResultItems in "Result Q"
"""

__author__='Brian Quinlan (brian@sweetapp.com)'

importatexit
importos
fromconcurrent.futuresimport_base
importqueue
fromqueueimportFull
importmultiprocessingasmp
frommultiprocessing.connectionimportwait
frommultiprocessing.queuesimportQueue
importthreading
importweakref
fromfunctoolsimportpartial
importitertools
importsys
importtraceback

# Workers are created as daemon threads and processes. This is done to allow the
# interpreter to exit when there are still idle processes in a
# ProcessPoolExecutor's process pool (i.e. shutdown() was not called). However,
# allowing workers to die with the interpreter has two undesirable properties:
# - The workers would still be running during interpreter shutdown,
# meaning that they would fail in unpredictable ways.
# - The workers could be killed while evaluating a work item, which could
# be bad if the callable being evaluated has external side-effects e.g.
# writing to a file.
#
# To work around this problem, an exit handler is installed which tells the
# workers to exit when their work queues are empty and then waits until the
# threads/processes finish.

_threads_wakeups=weakref.WeakKeyDictionary()
_global_shutdown=False


class_ThreadWakeup:
def__init__(self):
self._reader, self._writer=mp.Pipe(duplex=False)

defclose(self):
self._writer.close()
self._reader.close()

defwakeup(self):
self._writer.send_bytes(b"")

defclear(self):
whileself._reader.poll():
self._reader.recv_bytes()


def_python_exit():
global_global_shutdown
_global_shutdown=True
items=list(_threads_wakeups.items())
for_, thread_wakeupinitems:
thread_wakeup.wakeup()
fort, _initems:
t.join()

# Controls how many more calls than processes will be queued in the call queue.
# A smaller number will mean that processes spend more time idle waiting for
# work while a larger number will make Future.cancel() succeed less frequently
# (Futures in the call queue cannot be cancelled).
EXTRA_QUEUED_CALLS=1


# On Windows, WaitForMultipleObjects is used to wait for processes to finish.
# It can wait on, at most, 63 objects. There is an overhead of two objects:
# - the result queue reader
# - the thread wakeup reader
_MAX_WINDOWS_WORKERS=63-2

# Hack to embed stringification of remote traceback in local traceback

class_RemoteTraceback(Exception):
def__init__(self, tb):
self.tb=tb
def__str__(self):
returnself.tb

class_ExceptionWithTraceback:
def__init__(self, exc, tb):
tb=traceback.format_exception(type(exc), exc, tb)
tb=''.join(tb)
self.exc=exc
self.tb='\n"""\n%s"""'%tb
def__reduce__(self):
return_rebuild_exc, (self.exc, self.tb)

def_rebuild_exc(exc, tb):
exc.__cause__=_RemoteTraceback(tb)
returnexc

class_WorkItem(object):
def__init__(self, future, fn, args, kwargs):
self.future=future
self.fn=fn
self.args=args
self.kwargs=kwargs

class_ResultItem(object):
def__init__(self, work_id, exception=None, result=None):
self.work_id=work_id
self.exception=exception
self.result=result

class_CallItem(object):
def__init__(self, work_id, fn, args, kwargs):
self.work_id=work_id
self.fn=fn
self.args=args
self.kwargs=kwargs


class_SafeQueue(Queue):
"""Safe Queue set exception to the future object linked to a job"""
def__init__(self, max_size=0, *, ctx, pending_work_items):
self.pending_work_items=pending_work_items
super().__init__(max_size, ctx=ctx)

def_on_queue_feeder_error(self, e, obj):
ifisinstance(obj, _CallItem):
tb=traceback.format_exception(type(e), e, e.__traceback__)
e.__cause__=_RemoteTraceback('\n"""\n{}"""'.format(''.join(tb)))
work_item=self.pending_work_items.pop(obj.work_id, None)
# work_item can be None if another process terminated. In this case,
# the queue_manager_thread fails all work_items with BrokenProcessPool
ifwork_itemisnotNone:
work_item.future.set_exception(e)
else:
super()._on_queue_feeder_error(e, obj)


def_get_chunks(*iterables, chunksize):
""" Iterates over zip()ed iterables in chunks. """
it=zip(*iterables)
whileTrue:
chunk=tuple(itertools.islice(it, chunksize))
ifnotchunk:
return
yieldchunk

def_process_chunk(fn, chunk):
""" Processes a chunk of an iterable passed to map.

