def main():
taskflowdb = TaskFlowDB()
redisdb = RedisDB()
while True:
# 找到当前调度中有效调度,活动状态,有效数据
data = taskflowdb.get_sched_cron()
if len(data) == 0:
time.sleep(30)
continue
for item in data:
try:
# 精确到秒
trigger_time = datetime.datetime.strptime(
datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S'),
'%Y-%m-%d %H:%M:%S')
cron = croniter(item["cron_sched"], trigger_time)
args_python_code = item["args_python_code"]
args_json = get_arguments(args_python_code, trigger_time, True)
# 默认没有父任务
parent_id = 0
source_id = item["id"]
source_type = "schedule"
if "workflow" == item["task_type"]:
# 则先创建父任务
instance_id = taskflowdb.create_instance(
item["task_name"], source_id, source_type, parent_id,
"workflow", item["task_name"], args_json, 'running')
wf = WorkflowSpec(item["task_name"], taskflowdb,
instance_id, parent_id)
step_name = wf.get_step_name(wf.begin_step)
module_name = wf.steps[step_name].get("module")
args_json = wf.get_step_parameters(step_name, True)
parent_id = instance_id
elif "module" == item["task_type"]:
module_name = item["task_name"]
step_name = module_name
else:
raise ValueError("task_type is invalid")
instance_id = taskflowdb.create_instance(
step_name, source_id, source_type, parent_id, "module",
module_name, args_json, 'running')
trigger_next_time = cron.get_next(datetime.datetime)
redisdb.push_run_queue(instance_id)
taskflowdb.update_sched("start",
item["id"],
'running',
trigger_last_time=trigger_time,
trigger_next_time=trigger_next_time)
logging.info("shcedid=%s is running" % (item["id"]))
except:
logging.error(traceback.format_exc())
time.sleep(3)
def main():
logging.info("taskflow checker is running")
redisdb = RedisDB()
while True:
data = redisdb.fetch_check_queue()
if data:
for instance_id, score in data:
redisdb.push_run_queue(instance_id)
redisdb.del_check_queue(instance_id)
time.sleep(3)
else:
time.sleep(300)
def main():
logging.info("taskflow producer is running")
taskflowdb = TaskFlowDB()
redisdb = RedisDB()
while True:
data = taskflowdb.get_undo_taskforms()
if len(data) == 0:
time.sleep(30)
continue
for item in data:
try:
source_id = item["id"]
source_type = "form"
# 默认没有父任务
parent_id = 0
if "workflow" == item["task_type"]:
# 则先创建父任务
instance_id = taskflowdb.create_instance(
item["task_name"], source_id, source_type, parent_id,
"workflow", item["task_name"], item["args_json"],
'running')
wf = WorkflowSpec(item["task_name"], taskflowdb,
instance_id, parent_id)
step_name = wf.get_step_name(wf.begin_step)
module_name = wf.steps[step_name].get("module")
args_json = wf.get_step_parameters(step_name, True)
parent_id = instance_id
elif "module" == item["task_type"]:
module_name = item["task_name"]
step_name = module_name
args_json = item["args_json"]
else:
raise ValueError("task_type is invalid")
# 创建任务
instance_id = taskflowdb.create_instance(
step_name, source_id, source_type, parent_id, "module",
module_name, args_json, 'running')
redisdb.push_run_queue(instance_id)
taskflowdb.save_taskform_status(item["id"], 'running')
except:
logging.error(traceback.format_exc())
time.sleep(3)
class TestRedisDB(unittest.TestCase):
def setUp(self) -> None:
self.db = RedisDB()
def test_run_queue(self):
instance = 1
self.db.push_run_queue(instance)
pop_value = self.db.pop_run_queue()
self.assertEqual(int(pop_value), instance)
def test_check_queue(self):
instance = 1
times = 1
interval = 5
self.db.set_check(instance, times, interval)
data = self.db.get_check_times(instance)
self.assertEqual(int(data), times)
time.sleep(10)
data = self.db.fetch_check_queue()
self.assertTrue(len(data) > 0)
print('test check queue data:', type(data[0]), data[0])
self.db.del_check_queue(instance)
self.db.del_check(instance)
def main(instance_id: int):
"""
当前运行的一定是module
"""
try:
taskflowdb = TaskFlowDB()
# 获取基础数据信息
instance_data = taskflowdb.get_instance(instance_id)
if "module" != instance_data["task_type"]:
logging.error("当前运行的不是模块!")
