FastAPI+apSheduler动态定时任务

---

阅读目录

一、apSheduler

二、实战应用

apSheduler

• 官方链接

1.安装

pip install apscheduler

2.基础组件

• triggers 触发器
• job stores job存储
• executors 执行器
• schedulers 调度器

3.选择合适的调度器，存储器，执行器，触发器

3.1调度器（schedulers）

• BlockingScheduler: 进程中只有调度器
• BackgroundScheduler: 非以下框架，且希望运行在后台
• AsyncIOScheduler: 应用程序使用asyncio
• GeventScheduler: 应用程序使用gevent
• TornadoScheduler: 构建Tornado
• TwistedScheduler: 构建Twisted应用
• QtScheduler: 构建Qt应用

3.2存储器（job stores）

• 持久化存储job，通过SQLAlchemyJobStore设置存储链接
• 非持久化存储job，重启时重新创建job，默认MemoryJobStore内存存储

3.3执行器（executors）

• processpoolexecutor，CUP密集型业务，可选进程池，也可以同线程池同时使用
• threadpoolexecutor，默人线程池

3.4触发器（triggers ）

• date: 设置日期，针对某个时间点运行一次job
• interval: 固定时间间隔运行job
• cron: 类似linux-crontab，某个时间点定期运行job

4.配置调度器

from pytz import utc

from apscheduler.schedulers.background import BackgroundScheduler
from apscheduler.jobstores.sqlalchemy import SQLAlchemyJobStore
from apscheduler.executors.pool import ProcessPoolExecutor


jobstores = {
    # 可以配置多个存储
    'mongo': {'type': 'mongodb'}, 
    'default': SQLAlchemyJobStore(url='sqlite:///jobs.sqlite')  # SQLAlchemyJobStore指定存储链接
}
executors = {
    'default': {'type': 'threadpool', 'max_workers': 20},     # 最大工作线程数20
    'processpool': ProcessPoolExecutor(max_workers=5)         # 最大工作进程数为5
}
job_defaults = {
    'coalesce': False,   # 关闭新job的合并，当job延误或者异常原因未执行时
    'max_instances': 3   # 并发运行新job默认最大实例多少
}
scheduler = BackgroundScheduler()

# .. do something else here, maybe add jobs etc.

scheduler.configure(jobstores=jobstores, executors=executors, job_defaults=job_defaults, timezone=utc) # utc作为调度程序的时区

5.调度器的增删改查

import os
import time
from apscheduler.schedulers.background import BackgroundScheduler

def print_time(name):
    print(f'{name} - {time.ctime()}')

def add_job(job_id, func, args, seconds):
    """添加job"""
    print(f"添加job - {job_id}")
    scheduler.add_job(id=job_id, func=func, args=args, trigger='interval', seconds=seconds)

def remove_job(job_id):
    """移除job"""
    scheduler.remove_job(job_id)
    print(f"移除job - {job_id}")

def pause_job(job_id):
    """停止job"""
    scheduler.pause_job(job_id)
    print(f"停止job - {job_id}")

def resume_job(job_id):
    """恢复job"""
    scheduler.resume_job(job_id)
    print(f"恢复job - {job_id}")

def get_jobs():
    """获取所有job信息,包括已停止的"""
    res = scheduler.get_jobs()
    print(f"所有job - {res}")

def print_jobs():
    print(f"详细job信息")
    scheduler.print_jobs()

def start():
    """启动调度器"""
    scheduler.start()

def shutdown():
    """关闭调度器"""
    scheduler.shutdown()

if __name__ == '__main__':
    scheduler = BackgroundScheduler()
    start()
    print('Press Ctrl+{0} to exit 
'.format('Break' if os.name == 'nt' else 'C'))
    add_job('job_A', func=print_time, args=("A", ), seconds=1)
    add_job('job_B', func=print_time, args=("B", ), seconds=2)
    time.sleep(6)
    pause_job('job_A')
    get_jobs()
    time.sleep(6)
    print_jobs()
    resume_job('job_A')
    time.sleep(6)
    remove_job('job_A')
    time.sleep(6)
    try:
        shutdown()
    except RuntimeError:
        pass