 Runs the function passed to map() on a chunk of the
 iterable passed to map.

 This function is run in a separate process.

 """
return [fn(*args) forargsinchunk]


def_sendback_result(result_queue, work_id, result=None, exception=None):
"""Safely send back the given result or exception"""
try:
result_queue.put(_ResultItem(work_id, result=result,
exception=exception))
exceptBaseExceptionase:
exc=_ExceptionWithTraceback(e, e.__traceback__)
result_queue.put(_ResultItem(work_id, exception=exc))


def_process_worker(call_queue, result_queue, initializer, initargs):
"""Evaluates calls from call_queue and places the results in result_queue.

 This worker is run in a separate process.

 Args:
 call_queue: A ctx.Queue of _CallItems that will be read and
 evaluated by the worker.
 result_queue: A ctx.Queue of _ResultItems that will written
 to by the worker.
 initializer: A callable initializer, or None
 initargs: A tuple of args for the initializer
 """
ifinitializerisnotNone:
try:
initializer(*initargs)
exceptBaseException:
_base.LOGGER.critical('Exception in initializer:', exc_info=True)
# The parent will notice that the process stopped and
# mark the pool broken
return
whileTrue:
call_item=call_queue.get(block=True)
ifcall_itemisNone:
# Wake up queue management thread
result_queue.put(os.getpid())
return
try:
r=call_item.fn(*call_item.args, **call_item.kwargs)
exceptBaseExceptionase:
exc=_ExceptionWithTraceback(e, e.__traceback__)
_sendback_result(result_queue, call_item.work_id, exception=exc)
else:
_sendback_result(result_queue, call_item.work_id, result=r)

# Liberate the resource as soon as possible, to avoid holding onto
# open files or shared memory that is not needed anymore
delcall_item


def_add_call_item_to_queue(pending_work_items,
work_ids,
call_queue):
"""Fills call_queue with _WorkItems from pending_work_items.

 This function never blocks.

 Args:
 pending_work_items: A dict mapping work ids to _WorkItems e.g.
 {5: <_WorkItem...>, 6: <_WorkItem...>, ...}
 work_ids: A queue.Queue of work ids e.g. Queue([5, 6, ...]). Work ids
 are consumed and the corresponding _WorkItems from
 pending_work_items are transformed into _CallItems and put in
 call_queue.
 call_queue: A multiprocessing.Queue that will be filled with _CallItems
 derived from _WorkItems.
 """
whileTrue:
ifcall_queue.full():
return
try:
work_id=work_ids.get(block=False)
exceptqueue.Empty:
return
else:
work_item=pending_work_items[work_id]

ifwork_item.future.set_running_or_notify_cancel():
call_queue.put(_CallItem(work_id,
work_item.fn,
work_item.args,
work_item.kwargs),
block=True)
else:
delpending_work_items[work_id]
continue


def_queue_management_worker(executor_reference,
processes,
pending_work_items,
work_ids_queue,
call_queue,
result_queue,
thread_wakeup):
"""Manages the communication between this process and the worker processes.

 This function is run in a local thread.

 Args:
 executor_reference: A weakref.ref to the ProcessPoolExecutor that owns
 this thread. Used to determine if the ProcessPoolExecutor has been
 garbage collected and that this function can exit.
 process: A list of the ctx.Process instances used as
 workers.
 pending_work_items: A dict mapping work ids to _WorkItems e.g.
 {5: <_WorkItem...>, 6: <_WorkItem...>, ...}
 work_ids_queue: A queue.Queue of work ids e.g. Queue([5, 6, ...]).
 call_queue: A ctx.Queue that will be filled with _CallItems
 derived from _WorkItems for processing by the process workers.
 result_queue: A ctx.SimpleQueue of _ResultItems generated by the
 process workers.
 thread_wakeup: A _ThreadWakeup to allow waking up the
 queue_manager_thread from the main Thread and avoid deadlocks
 caused by permanently locked queues.
 """
executor=None

