def wrapped(*args):
"""
Initialize logs, make sure the job is still running, and run the task
code surrounded by a try-except. If any error occurs, log it as a
critical failure.
"""
# the last argument is assumed to be a monitor
monitor = args[-1]
job = models.OqJob.objects.get(id=monitor.job_id)
if job.is_running is False:
# the job was killed, it is useless to run the task
raise JobNotRunning(monitor.job_id)
# it is important to save the task id soon, so that
# the revoke functionality can work
with monitor("storing task id", task=tsk, autoflush=True):
pass
with logs.handle(job):
check_mem_usage() # warn if too much memory is used
# run the task
try:
total = "total " + task_func.__name__
with monitor(total, task=tsk):
with GroundShakingIntensityModel.forbid_instantiation():
return task_func(*args)
finally:
# save on the db
CacheInserter.flushall()
# the task finished, we can remove from the performance
# table the associated row 'storing task id'
models.Performance.objects.filter(
oq_job=job, operation="storing task id", task_id=tsk.request.id
).delete()
def aggregate_result_set(self, agg, acc):
"""
Loop on a set of celery AsyncResults and update the accumulator
by using the aggregation function.
:param agg: the aggregation function, (acc, val) -> new acc
:param acc: the initial value of the accumulator
:returns: the final value of the accumulator
"""
if not self.results:
return acc
backend = current_app().backend
amqp_backend = backend.__class__.__name__.startswith("AMQP")
rset = ResultSet(self.results)
for task_id, result_dict in rset.iter_native():
check_mem_usage() # warn if too much memory is used
result = result_dict["result"]
if isinstance(result, BaseException):
raise result
self.received.append(len(result))
acc = agg(acc, result.unpickle())
if amqp_backend:
# work around a celery bug
del backend._cache[task_id]
return acc
def wrapped(*args):
"""
Initialize logs, make sure the job is still running, and run the task
code surrounded by a try-except. If any error occurs, log it as a
critical failure.
"""
# job_id is always assumed to be the first argument
job_id = args[0]
job = models.OqJob.objects.get(id=job_id)
if job.is_running is False:
# the job was killed, it is useless to run the task
raise JobNotRunning(job_id)
# it is important to save the task id soon, so that
# the revoke functionality can work
EnginePerformanceMonitor.store_task_id(job_id, tsk)
with EnginePerformanceMonitor(
'total ' + task_func.__name__, job_id, tsk, flush=True):
# tasks write on the celery log file
logs.set_level(job.log_level)
check_mem_usage() # log a warning if too much memory is used
try:
# run the task
return task_func(*args)
finally:
# save on the db
CacheInserter.flushall()
# the task finished, we can remove from the performance
# table the associated row 'storing task id'
models.Performance.objects.filter(
oq_job=job,
operation='storing task id',
task_id=tsk.request.id).delete()
def aggregate_result_set(self, agg, acc):
"""
Loop on a set of celery AsyncResults and update the accumulator
by using the aggregation function.
:param agg: the aggregation function, (acc, val) -> new acc
:param acc: the initial value of the accumulator
:returns: the final value of the accumulator
"""
if isinstance(self.oqtask, types.FunctionType):
# don't use celery
return super(OqTaskManager, self).aggregate_result_set(
agg, acc)
if not self.results:
return acc
backend = current_app().backend
amqp_backend = backend.__class__.__name__.startswith('AMQP')
rset = ResultSet(self.results)
for task_id, result_dict in rset.iter_native():
idx = self.task_ids.index(task_id)
self.task_ids.pop(idx)
parallel.check_mem_usage() # warn if too much memory is used
result = result_dict['result']
if isinstance(result, BaseException):
raise result
self.received.append(len(result))
acc = agg(acc, result.unpickle())
if amqp_backend:
# work around a celery bug
del backend._cache[task_id]
return acc
def aggregate_result_set(self, agg, acc):
"""
Loop on a set of celery AsyncResults and update the accumulator
by using the aggregation function.
:param agg: the aggregation function, (acc, val) -> new acc
:param acc: the initial value of the accumulator
:returns: the final value of the accumulator
"""
if isinstance(self.oqtask, types.FunctionType):
# don't use celery
return super(OqTaskManager,
self).aggregate_result_set(agg, acc)
if not self.results:
return acc
backend = current_app().backend
amqp_backend = backend.__class__.__name__.startswith('AMQP')
rset = ResultSet(self.results)
for task_id, result_dict in rset.iter_native():
idx = self.task_ids.index(task_id)
self.task_ids.pop(idx)
parallel.check_mem_usage() # warn if too much memory is used
result = result_dict['result']
if isinstance(result, BaseException):
raise result
self.received.append(len(result))
acc = agg(acc, result.unpickle())
if amqp_backend:
# work around a celery bug
del backend._cache[task_id]
return acc
def wrapped(*args):
"""
Initialize logs, make sure the job is still running, and run the task
code surrounded by a try-except. If any error occurs, log it as a
critical failure.
"""
# the last argument is assumed to be a monitor
monitor = args[-1]
job = models.OqJob.objects.get(id=monitor.job_id)
if job.is_running is False:
# the job was killed, it is useless to run the task
raise JobNotRunning(monitor.job_id)
# it is important to save the task id soon, so that
# the revoke functionality can work
with monitor('storing task id', task=tsk, autoflush=True):
pass
with logs.handle(job):
check_mem_usage() # warn if too much memory is used
# run the task
try:
total = 'total ' + task_func.__name__
with monitor(total, task=tsk, autoflush=True):
return task_func(*args)
finally:
# save on the db
CacheInserter.flushall()
# the task finished, we can remove from the performance
# table the associated row 'storing task id'
models.Performance.objects.filter(
oq_job=job,
operation='storing task id',
task_id=tsk.request.id).delete()
def submit(self, *args):
"""
Submit an oqtask with the given arguments to celery and return
an AsyncResult. If the variable OQ_NO_DISTRIBUTE is set, the
task function is run in process and the result is returned.
"""
check_mem_usage() # log a warning if too much memory is used
if no_distribute():
res = safely_call(self.oqtask.task_func, args)
else:
piks = pickle_sequence(args)
self.sent += sum(len(p) for p in piks)
res = self.oqtask.delay(*piks)
self.results.append(res)
def aggregate_result_set(self, agg, acc):
"""
Loop on a set of celery AsyncResults and update the accumulator
by using the aggregation function.
:param agg: the aggregation function, (acc, val) -> new acc
:param acc: the initial value of the accumulator
:returns: the final value of the accumulator
"""
if not self.results:
return acc
backend = current_app().backend
rset = ResultSet(self.results)
for task_id, result_dict in rset.iter_native():
check_mem_usage() # log a warning if too much memory is used
result = result_dict['result']
if isinstance(result, BaseException):
raise result
acc = agg(acc, result.unpickle())
del backend._cache[task_id] # work around a celery bug
return acc
def aggregate_result_set(self, agg, acc):
"""
Loop on a set of celery AsyncResults and update the accumulator
by using the aggregation function.
:param agg: the aggregation function, (acc, val) -> new acc
:param acc: the initial value of the accumulator
:returns: the final value of the accumulator
"""
if not self.results:
return acc
backend = current_app().backend
rset = ResultSet(self.results)
for task_id, result_dict in rset.iter_native():
check_mem_usage() # warn if too much memory is used
result = result_dict['result']
if isinstance(result, BaseException):
raise result
self.received += len(result)
acc = agg(acc, result.unpickle())
del backend._cache[task_id] # work around a celery bug
return acc