raise ValueError("id %s is not module" % instance_id)
module_name = instance_data["task_name"]
# 动态导入运行模块
inner_func = importlib.import_module("modules.%s" % module_name)
inner_func_main = getattr(inner_func, "main")
# 实例获取到的参数
inner_func_main_full_arg_spec = inspect.getfullargspec(inner_func_main)
inner_func_main_argument_list = inner_func_main_full_arg_spec.args
# 处理参数数据
# 运行中的产生的参数
inner_func_kwargs = {}
# 处理输入参数别名的情况并设定模块运行数据
input_arguments = json.loads(instance_data["args_json"])
for arg_name in inner_func_main_argument_list:
if arg_name in input_arguments:
arg_value = input_arguments.get(arg_name)
inner_func_kwargs[arg_name] = arg_value
if inner_func_main_full_arg_spec.varkw:
inner_func_kwargs["sys_instance"] = instance_data
# 暂时关闭释放资源,因为连接串资源宝贵
taskflowdb.close()
# 运行模块
success = True
message = ""
return_data = {}
run_result = None
try:
logging.info("----------run module: %s start----------" % module_name)
run_result = inner_func_main(**inner_func_kwargs)
logging.info("----------run module: %s finish----------" % module_name)
if run_result is not None:
if type(run_result) is bool:
success = run_result
elif type(run_result) is tuple:
len_ret = len(run_result)
if len_ret > 0:
success = bool(run_result[0])
if len_ret > 1:
message = str(run_result[1])
if len_ret > 2:
return_data = dict(run_result[2])
except:
success = False
message = traceback.format_exc()
logging.error("run module err \n %s", message)
redisdb = RedisDB()
if str(module_name).startswith("check_"):
if run_result is None:
check_interval = inner_func_kwargs.get("check_interval", 300)
check_maxcount = inner_func_kwargs.get("check_maxcount", 0)
times = redisdb.get_check_times(instance_id)
# 这里需要出来下check的功能
if check_maxcount and times > check_maxcount:
redisdb.del_check(instance_id)
else:
redisdb.set_check(instance_id, times + 1, check_interval)
return
else:
redisdb.del_check(instance_id)
result_status = 'success' if success else 'failure'
# 重新开启db资源
taskflowdb = TaskFlowDB()
# 更新instance 数据
result_json = json.dumps(return_data, cls=CustomJSONEncoder)
taskflowdb.save_instance_status(instance_id, result_status, result_message=message, result_json=result_json)
# 处理执行结果
# 如果是工作流
source_id = instance_data["source_id"]
source_type = instance_data["source_type"]
parent_id = instance_data["parent_id"]
if parent_id > 0:
parent_instance = taskflowdb.get_instance(parent_id)
workflow_name = parent_instance["task_name"]
wf = WorkflowSpec(workflow_name, taskflowdb, instance_id, parent_id)
cur_step_name = instance_data["name"]
end_step_name = wf.get_step_name(wf.end_step)
if cur_step_name == end_step_name:
update_source_task_status(taskflowdb, source_type, source_id, result_status)
return
cur_step = wf.steps[cur_step_name]
if success:
# 判断是否需要进行成功后暂停
success_pause = cur_step.get("on-success-pause", False)
if success_pause:
update_source_task_status(taskflowdb, source_type, source_id, 'pause')
return
next_step_name = wf.get_step_name(cur_step.get("on-success"))
if not next_step_name:
update_source_task_status(taskflowdb, source_type, source_id, result_status)
return
else:
retry_count = int(cur_step.get("on-failure-retry", 0))
run_count = instance_data.get("retry_count", 0)
if retry_count > 0 and run_count <= retry_count:
redisdb.push_run_queue(instance_id)
taskflowdb.save_instance_status(parent_id, result_status, retry_count=run_count + 1,
result_message=message)
return
taskflowdb.save_instance_status(parent_id, result_status, result_message=message)
next_step_name = wf.get_step_name(cur_step.get("on-failure"))
if not next_step_name:
update_source_task_status(taskflowdb, source_type, source_id, result_status)
return
# 计算获取下一步骤的参数数据
next_module_name = wf.steps[next_step_name].get("module")
next_step_args_json = wf.get_step_parameters(next_step_name, True)
next_instance_id = taskflowdb.create_instance(next_step_name, source_id, source_type, parent_id,
"module", next_module_name, next_step_args_json, 'running')
redisdb.push_run_queue(next_instance_id)
else:
update_source_task_status(taskflowdb, source_type, source_id, result_status)
except:
logging.error("task run err \n %s", traceback.format_exc())