defshutting_down():
return (_global_shutdownorexecutorisNone
orexecutor._shutdown_thread)

defshutdown_worker():
# This is an upper bound on the number of children alive.
n_children_alive=sum(p.is_alive() forpinprocesses.values())
n_children_to_stop=n_children_alive
n_sentinels_sent=0
# Send the right number of sentinels, to make sure all children are
# properly terminated.
whilen_sentinels_sent<n_children_to_stopandn_children_alive>0:
foriinrange(n_children_to_stop-n_sentinels_sent):
try:
call_queue.put_nowait(None)
n_sentinels_sent+=1
exceptFull:
break
n_children_alive=sum(p.is_alive() forpinprocesses.values())

# Release the queue's resources as soon as possible.
call_queue.close()
# If .join() is not called on the created processes then
# some ctx.Queue methods may deadlock on Mac OS X.
forpinprocesses.values():
p.join()

result_reader=result_queue._reader
wakeup_reader=thread_wakeup._reader
readers= [result_reader, wakeup_reader]

whileTrue:
_add_call_item_to_queue(pending_work_items,
work_ids_queue,
call_queue)

# Wait for a result to be ready in the result_queue while checking
# that all worker processes are still running, or for a wake up
# signal send. The wake up signals come either from new tasks being
# submitted, from the executor being shutdown/gc-ed, or from the
# shutdown of the python interpreter.
worker_sentinels= [p.sentinelforpinprocesses.values()]
ready=wait(readers+worker_sentinels)

cause=None
is_broken=True
ifresult_readerinready:
try:
result_item=result_reader.recv()
is_broken=False
exceptBaseExceptionase:
cause=traceback.format_exception(type(e), e, e.__traceback__)

elifwakeup_readerinready:
is_broken=False
result_item=None
thread_wakeup.clear()
ifis_broken:
# Mark the process pool broken so that submits fail right now.
executor=executor_reference()
ifexecutorisnotNone:
executor._broken= ('A child process terminated '
'abruptly, the process pool is not '
'usable anymore')
executor._shutdown_thread=True
executor=None
bpe=BrokenProcessPool("A process in the process pool was "
"terminated abruptly while the future was "
"running or pending.")
ifcauseisnotNone:
bpe.__cause__=_RemoteTraceback(
f"\n'''\n{''.join(cause)}'''")
# All futures in flight must be marked failed
forwork_id, work_iteminpending_work_items.items():
work_item.future.set_exception(bpe)
# Delete references to object. See issue16284
delwork_item
pending_work_items.clear()
# Terminate remaining workers forcibly: the queues or their
# locks may be in a dirty state and block forever.
forpinprocesses.values():
p.terminate()
shutdown_worker()
return
ifisinstance(result_item, int):
# Clean shutdown of a worker using its PID
# (avoids marking the executor broken)
assertshutting_down()
p=processes.pop(result_item)
p.join()
ifnotprocesses:
shutdown_worker()
return
elifresult_itemisnotNone:
work_item=pending_work_items.pop(result_item.work_id, None)
# work_item can be None if another process terminated (see above)
ifwork_itemisnotNone:
ifresult_item.exception:
work_item.future.set_exception(result_item.exception)
else:
work_item.future.set_result(result_item.result)
# Delete references to object. See issue16284
delwork_item
# Delete reference to result_item
delresult_item

# Check whether we should start shutting down.
executor=executor_reference()
# No more work items can be added if:
# - The interpreter is shutting down OR
# - The executor that owns this worker has been collected OR
# - The executor that owns this worker has been shutdown.
ifshutting_down():
try:
# Flag the executor as shutting down as early as possible if it
# is not gc-ed yet.
ifexecutorisnotNone:
executor._shutdown_thread=True
# Since no new work items can be added, it is safe to shutdown
# this thread if there are no pending work items.
ifnotpending_work_items:
shutdown_worker()
return
exceptFull:
# This is not a problem: we will eventually be woken up (in
# result_queue.get()) and be able to send a sentinel again.
pass
executor=None


