def excTasks(tasks):
job_defaults = {'max_instances': 2}
scheduler = BlockingScheduler(job_defaults=job_defaults)
if tasks is not None:
if isinstance(tasks[0], list):
for task in tasks:
addTask(scheduler, task)
else:
addTask(scheduler, tasks)
scheduler.add_listener(my_listener, EVENT_JOB_EXECUTED | EVENT_JOB_ERROR)
try:
scheduler.start()
except (KeyboardInterrupt, SystemExit):
scheduler.pause()
def my_listener(event):
if event.exception:
print("保存出错信息")
# 如果出现失误，首先将出错的消息发给服务器
# 日志记录下来
#1.直接关闭任务调度器
scheduler.resume()
print("任务出错了!!!")
else:
print('任务照常进行...')
if __name__ == '__main__':
scheduler = BlockingScheduler()
scheduler.add_job(job2,
trigger='cron',
second='*/5',
args=['hello', 'world'])
scheduler.add_job(job1, trigger='cron', second='*/5')
#每天整点运行
# scheduler.add_job(
# get_craw_data,
# trigger='cron',
# #hour='0-23',
# minute='0-59',
# )
scheduler.add_listener(my_listener, EVENT_JOB_EXECUTED | EVENT_JOB_ERROR)
scheduler.start()
class Scheduler:
def __init__(self):
conf = configparser.ConfigParser()
conf.read("../agent.ini")
ip = conf.get("redis", "ip")
port = conf.getint("redis", "port")
timeout = conf.getint("redis", "timeout")
self.invoker_id = self._get_invoker_id()
self.max_tasks = conf.getint("invoker", "max_tasks")
self.live_seconds = conf.getint("invoker", "live_seconds")
self.db = SchedulerDb(ip, port, timeout)
logging.config.fileConfig("../logger.ini")
self.logger = logging.getLogger("main")
executors = {
'default': {'type': 'processpool', 'max_workers': self.max_tasks + 1}
}
self.blockScheduler = BlockingScheduler()
self.jobs = {}
self.lock = threading.Lock()
@staticmethod
def _get_invoker_id():
hostname = socket.gethostname()
pid = os.getpid()
return hostname + "-" + str(pid)
def task_invoke(self, task_instance, task_param):
if task_param.cmd.startswith('http'):
executor = HttpExecutor(self.db, task_instance, task_param)
executor.execute()
else:
pass
def break_heart(self):
"""
invoker每隔一段时间就心跳一下，看看是否有新任务，是否有任务需要更新
:param bs:
:return:
"""
# 先看看参数是否有变化的把调度重启或者关闭
try:
self.lock.acquire()
self.refresh_local_invoker()
self.refresh_other_invokers()
if len(self.jobs) >= self.max_tasks:
return
task_instances, task_params = self.db.query_waiting_run_tasks(self.invoker_id,
self.max_tasks - len(self.jobs),
True)
if len(task_instances) == 0:
return
for i in range(len(task_instances)):
task_instance = task_instances[i]
task_param = task_params[i]
if task_instance.id not in self.jobs.keys():
self.logger.info("分配了新任务%s", task_instance.id)
job = self.blockScheduler.add_job(self.task_invoke,
next_run_time=(
datetime.datetime.now() + datetime.timedelta(seconds=2)),
args=[task_instance, task_param], id=task_instance.id)
self.jobs[job.id] = job
self.db.lock_invoker_instance(self.invoker_id, task_instance.id, self.live_seconds)
else:
self.logger.error("%s任务已经在运行", task_instance.id)
finally:
self.lock.release()
def refresh_local_invoker(self):
"""
调度的参数是否发生变化，如有需要重启调度
:param bs:
:return:
"""
self.db.update_invoker_time(self.invoker_id, self.jobs.keys(), self.live_seconds)
self.logger.info("%s心跳更新成功！", self.invoker_id)
# 看看是否有需要停止的任务再自己这里，释放掉
stop_tasks = self.db.query_need_stop_tasks(self.invoker_id)
for stop_task in stop_tasks:
if stop_task in self.jobs.keys():
try:
job = self.jobs[stop_task]
task_instance = job.args[0]
task_instance.status = 'off'
job.pause()
