在web开发中我们经常会遇到一些耗时的操作,比如发送邮件/短信,执行各种任务等等,这时我们会采取异步的方式去执行这些任务,而celery就是这样的一个异步的分布式任务处理框架,官方文档
今天,我们的主题是celery如何与flask一起工作,我们都知道,flask是一个非常小巧的web框架,有许许多多的扩展,celery也不例外,我们先看下目前常用的几个flask-celery的扩展:
除这些扩展之外,其实flask的官方文档中已经给出了在flask中使用celery的方式,不过,那是一个单文件中运行flask的demo,在实际项目中使用,还是有许多需要注意的地方,接下来,我们就一起探究下如何在flask项目中使用celery。
├── celery_task # celery任务相关
│ ├── __init__.py
│ ├── tasks.py
│ └── test.py
├── manage.py # celery worker实例
├── requirements.txt # 依赖包
└── test_api # flask 项目
├── api # 蓝本相关
│ ├── __init__.py
│ └── v1
│ ├── __init__.py
│ └── views.py
├── extensions.py # 扩展初始化
├── __init__.py # flask app
├── models.py # 模型文件
└── settings.py # 配置文件
本项目中没有使用扩展,只是基于官方文档中的示例做进一步的应用。
from celery import Celery
def make_celery(app):
celery = Celery(
app.import_name,
backend=app.config['CELERY_RESULT_BACKEND'],
broker=app.config['CELERY_BROKER_URL']
)
celery.conf.update(app.config)
class ContextTask(celery.Task):
def __call__(self, *args, **kwargs):
with app.app_context():
return self.run(*args, **kwargs)
celery.Task = ContextTask
return celery
这是一个celery的工厂函数,使用flask app中的配置设置celery相关的属性,并且更改了celery对象的Task,使其能够使用flask的应用上下文,这一点非常重要。我们将这段代码放置到flask项目初始化文件中去也就是testapi/__init_\.py
celerytask/__init_\.py
rom test_api import create_app, make_celery
app = create_app()
celery = make_celery(app)
class MyTask(celery.Task): # celery 基类
def on_success(self, retval, task_id, args, kwargs):
# 执行成功的操作
print('MyTasks 基类回调,任务执行成功')
return super(MyTask, self).on_success(retval, task_id, args, kwargs)
def on_failure(self, exc, task_id, args, kwargs, einfo):
# 执行失败的操作
# 任务执行失败,可以调用接口进行失败报警等操作
print('MyTasks 基类回调,任务执行失败')
return super(MyTask, self).on_failure(exc, task_id, args, kwargs, einfo)
这里我对Task做了进一步的定制,用于添加一些任务信息。
import datetime
import time
import os
import random
from flask import current_app
from test_api.models import User
from test_api.extensions import db
from celery_task import celery, MyTask
@celery.task(bind=True, base=MyTask)
def apptask(self):
print(current_app.config)
print("==============%s " % current_app.config["SQLALCHEMY_DATABASE_URI"])
print("++++++++++++++%s " % os.getenv("DATABASE_URL"))
time.sleep(5)
user = User(username="user%s" % random.randint(1,100))
db.session.add(user)
db.session.commit()
return 'success'
这个任务很简单,使用User模型类异步向数据库中添加数据,为了体现耗时操作,使用sleep函数模拟。
test_api/api/v1/views.py
from flask import jsonify
from celery_task.tasks import apptask
from test_api.api.v1 import api_v1
from test_api.extensions import db
from flask import current_app
@api_v1.route("/", methods=["GET"])
def index():
r = apptask.apply_async()
return jsonify({"status": "success"})
视图函数非常的简单,只做了提交任务的操作。
为了避免循环导入问题,我们在项目根目录下新建manage.py
from test_api import create_app, make_celery
app = create_app()
celery = make_celery(app)
if __name__ == '__main__':
app.run()
这个文件只用来启动celery,启动命令如下:
# celery worker -A manage:celery -l debug
看到如下输出,表明启动成功:
-------------- celery@test-3 v4.4.0 (cliffs)
--- ***** -----
-- ******* ---- Linux-3.10.0-693.2.2.el7.x86_64-x86_64-with-centos-7.4.1708-Core 2020-03-03 21:14:13
- *** --- * ---
- ** ---------- [config]
- ** ---------- .> app: test_api:0x7f87c31a4e48
- ** ---------- .> transport: redis://127.0.0.1:6379/3
- ** ---------- .> results: redis://127.0.0.1:6379/4
- *** --- * --- .> concurrency: 2 (prefork)
-- ******* ---- .> task events: OFF (enable -E to monitor tasks in this worker)
--- ***** -----
-------------- [queues]
.> celery exchange=celery(direct) key=celery
[tasks]
. celery.accumulate
. celery.backend_cleanup
. celery.chain
. celery.chord
. celery.chord_unlock
. celery.chunks
. celery.group
. celery.map
. celery.starmap
. celery_task.tasks.apptask
[2020-03-03 21:14:13,632: DEBUG/MainProcess] | Worker: Starting Hub
[2020-03-03 21:14:13,632: DEBUG/MainProcess] ^-- substep ok
[2020-03-03 21:14:13,632: DEBUG/MainProcess] | Worker: Starting Pool
[2020-03-03 21:14:13,690: DEBUG/MainProcess] ^-- substep ok
[2020-03-03 21:14:13,691: DEBUG/MainProcess] | Worker: Starting Consumer
[2020-03-03 21:14:13,691: DEBUG/MainProcess] | Consumer: Starting Connection
[2020-03-03 21:14:13,708: INFO/MainProcess] Connected to redis://127.0.0.1:6379/3
[2020-03-03 21:14:13,708: DEBUG/MainProcess] ^-- substep ok
[2020-03-03 21:14:13,708: DEBUG/MainProcess] | Consumer: Starting Events
[2020-03-03 21:14:13,718: DEBUG/MainProcess] ^-- substep ok
[2020-03-03 21:14:13,718: DEBUG/MainProcess] | Consumer: Starting Mingle
[2020-03-03 21:14:13,718: INFO/MainProcess] mingle: searching for neighbors
[2020-03-03 21:14:14,743: INFO/MainProcess] mingle: all alone
[2020-03-03 21:14:14,743: DEBUG/MainProcess] ^-- substep ok
[2020-03-03 21:14:14,744: DEBUG/MainProcess] | Consumer: Starting Gossip
[2020-03-03 21:14:14,748: DEBUG/MainProcess] ^-- substep ok
[2020-03-03 21:14:14,748: DEBUG/MainProcess] | Consumer: Starting Heart
[2020-03-03 21:14:14,750: DEBUG/MainProcess] ^-- substep ok
[2020-03-03 21:14:14,750: DEBUG/MainProcess] | Consumer: Starting Tasks
[2020-03-03 21:14:14,756: DEBUG/MainProcess] ^-- substep ok
[2020-03-03 21:14:14,756: DEBUG/MainProcess] | Consumer: Starting Control
[2020-03-03 21:14:14,759: DEBUG/MainProcess] ^-- substep ok
[2020-03-03 21:14:14,759: DEBUG/MainProcess] | Consumer: Starting event loop
[2020-03-03 21:14:14,759: DEBUG/MainProcess] | Worker: Hub.register Pool...