6.调度事件

可以将事件侦听器附加到调度程序。调度程序事件在某些情况下会被触发，并且可能会在其中携带有关该特定事件详细信息的附加信息。通过给add_listener()提供适当的掩码参数或者将不同的常量放在一起，可以只监听特定类型的事件。用一个参数调用侦听器callable，即event对象。

def my_listener(event):
    if event.exception:
        print('The job crashed :(')
    else:
        print('The job worked :)')

scheduler.add_listener(my_listener, EVENT_JOB_EXECUTED | EVENT_JOB_ERROR)

7.配置日志

import logging

logging.basicConfig()
logging.getLogger('apscheduler').setLevel(logging.DEBUG)

实战应用

1.fastapi动态添加定时任务

import asyncio
import datetime
import uvicorn
from fastapi import FastAPI, Body
from apscheduler.schedulers.asyncio import AsyncIOScheduler
from apscheduler.jobstores.sqlalchemy import SQLAlchemyJobStore
from apscheduler.executors.pool import ProcessPoolExecutor
from apscheduler.triggers.cron import CronTrigger

app = FastAPI(title='fast-api')

scheduler = None


@app.on_event('startup')
def init_scheduler():
    """初始化"""
    jobstores = {
        'default': SQLAlchemyJobStore(url='sqlite:///jobs.sqlite')  # SQLAlchemyJobStore指定存储链接
    }
    executors = {
        'default': {'type': 'threadpool', 'max_workers': 20},  # 最大工作线程数20
        'processpool': ProcessPoolExecutor(max_workers=5)  # 最大工作进程数为5
    }
    global scheduler
    scheduler = AsyncIOScheduler()
    scheduler.configure(jobstores=jobstores, executors=executors)
    # 添加一个coroutine执行，结果很不理想...
    scheduler.add_job(tick, 'interval', seconds=3)
    print("启动调度器...")

    scheduler.start()


def print_time(name):
    print(f'{name} - {datetime.datetime.now()}')


async def tick():
    print('Tick! The time is: %s' % datetime.datetime.now())
    await asyncio.sleep(1)


@app.post('/add-job')
async def add_job(job_id: str = Body(...), cron: str = Body(...)):
    """添加job"""
    scheduler.add_job(id=job_id, func=print_time, args=(job_id, ), trigger=CronTrigger.from_crontab(cron))
    return {"msg": "success!"}


@app.post('/remove-job')
async def remove_job(job_id: str = Body(..., embed=True)):
    """移除job"""
    scheduler.remove_job(job_id)
    return {"msg": "success!"}


if __name__ == '__main__':
    uvicorn.run(app, host='127.0.0.1', port=5566)

1.1问题

1. AsyncIOScheduler()添加协程job会有问题
2. apscheduler在多进程中会多次加载job，导致job重复执行，怎么解决呢？
   • 官方推荐可以用http、rpyc、grpc等方式解决

2.搭建rpyc服务添加定时任务

2.1定时业务代码在rpyc服务中

• 见官方demo

• 附rpyc官网查看最新rpyc接口调用

2.2通过rpyc服务回调添加定时任务

server.py

import datetime
import uvicorn
from fastapi import FastAPI, Body
import rpyc

app = FastAPI(title='fast-api')

conn = None
bgsrv = None
mon = None

@app.on_event('startup')
def init_scheduler():
    """初始化"""
    global conn,bgsrv,mon
    conn = rpyc.connect("localhost", 12345)
    # create a bg thread to process incoming events
    bgsrv = rpyc.BgServingThread(conn)  # 执行回调函数必须在开始线程
    mon = conn.root.Monitor(print_time)

def print_time(name):
    print(f'{name} - {datetime.datetime.now()}')

def from_crontab(cron):
    values = cron.split(' ')
    return {
        'minute': values[0],
        'hour': values[1],
        'day': values[2],
        'month': values[3],
        'day_of_week': values[4],
    }


@app.post('/add-job')
async def add_job(job_id: str = Body(...), cron: str = Body(...)):
    """添加job"""
    mon.add_job(id=job_id, args=(job_id, ), trigger='cron',  **from_crontab(cron))
    return {"msg": "success!"}

