def check_scheduled_but_unprocessed_items(self):
query = {"status":{"$in":["proc_schedule","quit","stop"] }}
count = self.__logs_collection__.find(query).count()
self.inform("Found %d scheduled but Unprocessed items" % count)
if count <= 0 :
return
logs_cursor = self.__logs_collection__.find(query)
for log in logs_cursor:
file = log.get("file")
self.__logs_collection__.update({"_id":log.get("_id")},{"$set":{"status":"proc_schedule","proc_timeout":five_mins_later()}},safe=True)
self.schedule_job(log.get("_id"),file)
def check_new_files_to_be_processed(self):
self.inform( "{0}Checking New Files to Process{0}".format("="*10))
logs_cursor = self.__logs_collection__.find({"status":"ready"})
count = 0
for file in logs_cursor:
count += 1
self.__logs_collection__.update({"_id":file.get("_id")},{"$set":{"status":"proc_schedule","proc_timeout":five_mins_later()}},safe=True)
self.schedule_job(file.get("_id"),file.get("file"))
self.inform( "Found %s files to process " % ( str(count) if count > 0 else "NO" ) )
def check_scheduled_but_left_behind_jobs(self):
self.inform( "{0}Checking Left Behind Tasks to Process{0}".format("="*10))
# qsize = self.__worker_pool__._taskqueue.qsize()
# self.inform("Process Queue Length is %s" % (qsize))
query = { "status":{"$in": ["proc_schedule","working"] } }
count = self.__logs_collection__.find(query).count()
if count <= 0:
self.inform( "Nothing seems to be active or left behind" )
else:
self.inform("Possibly %d working. Let's check their status" % count)
logs_cursor = self.__logs_collection__.find(query)
def_timeout = dt.utcnow() - timedelta(hours=1)
for log in logs_cursor:
shouldReschedule = False
file = log.get("file")
status = log.get("status")
hbeat = log.get("hbeat")
timeout = log.get("proc_timeout",def_timeout)
if status in ["working"] :
if hbeat is None:
shouldReschedule = True
elif hbeat < dt.utcnow() - timedelta(minutes=1):
shouldReschedule = True
elif status in ["proc_schedule"] and timeout <= dt.utcnow():
shouldReschedule = True
if shouldReschedule:
self.inform("Rescheduling %s - was %s" % (file,status))
self.__logs_collection__.update({"_id":log.get("_id")},{"$set":{"status":"proc_schedule","proc_timeout":five_mins_later()}},safe=True)
self.schedule_job(log.get("_id"),file)