job.remove()
except Exception as e:
self.logger.error(e)
self.jobs.pop(stop_task)
try:
self.blockScheduler.remove_job(stop_task)
except Exception as e1:
self.logger.error(e1)
self.logger.info("人工停止了任务%s", stop_task)
self.db.unlock_invoker_instance(self.invoker_id, stop_task, self.live_seconds)
# 是否有参数变化的任务需要重启
c_jobs = copy.copy(self.jobs)
for key in c_jobs.keys():
if key not in self.jobs.keys():
continue
job = self.jobs[key]
task_instance = job.args[0]
old_task_param = job.args[1]
# 判断参数是否发生变化，如果有变化重新执行任务
new_task_param = self.db.query_task_param(task_instance.task_param_id)
# if new_task_param
if not new_task_param.has_diff(old_task_param):
continue
try:
task_instance.status = 'off'
job.pause()
job.remove()
except Exception as e:
self.logger.error(e)
self.jobs.pop(key)
try:
self.blockScheduler.remove_job(key)
except Exception as e1:
self.logger.error(e1)
self.logger.info("参数变化停止了任务%s", task_instance.id)
self.db.unlock_invoker_instance(self.invoker_id, task_instance.id, self.live_seconds)
self.db.add_task_waiting_run(task_instance.id)
def refresh_other_invokers(self):
"""
遍历所有的invoker，判断invoker是否超过存活期
:return:
"""
invokers = self.db.query_all_invokers()
for invoker_id in invokers.keys():
if not self.db.invoker_is_live(self.invoker_id):
task_instance_list = self.db.query_invoker_tasks(self.invoker_id)
for task_instance_id in task_instance_list:
self.db.add_task_waiting_run(task_instance_id)
def main(self):
try:
self.db.register_invoker(self.invoker_id, self.max_tasks, self.live_seconds);
self.blockScheduler.add_listener(self._job_listener,
events.EVENT_JOB_ERROR | events.EVENT_JOB_MISSED)
self.blockScheduler.add_job(self.break_heart, "interval", seconds=self.live_seconds / 2,
id="break_heart")
self.logger.info("开始启动调度...")
self.blockScheduler.start()
self.logger.info("启动调度成功！")
except KeyboardInterrupt as e:
self.logger.info(e)
self.blockScheduler.shutdown()
def _job_listener(self, ev):
"""
监听job的事件，job完成后再发起下次调用，对于异常也要处理
:param ev:
:return:
"""
if ev.code == events.EVENT_JOB_ERROR:
self.logger.error(ev.exception)
self.logger.error(ev.traceback)
else:
pass
class EventScheduler:
def __init__(self, reporter: ResultReporter):
self.reporter = reporter
self.scheduler = BlockingScheduler()
self.events = list()
log_path = static_setting.settings["CaseRunner"].log_path
log_file = os.path.join(log_path, "event_scheduler_log.log")
self.log = logger.register("EventScheduler",
filename=log_file,
for_test=True)
self.scheduler.add_listener(self._event_listen, EVENT_JOB_EXECUTED)
def add_event(self,
event,
package,
args,
is_background,
need_lock,
start_time,
interval=5,
loop_count=1,
description=""):
m = importlib.import_module(package)
event_cls = getattr(m, event)
new_event = event_cls(description, log=self.log)
new_event.need_lock = need_lock
new_event.back_ground = is_background
new_event.arguments = args
new_event.interval = interval
new_event.loop_count = loop_count
# 生成一个STEP 的节点给Event操作
new_event.reporter = self.reporter.add_event_group(f"Event: {event}")
if is_background:
new_event.job = self.scheduler.add_job(new_event.run,
"interval",
seconds=interval,
start_date=start_time,
id=f"{event}{uuid.uuid4()}")
else:
new_event.job = self.scheduler.add_job(new_event.run,
"date",
run_date=start_time,
id=f"{event}{uuid.uuid4()}")
self.events.append(new_event)
def remove_event(self, event_id):