@app.post('/remove-job')
async def remove_job(job_id: str = Body(..., embed=True)):
    """移除job"""
    mon.remove_job(job_id)
    return {"msg": "success!"}

if __name__ == '__main__':
    uvicorn.run(app, host='127.0.0.1', port=5566)

rpc.py

import rpyc
from rpyc.utils.server import ThreadedServer
from apscheduler.schedulers.background import BackgroundScheduler

class SchedulerService(rpyc.Service):

    class exposed_Monitor(object):   # exposing names is not limited to methods :)
        def __init__(self, callback):
            # 这里需要用rpyc.async_异步加载回调函数
            self.callback = rpyc.async_(callback)

        def exposed_add_job(self, *args, **kwargs):
            print("添加任务："， args, kwargs)
            return scheduler.add_job(self.callback, *args, **kwargs)

        def exposed_pause_job(self, job_id, jobstore=None):
            return scheduler.pause_job(job_id, jobstore)

        def exposed_resume_job(self, job_id, jobstore=None):
            return scheduler.resume_job(job_id, jobstore)

        def exposed_remove_job(self, job_id, jobstore=None):
            scheduler.remove_job(job_id, jobstore)


if __name__ == '__main__':
    scheduler = BackgroundScheduler()
    scheduler.start()
    protocol_config = {'allow_public_attrs': True}
    server = ThreadedServer(SchedulerService, port=12345, protocol_config=protocol_config)
    try:
        server.start()
    except (KeyboardInterrupt, SystemExit):
        pass
    finally:
        scheduler.shutdown()

2.3 rpyc服务添加定时任务为http请求

2.3.1rpyc服务后端

• 见 rpyc调度器服务

2.3.2调用示例

#!usr/bin/env python

import os
import rpyc
import logging

from app.conf import common_config

fast_api_env = os.environ.get('FAST_API_ENV')


def from_crontab(cron):
    values = cron.split(' ')
    return {
        'minute': values[0],
        'hour': values[1],
        'day': values[2],
        'month': values[3],
        'day_of_week': values[4],
    }


def run_task():
    """启动任务"""
    try:
        rpyc_conn = rpyc.connect(common_config.HOST, common_config.PORT)
    except ConnectionRefusedError:
        logging.error('run scheduler tasks err, connect err!')
        return False
    # bgsrv = rpyc.BgServingThread(rpyc_conn)   # 异步回调需要开启线程处理器
    # creates a bg thread to process incoming events
    logging.info('run scheduler tasks')
    # 添加job1
    rpyc_conn.root.add_request_job(id=f'job1', task_id='job1',, trigger='cron',**from_crontab('* * * * *'))
    # 添加job2
    rpyc_conn.root.add_request_job(id=f'job2', task_id='job2', trigger='cron', **from_crontab('* * * * *'))

    # bgsrv.stop()
    rpyc_conn.close()
    logging.info('run scheduler tasks success!')

2.3.3服务端接口

from fastapi import APIRouter, Depends, Query
from sqlalchemy.orm import Session

from app.schemas.response import Response
from app.api.utils.common import get_db

# 具体的定时任务内容
from app.service.jobs import job1, job2

router = APIRouter()

@router.get("/run-task", response_model=Response)
def run_task(
        task_id: str = Query(...),
        db: Session = Depends(get_db),
):
    """调用接口，自定义的task_id对应一个job任务，可选择数据库存储映射关系"""
    if task_id == 'job1':
        job1(db)
    elif task_id == 'job2':
        job2(db)
    else:
        return {'code': 1, 'error': '无效任务类型'}
    return {'msg': 'completed!'}

2.4问题

1. 2.1这种方式会导致业务代码需要在rpyc服务在写一遍，肯定造成业务代码脱离项目

• 如果定时业务代码和原项目代码不关联，可以采用此种方案

2. 2.2通过rpyc服务回调添加定时任务，这种方式的弊端是FastAPI服务端与rpyc服务需要长连接，且不能做持久化存储，服务之间强关联

• 如果不考虑长连接影响可以采用此方案

3. 2.3rpyc服务添加定时任务为http请求，这种方式会绕一个大弯

• 解决了服务之间强关联，个人推荐使用方案，如果觉得low，建议采用其它方案(celery+MQ)做定时任务