【记录】celery + mongodb 动态添加、更新定时任务时间

katja · 老王

已于 2023-06-12 16:54:01 修改

阅读量1.3k

点赞数

分类专栏：笔记文章标签： mongodb celery

于 2022-11-05 16:26:08 首次发布

本文链接：https://blog.csdn.net/qq_35567179/article/details/127704085

版权

笔记专栏收录该内容

5 篇文章 1 订阅

订阅专栏

更新时间

2023-06-12

一、目录结构

├── run_celery.py
├── celery_tasks
    │   ├── sync_task.py
    │   ├── task1.py
    │   └── task2.py
    └──
├── service
    │   ├── add_task.py
    └──
├── database
    │   ├── mongodb_class.py
    └──
├── config
    │   ├── config.py
    └──
└──

功能说明

run_celery.py 用于定义celery应用
celery_tasks 用于存放需要定时执行的任务方法
sync_task.py 用于触发动态添加任务
task1.py， task2.py 被动态添加的任务
service add_task.py 添加任务方法
database mongodb_class.py 被封装的mongodb操作类
config config.py 用于定义celery、mongodb 配置

二、逻辑

在这里插入图片描述

三、依赖

1. mongodb 安装

https://blog.csdn.net/qq_21103471/article/details/124706915

2. pip 依赖包安装

pip install celery==5.2.7

pip install celerybeat-mongo==0.2.0

pip install eventlet==0.33.1

pip install mongoengine==0.24.2

pip install pymongo==3.12.0

四、代码

1. 配置项

config.py

# celery 任务数据库
CELERY_DB_NAME_BROKE = 'celery_broke'  # mongodb 任务存储 - 可自定义
CELERY_DB_NAME_BACKEND = 'celery_backend'  # mongodb celery 执行结果存储
CELERY_COLLECT_NAME = 'schedules'  # 任务所在表名
CELERY_DRIVER = 'mongodb'  # celery 数据存储方式
CELERY_DB_HOST = 'localhost'  
CELERY_DB_PORT = 27017
CELERY_TASK_DB_URI = f'{CELERY_DRIVER}://{CELERY_DB_HOST}:{CELERY_DB_PORT}'  # 用于本地操作mongodb库
CELERY_BROKER = f'{CELERY_TASK_DB_URI}/{CELERY_DB_NAME}'  # 用于给celery存储消息中间件，任务id
CELERY_BACKEND = f'{CELERY_TASK_DB_URI}/{CELERY_DB_NAME}' # 用于给celery存储执行结果
CELERY_DB_ALIAS = 'celery_alias'  # 别名
CELERY_TIMEZONE = 'Asia/Shanghai'  # 时区
CELERY_UTC = True  # 是否UTC时间

2. celery 任务定义


from celery import Celery
from config.config import (
    CELERY_TASK_DB_URI, CELERY_BACKEND, CELERY_BROKER, CELERY_TIMEZONE,
    CELERY_UTC, CELERY_DB_NAME_BROKE
)
from service.add_task import TaskScheduler

task_path = 'celery_tasks'
check_task_path = f'{task_path}.sync_task.sync_time'
sync_task_path1 = f'{task_path}.task1.task_a'
sync_task_path2 = f'{task_path}.task2.task_b'

celery_app = Celery(
    'task_name_whatever',
    backend=CELERY_BACKEND,
    broker=CELERY_BROKER,
)

celery_app.conf.update(
    {
        'beat_dburi': CELERY_TASK_DB_URI,
        'broker_url': CELERY_BROKER,
        'result_backend': CELERY_BACKEND,
        'timezone': CELERY_TIMEZONE,
        'enable_utc': CELERY_UTC,
        'mongodb_scheduler_db': CELERY_DB_NAME_BROKE, # mongodb schedule 所在数据库,与celery beat相关
        'include': [
            f'{task_path}.sync_task',
            f'{task_path}.task1',
            f'{task_path}.task2',
        ],
    }
)
# 每3分钟检查一下是否需要动态添加/更新任务时间
TaskScheduler.update(
    'sync_task',
    check_task_path,
    'sync_task',
    'minutes',
    3,
)

3. mongodb 操作封装

PS：celerybeatmongo PeriodicTask 用的是 mongoengine
而我自己项目操作数据库用的是pymongo
这一步记录是为了连贯，可以不用看，只是按照我的业务逻辑重写了一下方法
其实就是 mongodb 的增删改查

import pymongo
from configs.config import CELERY_DB_NAME, CELERY_TASK_DB_URI, CELERY_COLLECT_NAME


class MongoDBClass:

    def __init__(self, opt, collect_name=CELERY_COLLECT_NAME, value=None, condition=None, order='_id'):
        # 连接mongodb
        self.client = pymongo.MongoClient(CELERY_TASK_DB_URI)
        self.db = self.client[CELERY_DB_NAME]
        self.col = self.db[collect_name]
        self.opt = opt
        self.value = value
        self.condition = condition
        self.order = order

    def query(self):
        return self.col.find_one(self.condition)

    def insert(self):
        ...

    def update(self, is_insert=False):
        ...

    def upsert(self):
        ...

    def delete(self):
        ...

    def close(self):
        self.client.close()

    def __enter__(self):
        try:
            for attr in dir(MongoDBClass):
                if self.opt == attr:
                    return eval(f'self.{attr}')()
        except Exception as e:
            print(f'execute mongodb error, error option:{self.opt}, error info:', e)
            return None

    def __exit__(self, exc_type, exc_val, exc_tb):
        self.close()
        # 异常
        if exc_type is not None:
            print(exc_type, exc_val)
            return False
        return True


class MongoDBOperate:
    @staticmethod
    def query(collect_name=CELERY_COLLECT_NAME, condition=None):
        with MongoDBClass('query', collect_name, condition=condition) as res:
            return res