job = self.scheduler.get_job(event_id)
if job:
event_to_remove = None
for event in self.events:
if event.job == job:
event_to_remove = event
self.scheduler.remove_job(event_id)
break
if event_to_remove:
self.events.remove(event_to_remove)
def start(self):
self.scheduler.start()
def _event_listen(self, job):
for event in self.events:
if event.job.id == job.job_id:
if event.back_ground:
return
else:
if event.loop_count == 1:
return
delta = datetime.timedelta(seconds=event.interval)
next_date = job.scheduled_run_time + delta
event.job = self.scheduler.add_job(
event.run,
"date",
run_date=next_date,
id=f"{event.name}{uuid.uuid4()}")
event.loop_count -= 1
return
class ActionScheduler:
def __init__(self):
self.scheduler = BlockingScheduler(jobstores=APSCHEDULER_SETTINGS['jobstores'],
executors=APSCHEDULER_SETTINGS['executors'],
job_defaults=APSCHEDULER_SETTINGS['job_defaults'],
timezone=TIMEZONE_PST8PDT)
pass
def start(self):
self._add_event_listener()
# self._add_example_jobs()
self._add_jobs()
self.scheduler.start()
def shutdown(self):
# self.scheduler.remove_all_jobs() # save all jobs into sqlite, do not remove them
self.scheduler.shutdown()
def _add_event_listener(self):
self.scheduler.add_listener(ActionScheduler.listener_jobs_status, EVENT_JOB_EXECUTED | EVENT_JOB_ERROR)
self.scheduler.add_listener(ActionScheduler.listener_all_jobs_finished, EVENT_ALL_JOBS_REMOVED)
# examples
def _add_example_jobs(self):
import datetime
self.scheduler.add_job(func=ActionScheduler.job_example, args=["cron", ], trigger='cron', second='*/5',
misfire_grace_time=DEFAULT_MISFIRE_GRACE_TIME, replace_existing=True, id="cron")
self.scheduler.add_job(func=ActionScheduler.job_example, args=["interval", ], trigger='interval', seconds=60,
misfire_grace_time=DEFAULT_MISFIRE_GRACE_TIME, replace_existing=True, id="interval")
self.scheduler.add_job(func=ActionScheduler.job_example, args=["date", ], trigger='date',
run_date=get_cur_time()+datetime.timedelta(seconds=12), id="date")
# examples
@staticmethod
def job_example(job_type):
print("job_example: {}".format(job_type))
def _add_jobs(self):
# add reap alerts immediate job TODO test
# self.scheduler.add_job(id="reap_alerts_immediate", func=ActionScheduler.job_reap_alerts_and_start_action_tasks, args=[],
# misfire_grace_time=DEFAULT_MISFIRE_GRACE_TIME, replace_existing=True, )
# add reap alerts interval job
# self.scheduler.add_job(id="reap_alerts", func=ActionScheduler.job_reap_alerts_and_start_action_tasks,
# args=[], trigger='interval', seconds=REAP_INTERVAL_SECONDS,
# misfire_grace_time=DEFAULT_MISFIRE_GRACE_TIME, replace_existing=True, )
# add gather & retry failed action tasks immediate job TODO test
# self.scheduler.add_job(id="check_tasks_immediate", func=ActionScheduler.job_gather_and_retry_failed_action_tasks, args=[],
# misfire_grace_time=DEFAULT_MISFIRE_GRACE_TIME, replace_existing=True, )
# add gather & retry failed action tasks interval job
# self.scheduler.add_job(id="check_tasks", func=ActionScheduler.job_gather_and_retry_failed_action_tasks,
# args=[], trigger='interval', seconds=GATHER_FAILED_TASKS_INTERVAL_SECONDS,
# misfire_grace_time=DEFAULT_MISFIRE_GRACE_TIME, replace_existing=True, )
pass
@staticmethod
def listener_all_jobs_finished(event): # this would hardly be invoked
logger_.info('All jobs are done.')
