def build_task_descriptor(cls,
task_descriptor,
parent_task_record,
args,
n_outputs,
is_tail_spawn=False,
handler_name=None):
# This is needed to work around the fact that stdinout has its own implementation of build_task_descriptor, so
# we can't rely using cls.handler_name to find the actual executor.
if handler_name is None:
handler_name = cls.handler_name
if is_tail_spawn and len(
task_descriptor["expected_outputs"]) != n_outputs:
raise BlameUserException(
"SimpleExecutor being built with delegated outputs %s but n_outputs=%d"
% (task_descriptor["expected_outputs"], n_outputs))
# Throw early if the args are bad
cls.check_args_valid(args, n_outputs)
# Discover required ref IDs for this executor
reqd_refs = cls.get_required_refs(args)
task_descriptor["dependencies"].extend(reqd_refs)
sha = hashlib.sha1()
hash_update_with_structure(sha, [args, n_outputs])
name_prefix = "%s:%s:" % (handler_name, sha.hexdigest())
# Name our outputs
if not is_tail_spawn:
task_descriptor["expected_outputs"] = [
"%s%d" % (name_prefix, i) for i in range(n_outputs)
]
# Add the args dict
args_name = "%ssimple_exec_args" % name_prefix
args_ref = ref_from_object(args, "pickle", args_name)
parent_task_record.publish_ref(args_ref)
task_descriptor["dependencies"].append(args_ref)
task_descriptor["task_private"]["simple_exec_args"] = args_ref
BaseExecutor.build_task_descriptor(task_descriptor, parent_task_record)
if is_tail_spawn:
return None
else:
return [
SW2_FutureReference(x)
for x in task_descriptor["expected_outputs"]
]
def run(self):
ciel.engine.publish("worker_event", "Start execution " + repr(self.task_descriptor['task_id']) + " with handler " + self.task_descriptor['handler'])
ciel.log.error("Starting task %s with handler %s" % (str(self.task_descriptor['task_id']), self.task_descriptor['handler']), 'TASK', logging.DEBUG, False)
try:
self.start_time = datetime.datetime.now()
# Need to do this to bring task_private into the execution context.
BaseExecutor.prepare_task_descriptor_for_execute(self.task_descriptor, self, self.block_store)
if "package_ref" in self.task_descriptor["task_private"]:
self.package_ref = self.task_descriptor["task_private"]["package_ref"]
else:
self.package_ref = None
with self._executor_lock:
if self.aborted:
raise AbortedException()
else:
self.executor = self.execution_features.get_executor(self.task_descriptor["handler"], self.worker)
self.executor.run(self.task_descriptor, self)
self.finish_time = datetime.datetime.now()
self.success = not self.failed
ciel.engine.publish("worker_event", "Completed execution " + repr(self.task_descriptor['task_id']))
ciel.log.error("Completed task %s with handler %s" % (str(self.task_descriptor['task_id']), self.task_descriptor['handler']), 'TASK', logging.DEBUG, False)
except MissingInputException as mie:
ciel.log.error('Missing input in task %s with handler %s' % (str(self.task_descriptor['task_id']), self.task_descriptor['handler']), 'TASKEXEC', logging.ERROR, True)
self.failure_bindings = mie.bindings
self.failure_details = ""
self.failure_reason = "MISSING_INPUT"
self.finish_time = datetime.datetime.now()
self.success = False
raise
except AbortedException:
self.finish_time = datetime.datetime.now()
self.success = False
raise
except Exception, e:
ciel.log.error("Error in task %s with handler %s" % (str(self.task_descriptor['task_id']), self.task_descriptor['handler']), 'TASK', logging.ERROR, True)
self.failure_bindings = dict()
self.failure_details = getattr(e, "message", '')
self.failure_reason = "RUNTIME_EXCEPTION"
self.finish_time = datetime.datetime.now()
self.success = False
raise
def build_task_descriptor(cls, task_descriptor, parent_task_record, args, n_outputs, is_tail_spawn=False, handler_name=None):