_system_limits_checked=False
_system_limited=None


def_check_system_limits():
global_system_limits_checked, _system_limited
if_system_limits_checked:
if_system_limited:
raiseNotImplementedError(_system_limited)
_system_limits_checked=True
try:
nsems_max=os.sysconf("SC_SEM_NSEMS_MAX")
except (AttributeError, ValueError):
# sysconf not available or setting not available
return
ifnsems_max==-1:
# indetermined limit, assume that limit is determined
# by available memory only
return
ifnsems_max>=256:
# minimum number of semaphores available
# according to POSIX
return
_system_limited= ("system provides too few semaphores (%d"
" available, 256 necessary)"%nsems_max)
raiseNotImplementedError(_system_limited)


def_chain_from_iterable_of_lists(iterable):
"""
 Specialized implementation of itertools.chain.from_iterable.
 Each item in *iterable* should be a list. This function is
 careful not to keep references to yielded objects.
 """
forelementiniterable:
element.reverse()
whileelement:
yieldelement.pop()


classBrokenProcessPool(_base.BrokenExecutor):
"""
 Raised when a process in a ProcessPoolExecutor terminated abruptly
 while a future was in the running state.
 """


classProcessPoolExecutor(_base.Executor):
def__init__(self, max_workers=None, mp_context=None,
initializer=None, initargs=()):
"""Initializes a new ProcessPoolExecutor instance.

 Args:
 max_workers: The maximum number of processes that can be used to
 execute the given calls. If None or not given then as many
 worker processes will be created as the machine has processors.
 mp_context: A multiprocessing context to launch the workers. This
 object should provide SimpleQueue, Queue and Process.
 initializer: A callable used to initialize worker processes.
 initargs: A tuple of arguments to pass to the initializer.
 """
_check_system_limits()

ifmax_workersisNone:
self._max_workers=os.cpu_count() or1
ifsys.platform=='win32':
self._max_workers=min(_MAX_WINDOWS_WORKERS,
self._max_workers)
else:
ifmax_workers<=0:
raiseValueError("max_workers must be greater than 0")
elif (sys.platform=='win32'and
max_workers>_MAX_WINDOWS_WORKERS):
raiseValueError(
f"max_workers must be <= {_MAX_WINDOWS_WORKERS}")

self._max_workers=max_workers

ifmp_contextisNone:
mp_context=mp.get_context()
self._mp_context=mp_context

ifinitializerisnotNoneandnotcallable(initializer):
raiseTypeError("initializer must be a callable")
self._initializer=initializer
self._initargs=initargs

# Management thread
self._queue_management_thread=None

# Map of pids to processes
self._processes= {}

# Shutdown is a two-step process.
self._shutdown_thread=False
self._shutdown_lock=threading.Lock()
self._broken=False
self._queue_count=0
self._pending_work_items= {}

# Create communication channels for the executor
# Make the call queue slightly larger than the number of processes to
# prevent the worker processes from idling. But don't make it too big
# because futures in the call queue cannot be cancelled.
queue_size=self._max_workers+EXTRA_QUEUED_CALLS
self._call_queue=_SafeQueue(
max_size=queue_size, ctx=self._mp_context,
pending_work_items=self._pending_work_items)
# Killed worker processes can produce spurious "broken pipe"
# tracebacks in the queue's own worker thread. But we detect killed
# processes anyway, so silence the tracebacks.
self._call_queue._ignore_epipe=True
self._result_queue=mp_context.SimpleQueue()
self._work_ids=queue.Queue()