@staticmethod
def listener_jobs_status(event):
if event.exception:
logger_.warn('Job {} crashed.'.format(event.job_id))
else:
logger_.info('Job {} executed.'.format(event.job_id))
'''
if isinstance(event, JobSubmissionEvent):
print("任务{}触发执行".format(event.job_id))
if isinstance(event, JobExecutionEvent) and event.exception:
print("任务{}抛出异常:{}".format(event.job_id, event.exception))
jobstores = {
'default': SQLAlchemyJobStore(url='sqlite:///db.sqlite3') #定时器任务持久化地址
}
scheduler = BlockingScheduler(jobstores=jobstores)
scheduler.add_job(func=aps_test,
trigger=CronTrigger(hour=20, minute=48),
replace_existing=True,
id="sb")
scheduler.add_job(func=aps_test,
replace_existing=True,
trigger=DateTrigger(run_date=datetime.datetime.now() + datetime.timedelta(seconds=12)),
id="sbsb")
scheduler.add_job(func=aps_test, # 任务函数
trigger=IntervalTrigger(seconds=5), # 执行计划
replace_existing=True,
id="sbsbsb" # 唯一的
)
scheduler.add_listener(my_listener)
scheduler.start()
class TaskExecutor:
def __init__(self, db, task_instance, task_param):
self.task_instance = task_instance
self.task_param = task_param
self.db = db
# invoke log
self.invoke_log_map = {}
self.jobs = {}
logging.config.fileConfig("../logger.ini")
self.logger = logging.getLogger("taskExecutor")
invoke_count = int(self.task_param.get_invoke_args()['invoke_count'])
executors = {
'default': {
'type': 'threadpool',
'max_workers': invoke_count + 1
}
}
self.scheduler = BlockingScheduler(executors=executors)
def execute(self):
self.scheduler.add_listener(
self._job_listener,
events.EVENT_JOB_EXECUTED | events.EVENT_JOB_ERROR
| events.EVENT_JOB_ADDED | events.EVENT_JOB_MISSED)
# invoke_log_map up server
self.scheduler.add_job(self._invoke_break_heart, "interval", seconds=2)
try:
self.scheduler.start()
except Exception as e:
print(e)
self.scheduler.shutdown(wait=True)
def _job_listener(self, ev):
"""
监听job的事件，job完成后再发起下次调用，对于异常也要处理
:param ev:
:return:
"""
if self.task_instance.status == 'off':
return
if ev.code == events.EVENT_JOB_ADDED:
self.jobs[ev.job_id] = self.scheduler.get_job(ev.job_id)
elif ev.code == events.EVENT_JOB_EXECUTED or ev.code == events.EVENT_JOB_ERROR:
if ev.code == events.EVENT_JOB_ERROR:
self.logger.error(ev.exception)
self.logger.error(ev.traceback)
job = self.jobs[ev.job_id]
self.scheduler.add_job(
job.func,
next_run_time=(datetime.datetime.now() +
datetime.timedelta(seconds=1)),
id=ev.job_id,
args=job.args)
else:
pass
def _invoke_break_heart(self):
if self.task_instance.status == 'off':
jobs = self.scheduler.get_jobs()
for job in jobs:
try:
job.pause()
job.remove()
except Exception as e:
self.logger.error(e)
self.db.save_task_logs(self.invoke_log_map)
parser.add_argument(
'--log',
choices=['DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL'],
default='WARNING',
help="Log level")
args = parser.parse_args()
logging.basicConfig(level=getattr(logging, args.log))
config = configparser.ConfigParser()
config.read_file(args.config_file)
scheduler = BlockingScheduler()
scheduler.add_listener(
lambda event: logging.error("Exception in feed fetch:",
exc_info=event.exception), EVENT_JOB_ERROR)
interval = int(config['interval']['interval'], 10)
agency_ids = [
key.split(':')[1] for key in config.keys() if key.startswith('agency:')
]
for agency_id in agency_ids:
for feed in config['agency:' + agency_id].items():
(feed_id, feed_url) = feed
scheduler.add_job(get,
'interval', (agency_id, feed_id, feed_url,
config['influxdb'], interval / 2),
seconds=interval,
id=agency_id + ":" + feed_id)
class TimerWork():
def __init__(self):
self.jobs = []
self.sched = BlockingScheduler()
self.sched.add_listener(self.listener, EVENT_JOB_ERROR | EVENT_JOB_EXECUTED)
def listener(self, event):
if event.exception:
print 'JOB :<'
else:
print 'JOB :>'
pass
def exit(self):
print 'shutdown now'