# This is needed to work around the fact that stdinout has its own implementation of build_task_descriptor, so
# we can't rely using cls.handler_name to find the actual executor.
if handler_name is None:
handler_name = cls.handler_name
if is_tail_spawn and len(task_descriptor["expected_outputs"]) != n_outputs:
raise BlameUserException("SimpleExecutor being built with delegated outputs %s but n_outputs=%d" % (task_descriptor["expected_outputs"], n_outputs))
# Throw early if the args are bad
cls.check_args_valid(args, n_outputs)
# Discover required ref IDs for this executor
reqd_refs = cls.get_required_refs(args)
task_descriptor["dependencies"].extend(reqd_refs)
sha = hashlib.sha1()
hash_update_with_structure(sha, [args, n_outputs])
name_prefix = "%s:%s:" % (handler_name, sha.hexdigest())
# Name our outputs
if not is_tail_spawn:
task_descriptor["expected_outputs"] = ["%s%d" % (name_prefix, i) for i in range(n_outputs)]
# Add the args dict
args_name = "%ssimple_exec_args" % name_prefix
args_ref = ref_from_object(args, "pickle", args_name)
parent_task_record.publish_ref(args_ref)
task_descriptor["dependencies"].append(args_ref)
task_descriptor["task_private"]["simple_exec_args"] = args_ref
BaseExecutor.build_task_descriptor(task_descriptor, parent_task_record)
if is_tail_spawn:
return None
else:
return [SW2_FutureReference(x) for x in task_descriptor["expected_outputs"]]
def __init__(self, worker):
BaseExecutor.__init__(self, worker)
def __init__(self, worker):
BaseExecutor.__init__(self, worker)
self.process_pool = worker.process_pool
self.ongoing_fetches = []
self.ongoing_outputs = dict()
self.transmit_lock = threading.Lock()
def __init__(self, worker):
BaseExecutor.__init__(self, worker)
def __init__(self, worker):
BaseExecutor.__init__(self, worker)
self.process_pool = worker.process_pool
self.ongoing_fetches = []
self.ongoing_outputs = dict()
self.transmit_lock = threading.Lock()
def run(self):
ciel.engine.publish(
"worker_event",
"Start execution " + repr(self.task_descriptor['task_id']) +
" with handler " + self.task_descriptor['handler'])
ciel.log.error(
"Starting task %s with handler %s" %
(str(self.task_descriptor['task_id']),
self.task_descriptor['handler']), 'TASK', logging.DEBUG, False)
try:
self.start_time = datetime.datetime.now()
# Need to do this to bring task_private into the execution context.
BaseExecutor.prepare_task_descriptor_for_execute(
self.task_descriptor, self, self.block_store)
if "package_ref" in self.task_descriptor["task_private"]:
self.package_ref = self.task_descriptor["task_private"][
"package_ref"]
else:
self.package_ref = None
with self._executor_lock:
if self.aborted:
raise AbortedException()
else:
self.executor = self.execution_features.get_executor(
self.task_descriptor["handler"], self.worker)
self.executor.run(self.task_descriptor, self)
self.finish_time = datetime.datetime.now()
self.success = not self.failed
ciel.engine.publish(
"worker_event",
"Completed execution " + repr(self.task_descriptor['task_id']))
ciel.log.error(
"Completed task %s with handler %s" %
(str(self.task_descriptor['task_id']),
self.task_descriptor['handler']), 'TASK', logging.DEBUG,
False)
except MissingInputException as mie:
ciel.log.error(
'Missing input in task %s with handler %s' %
(str(self.task_descriptor['task_id']),
self.task_descriptor['handler']), 'TASKEXEC', logging.ERROR,
True)
self.failure_bindings = mie.bindings
self.failure_details = ""
self.failure_reason = "MISSING_INPUT"
self.finish_time = datetime.datetime.now()
self.success = False
raise
except AbortedException:
self.finish_time = datetime.datetime.now()
self.success = False
raise
except Exception, e:
ciel.log.error(
"Error in task %s with handler %s" %
(str(self.task_descriptor['task_id']),
self.task_descriptor['handler']), 'TASK', logging.ERROR, True)
self.failure_bindings = dict()
self.failure_details = getattr(e, "message", '')
self.failure_reason = "RUNTIME_EXCEPTION"
self.finish_time = datetime.datetime.now()
self.success = False
raise