# _ThreadWakeup is a communication channel used to interrupt the wait
# of the main loop of queue_manager_thread from another thread (e.g.
# when calling executor.submit or executor.shutdown). We do not use the
# _result_queue to send the wakeup signal to the queue_manager_thread
# as it could result in a deadlock if a worker process dies with the
# _result_queue write lock still acquired.
self._queue_management_thread_wakeup=_ThreadWakeup()

def_start_queue_management_thread(self):
ifself._queue_management_threadisNone:
# When the executor gets garbarge collected, the weakref callback
# will wake up the queue management thread so that it can terminate
# if there is no pending work item.
defweakref_cb(_,
thread_wakeup=self._queue_management_thread_wakeup):
mp.util.debug('Executor collected: triggering callback for'
' QueueManager wakeup')
thread_wakeup.wakeup()
# Start the processes so that their sentinels are known.
self._adjust_process_count()
self._queue_management_thread=threading.Thread(
target=_queue_management_worker,
args=(weakref.ref(self, weakref_cb),
self._processes,
self._pending_work_items,
self._work_ids,
self._call_queue,
self._result_queue,
self._queue_management_thread_wakeup),
name="QueueManagerThread")
self._queue_management_thread.daemon=True
self._queue_management_thread.start()
_threads_wakeups[self._queue_management_thread] = \
self._queue_management_thread_wakeup

def_adjust_process_count(self):
for_inrange(len(self._processes), self._max_workers):
p=self._mp_context.Process(
target=_process_worker,
args=(self._call_queue,
self._result_queue,
self._initializer,
self._initargs))
p.start()
self._processes[p.pid] =p

defsubmit(*args, **kwargs):
iflen(args) >=2:
self, fn, *args=args
elifnotargs:
raiseTypeError("descriptor 'submit' of 'ProcessPoolExecutor' object "
"needs an argument")
elif'fn'inkwargs:
fn=kwargs.pop('fn')
self, *args=args
else:
raiseTypeError('submit expected at least 1 positional argument, '
'got %d'% (len(args)-1))

withself._shutdown_lock:
ifself._broken:
raiseBrokenProcessPool(self._broken)
ifself._shutdown_thread:
raiseRuntimeError('cannot schedule new futures after shutdown')
if_global_shutdown:
raiseRuntimeError('cannot schedule new futures after '
'interpreter shutdown')

f=_base.Future()
w=_WorkItem(f, fn, args, kwargs)

self._pending_work_items[self._queue_count] =w
self._work_ids.put(self._queue_count)
self._queue_count+=1
# Wake up queue management thread
self._queue_management_thread_wakeup.wakeup()

self._start_queue_management_thread()
returnf
submit.__doc__=_base.Executor.submit.__doc__

defmap(self, fn, *iterables, timeout=None, chunksize=1):
"""Returns an iterator equivalent to map(fn, iter).

 Args:
 fn: A callable that will take as many arguments as there are
 passed iterables.
 timeout: The maximum number of seconds to wait. If None, then there
 is no limit on the wait time.
 chunksize: If greater than one, the iterables will be chopped into
 chunks of size chunksize and submitted to the process pool.
 If set to one, the items in the list will be sent one at a time.

 Returns:
 An iterator equivalent to: map(func, *iterables) but the calls may
 be evaluated out-of-order.

 Raises:
 TimeoutError: If the entire result iterator could not be generated
 before the given timeout.
 Exception: If fn(*args) raises for any values.
 """
ifchunksize<1:
raiseValueError("chunksize must be >= 1.")

results=super().map(partial(_process_chunk, fn),
_get_chunks(*iterables, chunksize=chunksize),
timeout=timeout)
return_chain_from_iterable_of_lists(results)

defshutdown(self, wait=True):
withself._shutdown_lock:
self._shutdown_thread=True
ifself._queue_management_thread:
# Wake up queue management thread
self._queue_management_thread_wakeup.wakeup()
ifwait:
self._queue_management_thread.join()
# To reduce the risk of opening too many files, remove references to
# objects that use file descriptors.
self._queue_management_thread=None
ifself._call_queueisnotNone:
self._call_queue.close()
ifwait:
self._call_queue.join_thread()
self._call_queue=None
self._result_queue=None
self._processes=None

ifself._queue_management_thread_wakeup:
self._queue_management_thread_wakeup.close()
self._queue_management_thread_wakeup=None

shutdown.__doc__=_base.Executor.shutdown.__doc__

atexit.register(_python_exit)