self.sched.shutdown(wait=True)
print 'shutdown ok'
pass
def pause(self):
BaseScheduler.pause_job()
pass
def resume(self):
BaseScheduler.remove_job()
pass
def add_job(self, func, gap_time=0, first_run_time=None, args=None, kwargs=None):
'''
添加定时任务
:param func: 执行函数
:param gap_time: 间隔时间,如果设置小于等于0的值忽略 单位s
:param first_run_time: 第一次执行时间 datetime.datetime 如果不设置就按正常处理
:param args: 参数
:param kwargs: 参数
:return:
'''
self.jobs.append(func)
if type(first_run_time) == types.IntType:
first_run_time = datetime.datetime.now() + datetime.timedelta(seconds=first_run_time)
if gap_time > 0:
if first_run_time:
self.sched.add_job(func, 'interval', seconds=gap_time, next_run_time=first_run_time, args=args, kwargs=kwargs)
else:
self.sched.add_job(func, 'interval', seconds=gap_time, args=args,
kwargs=kwargs)
else:
if first_run_time:
self.sched.add_job(func, next_run_time=first_run_time, args=args, kwargs=kwargs)
pass
else:
self.sched.add_job(func, args=args, kwargs=kwargs)
pass
pass
def run(self):
if len(self.jobs) > 0:
self.sched.start()
else:
print 'no add jobs , so check'
class JobLauncher(object):
def __init__(self, background=False, deamon=True, **kwargs):
logging.basicConfig(format="[%(asctime)s] %(message)s",
atefmt="%Y-%m-%d %H:%M:%S")
logging.getLogger('apscheduler').setLevel(logging.DEBUG)
if background:
self.sched = BackgroundScheduler(deamon=deamon) # background
else:
self.sched = BlockingScheduler(deamon=deamon) # foreground
# TODO: Read from configuration file.
self.sched.configure(
jobstores={
# "sqlite": SQLAlchemyJobStore(url='sqlite:///app/database/example.db'),
# "default": MemoryJobStore()
"default":
SQLAlchemyJobStore(url='sqlite:///app/database/example.db')
},
executors={
'default': ThreadPoolExecutor(20),
'processpool': ProcessPoolExecutor(5)
},
job_defaults={
'coalesce': False,
'max_instances': 3
},
timezone=get_localzone() # Asia/Seoul
)
self.retried = 0
self.logger = logging.getLogger('apscheduler')
super(JobLauncher, self).__init__()
def start(self):
try:
if self.sched.state != STATE_RUNNING:
self.printJobs(jobstore='default')
started = self.sched.start()
except ConflictingIdError as e:
traceback.print_exc()
except KeyboardInterrupt as e:
traceback.print_exc()
finally:
pass
# Remove all remained store.
# self.sched.remove_all_jobs()
# for job in self.getJobs():
# if job.pending:
# job.pause()
self.logger.info('Finished')
self.logger.info(self.getJobs())
self.printJobs()
def stop(self, wait=False):
if self.sched.state == STATE_RUNNING:
self.sched.shutdown(wait=wait)
def resume(self):
if self.sched.state == STATE_RUNNING:
self.sched.resume()
def pause(self):
if self.sched.state == STATE_RUNNING:
self.sched.pause()
def addListener(self, listener, types):
self.sched.add_listener(listener, types)
def addJob(self, job, **kwargs):
execute, trigger, options = job.build(**kwargs)
added_job = self.sched.add_job(execute, trigger, **options)
self.printJobs()
return added_job
def getJob(self, job_id):
return self.sched.get_job(job_id)
def getJobs(self, jobstore=None):
return self.sched.get_jobs(jobstore=jobstore)
def removeJob(self, job_id, jobstore=None):
return self.sched.remove_job(job_id, jobstore=jobstore)
def removeAllJob(self, jobstore=None):
return self.sched.remove_all_jobs(jobstore=jobstore)
def printJobs(self, jobstore=None, out=None):
return self.sched.print_jobs(jobstore=jobstore, out=None)
def getJobState(self, job_id=None, jobstore=None):
state = list()
if job_id is not None:
job = self.sched.get_job(job_id, jobstore=jobstore)
if job is not None:
temp = dict()
temp[job.id] = {
"next_run_time": job.next_run_time,
"state": job.pending,
}
state.append(temp)
else:
for job in self.sched.get_jobs(jobstore=jobstore):
temp = dict()
temp[job.id] = {
"next_run_time": job.next_run_time,
"state": job.pending,
}
state.append(temp)
return state
