def test_MPEngine_dropProducers(self):
try:
print "Starting test"
maxCores = 6
mpe = MPEngineProdCons(maxCores, WkrTestProd, WkrTestCons)
# Add tasks
task_list = [i for i in xrange(1, 10)]
mpe.addTaskList(task_list)
mpe.startProducers(4)
mpe.addProducer()
mpe.addProducer()
mpe.removeProducer()
mpe.removeProducer()
mpe.startConsumers(1)
while mpe.working():
print("Prod: %d / Cons: %d | %s -> %s -> %s" %
mpe.getProgress())
time.sleep(1)
mpe.rebalance()
print "Test ended"
except Exception:
traceback.print_exc(file=sys.stdout)
self.fail("Exception triggered")
# Pass
self.assertEquals(1, 1, "test_MPEngine")
def test_MPEngine_regressionBug(self):
try:
print "Starting test"
maxCores = 4
mpe = MPEngineProdCons(maxCores, WkrTestProd, WkrTestCons)
# Add tasks
task_list = [i for i in xrange(1, 8)]
mpe.addTaskList(task_list)
mpe.startProducers(3)
time.sleep(2)
mpe.startConsumers(1)
time.sleep(2)
mpe.endProducers()
mpe.endConsumers()
time.sleep(2)
mpe.addConsumer()
mpe.addProducer()
time.sleep(2)
mpe.addProducer()
time.sleep(2)
logger.debug("restartConsumers")
mpe.restartConsumers()
logger.debug("restartProducers")
mpe.restartProducers()
time.sleep(2)
while mpe.working():
print("Prod: %d / Cons: %d | %s -> %s -> %s" %
mpe.getProgress())
time.sleep(1)
# mpe.rebalance()
del mpe
print "Test ended"
except Exception:
traceback.print_exc(file=sys.stdout)
self.fail("Exception triggered")
# Pass
self.assertEquals(1, 1, "test_MPEngine")
def appLoadMP(pathToLoad, dbfilenameFullPath, maxCores, governorOffFlag):
global _tasksPerJob
files_to_process = []
conn = None
# Start timer
t0 = datetime.now()
logger.debug("Starting appLoadMP")
# Calculate aggreagate file_filter for all ingest types supported:
file_filter = '|'.join([v.getFileNameFilter() for k,v in ingest_plugins.iteritems()])
# Add zip extension
file_filter += "|.*\.zip"
# Check if we're loading Redline data
if os.path.isdir(pathToLoad) and os.path.basename(pathToLoad).lower() == 'RedlineAudits'.lower():
files_to_process = searchRedLineAudits(pathToLoad)
else:
# Search for all files to be processed
if os.path.isdir(pathToLoad):
files_to_process = searchFolders(pathToLoad, file_filter)
else:
files_to_process = processArchives(pathToLoad, file_filter)
if files_to_process:
# Init DB if required
DB = appDB.DBClass(dbfilenameFullPath, True, settings.__version__)
conn = DB.appConnectDB()
# Extract hostnames, grab existing host IDs from DB and calculate instance ID for new IDs to be ingested:
instancesToProcess = []
instancesToProcess += GetIDForHosts(files_to_process, DB)
countInstancesToProcess = len(instancesToProcess)
logger.info("Found %d new instances" % (countInstancesToProcess))
# Setup producers/consumers initial counts
num_consumers = 1
num_producers = 1
# Setup MPEngine
mpe = MPEngineProdCons(maxCores, appLoadProd, appLoadCons, governorOffFlag)
# Reduce _tasksPerJob for small jobs
if countInstancesToProcess < _tasksPerJob: _tasksPerJob = 1
# Create task list
task_list = []
instancesPerJob = _tasksPerJob
num_tasks = 0
for chunk in chunks(instancesToProcess, instancesPerJob):
# todo: We no longer need pathToLoad as tasks include the fullpath now
task_list.append(Task(pathToLoad, chunk))
num_tasks += 1
if num_tasks > 0:
# Check if we have to drop indexes to speedup insertions
# todo: Research ratio of existing hosts to new hosts were this makes sense
if countInstancesToProcess > 1000 or DB.CountHosts() < 1000:
DB.appDropIndexesDB()
# Queue tasks for Producers
mpe.addTaskList(task_list)
# Start procs
mpe.startProducers(num_producers)
mpe.startConsumers(num_consumers, [dbfilenameFullPath])
# mpe.addProducer()
# Control loop
while mpe.working():
time.sleep(1.0)
(num_producers,num_consumers,num_tasks,progress_producers,progress_consumers) = mpe.getProgress()
elapsed_time = datetime.now() - t0
mean_loadtime_per_host = (elapsed_time) / max(1, _tasksPerJob * progress_consumers)
pending_hosts = ((num_tasks * _tasksPerJob) - (_tasksPerJob * progress_consumers))
etr = (mean_loadtime_per_host * pending_hosts)
eta = t0 + elapsed_time + etr
ett = (eta - t0)
if settings.logger_getDebugMode(): status_extra_data = " Prod: %s Cons: %s (%d -> %d -> %d: %d) [RAM: %d%% / Obj: %d / ETH: %s / ETA: %s / ETT: %s]" % \
(num_producers, num_consumers, num_tasks, progress_producers, progress_consumers, progress_producers - progress_consumers,
psutil_phymem_usage(), len(gc.get_objects()),
mean_loadtime_per_host if progress_consumers * _tasksPerJob > 100 else "N/A",
str(eta.time()).split(".")[0] if progress_consumers * _tasksPerJob > 100 else "N/A",
str(ett).split(".")[0] if progress_consumers * _tasksPerJob > 100 else "N/A")
else: status_extra_data = ""
# logger.info("Parsing files%s" % status_extra_data)
logger.info(update_progress(min(1,float(progress_consumers) / float(num_tasks)), "Parsing files%s" % status_extra_data, True))
mpe.rebalance()
del mpe
# Stop timer
elapsed_time = datetime.now() - t0
mean_loadtime_per_host = (elapsed_time) / max(1, countInstancesToProcess)
logger.info("Load speed: %s seconds / file" % (mean_loadtime_per_host))
logger.info("Load time: %s" % (str(elapsed_time).split(".")[0]))
else:
logger.info("Found no files to process!")
