def build_detectors(self, inbox):
logs.context['step'] = 'build_detectors'
detectors = [Detector(inbox) for Detector in self.features]
verbose = self.should_enable_extra_debugging(inbox, detectors)
for i, detector in enumerate(detectors):
logs.context['detector'] = type(detector).__name__
detector.create_sender_profile(self.sender_profile_num_emails)
logs.Watchdog.reset()
if verbose:
progress_logger.info('Finished creating {} sender profile, RSS = {}'.format(type(detector).__name__, MemTracker.cur_mem_usage()))
MemTracker.logMemory('finished creating {} sender profile'.format(type(detector).__name__))
else:
logs.RateLimitedMemTracker.checkmem('finished creating {} sender profile'.format(type(detector).__name__))
del logs.context['step']
del logs.context['detector']
return detectors
def should_enable_extra_debugging(self, inbox, detectors):
if len(inbox) < 10000:
return False
detector_names = ', '.join([type(d).__name__ for d in detectors])
progress_logger.info('Enabling extra debugging for large inbox with {} messages, {}; RSS = {}, creating sender profiles for {}'.format(len(inbox), logs.context, MemTracker.cur_mem_usage(), detector_names))
return True
def build_detectors(self, inbox):
logs.context['step'] = 'build_detectors'
detectors = [Detector(inbox) for Detector in self.features]
verbose = self.should_enable_extra_debugging(inbox, detectors)
for i, detector in enumerate(detectors):
logs.context['detector'] = type(detector).__name__
detector.create_sender_profile(self.sender_profile_indeces)
logs.Watchdog.reset()
if verbose:
progress_logger.info('Finished creating {} sender profile, RSS = {}'.format(type(detector).__name__, MemTracker.cur_mem_usage()))
MemTracker.logMemory('finished creating {} sender profile'.format(type(detector).__name__))
else:
logs.RateLimitedMemTracker.checkmem('finished creating {} sender profile'.format(type(detector).__name__))
del logs.context['step']
del logs.context['detector']
return detectors
def should_enable_extra_debugging(self, inbox, detectors):
if len(inbox) < 10000:
return False
detector_names = ', '.join([type(d).__name__ for d in detectors])
progress_logger.info('Enabling extra debugging for large inbox with {} messages, {}; RSS = {}, creating sender profiles for {}'.format(len(inbox), logs.context, MemTracker.cur_mem_usage(), detector_names))
return True
def checkmem_rate_limited(self):
cur = MemTracker.cur_mem_usage()
if cur >= 2*self.max_mem_logged:
self.max_mem_logged = cur
MemTracker.logMemory(self.task + "; " + str(context))
def checkmem_rate_limited(self):
cur = MemTracker.cur_mem_usage()
if cur >= 2 * self.max_mem_logged:
self.max_mem_logged = cur
MemTracker.logMemory(self.task + "; " + str(context))