    @staticmethod
    def is_exist(collect_name=CELERY_COLLECT_NAME):
        with MongoDBClass('find', collect_name) as res:
            return True if res else False

    @staticmethod
    def insert(value, collect_name=CELERY_COLLECT_NAME):
        with MongoDBClass('insert', collect_name, value=value) as res:
            return res

    @staticmethod
    def update(condition, value, collect_name=CELERY_COLLECT_NAME):
        with MongoDBClass('update', collect_name, value=value, condition=condition) as res:
            return res

    @staticmethod
    def delete(condition, collect_name=CELERY_COLLECT_NAME):
        with MongoDBClass('delete', collect_name, condition=condition) as res:
            return res

    @staticmethod
    def upsert(condition, value, collect_name=CELERY_COLLECT_NAME):
        # 更新，不存在则添加
        with MongoDBClass('upsert', collect_name, value=value, condition=condition) as res:
            return res

4. 定义任务添加方法

from celerybeatmongo.models import PeriodicTask, get_periodic_task_collection
from database.mongo_class import MongoDBClass
from configs.config import (
    CELERY_DB_NAME_BROKE, CELERY_DB_HOST, CELERY_DB_PORT, CELERY_DB_ALIAS
)
import mongoengine


# 重写meta，定义PeriodicTask mongodb 默认连接
def celery_get_periodic_task_collection():
    try:
        return get_periodic_task_collection()
    except Exception as e:
        print('celery_get_periodic_task_collection err:', e)
        return "schedules"


class SubPeriodicTask(PeriodicTask):
    meta = {
        'collection': celery_get_periodic_task_collection(),
        'allow_inheritance': True,
        'db_alias': CELERY_DB_ALIAS,
    }


class TaskScheduler:

    @staticmethod
    def add(task_name: str, task_path: str, period: str, every: int, args=None, kwargs=None):
        permissible_periods = ['days', 'hours', 'minutes', 'seconds']
        if period not in permissible_periods:
            raise Exception('Invalid period specified')
        # 判断任务是否存在
        if TaskScheduler.is_exist(task_name):
            print(f'任务{task_name}已存在')
            return False
        # 创建任务
        mongoengine.connect(
        CELERY_DB_NAME_BROKE, host=CELERY_DB_HOST, port=CELERY_DB_PORT, alias=CELERY_DB_ALIAS)
        
        periodic = SubPeriodicTask(
            name=task_name,
            task=task_path,
            interval=SubPeriodicTask.Interval(every=every, period=period),
            run_immediately=True,
            enabled=True,
            args=args,
            kwargs=kwargs,
            connection_alias=CELERY_DB_ALIAS,
        )
        periodic.save()
        mongoengine.disconnect()
        return True

    @staticmethod
    def update(task_name: str, task_path: str, search_name: str, period: str, every: int, args=None, kwargs=None):
        # 判断相关任务是否已经存在
        with MongoDBClass('query', condition={'name': search_name}) as tasks:
        	if task and isinstance(task, dict):
                task_id = task.get('_id')
                if task_id:
                	TaskScheduler.delete({'_id': task_id})
        return TaskScheduler.add(task_name, task_path, period, every, args, kwargs)

    @staticmethod
    def query(condition):
        return MongoDBOperate.query(condition=condition)

    @staticmethod
    def delete(condition):
        return MongoDBOperate.delete(condition=condition)
        
	@staticmethod
    def is_exist(task_name):
        task = TaskScheduler.query({'name': task_name})
        return True if task and task.get('_id') else False

5. 检查方法调用任务添加

sync_task.py


from celery_run import celery_app, sync_task_path1, sync_task_path2, check_task_path
from service.add_task import TaskScheduler
from datetime import datetime
import re


@celery_app.task(name=check_task_path)
def check_time():
	# 假设此处 查询了数据库，字段test=1就执行task1,=2 则执行task2
	# ...
	test = 1
	task_name = ''
	task_path = ''
	sync_time = 0
	if test == 1:
		# 以下为业务逻辑，省略
		task_name = 'task1'
        task_path = sync_task_path1
        sync_time = 20
    elif test == 2:
		# 以下为业务逻辑，省略
    	task_name = 'task2'
        task_path = sync_task_path2
        sync_time = 15
    # 如果需要修改任务，则调用任务修改方法
    if task_name:
    	# 任务名统一为：
    	task_name = f'{task_name}-every-{sync_time}-minute'
    	# 任务在mongodb 里的查找方法我用的是正则匹配任务名
        search_name = f'$regex:/{task_name}-every-(.*?)-minute/'
        # 任务修改
        TaskScheduler.update(
            task_name, task_path, search_name, 'minutes', sync_time,
            ['传参1', '传参2'])

6. 最终任务方法

task1.py

from celery_run import celery_app, sync_task_path1

@celery_app.task(name=sync_task_path1)
def task_a(arg1, arg2):
    # 业务逻辑
    ...

task2.py

from celery_run import celery_app, sync_task_path2

@celery_app.task(name=sync_task_path2)
def task_b(arg1, arg2):
    # 业务逻辑
    ...

7. 执行

celery -A celery_run.celery_app worker --loglevel=debug -P eventlet -f ./logs/celery.log

celery -A celery_run.celery_app beat -S celerybeatmongo.schedulers.MongoScheduler -l info -f ./logs/celery2.log