[2020-03-03 21:14:14,760: INFO/MainProcess] celery@test-3 ready.
[2020-03-03 21:14:14,760: DEBUG/MainProcess] basic.qos: prefetch_count->8
启动flask:
# flask run
* Serving Flask app "test_api" (lazy loading)
* Environment: development
* Debug mode: on
* Running on http://0.0.0.0:5000/ (Press CTRL+C to quit)
* Restarting with stat
* Debugger is active!
* Debugger PIN: 237-492-852
调试接口:
# curl http://127.0.0.1:5000/api/v1/
{
"status": "success"
}
查看celery日志:
[2020-03-03 21:17:31,330: WARNING/ForkPoolWorker-2]
[2020-03-03 21:17:31,330: DEBUG/MainProcess] Task accepted: celery_task.tasks.apptask[5f27a148-161f-4485-931f-17d94637168e] pid:2341
[2020-03-03 21:17:36,391: WARNING/ForkPoolWorker-2] MyTasks 基类回调,任务执行成功
[2020-03-03 21:17:36,392: INFO/ForkPoolWorker-2] Task celery_task.tasks.apptask[5f27a148-161f-4485-931f-17d94637168e] succeeded in 5.0624741315841675s: 'success'
任务执行成功,查看数据库数据:
mysql> select * from user order by id;
+----+----------+
| id | username |
+----+----------+
| 1 | user26 |
| 2 | user69 |
| 3 | user71 |
| 4 | user35 |
| 5 | user13 |
| 6 | user54 |
| 7 | user88 |
| 8 | user63 |
| 9 | user87 |
| 10 | user90 |
| 11 | user3 |
| 12 | user18 |
| 13 | user65 |
+----+----------+
数据已被插入,实验成功!
有几个坑希望大家注意下
出错文件: testapi/__init_\.py
import os
import click
from flask import Flask, jsonify
from test_api.api.v1 import api_v1 # 蓝图在上方导入,循环报错产生
from test_api.settings import config
from test_api.models import User
from celery import Celery
def make_celery(app):
...
def create_app(config_name=None):
if config_name is None:
config_name = os.getenv('FLASK_ENV', 'development')
app = Flask('test_api')
app.config.from_object(config[config_name])
register_extensions(app)
register_blueprints(app)
register_commands(app)
register_errors(app)
return app
# 注册蓝图函数
def register_blueprints(app):
app.register_blueprint(api_v1, url_prefix='/api/v1')
启动celery和请求接口时均会报错,错误堆栈如下:
from test_api import create_app, make_celery
File "/tmp/test/test_api/__init__.py", line 5, in <module>
from test_api.api.v1 import api_v1
File "/tmp/test/test_api/api/v1/__init__.py", line 9, in <module>
from test_api.api.v1 import views
File "/tmp/test/test_api/api/v1/views.py", line 2, in <module>
from celery_task.tasks import apptask
File "/tmp/test/celery_task/__init__.py", line 1, in <module>
from test_api import create_app, make_celery
ImportError: cannot import name 'create_app'
将蓝图的导入下放置蓝图注册函数中testapi/__init_\.py:
...
def register_blueprints(app):
from test_api.api.v1 import api_v1
app.register_blueprint(api_v1, url_prefix='/api/v1')
...
提交任务,celery报错如下:
...
options = self.get_options(sa_url, echo)
File "/tmp/py3/lib/python3.6/site-packages/flask_sqlalchemy/__init__.py", line 575, in get_options
self._sa.apply_driver_hacks(self._app, sa_url, options)
File "/tmp/py3/lib/python3.6/site-packages/flask_sqlalchemy/__init__.py", line 877, in apply_driver_hacks
if sa_url.drivername.startswith('mysql'):
AttributeError: 'NoneType' object has no attribute 'drivername'
通过调试我发现,flask的app的配置是可以拿到的,因为我们在工厂函数中推送了应用上下文,我的数据库配置信息是以键值的形式写在了.env文件中,这也是目前flask推荐的方式。那为什么celery取不到数据库连接配置呢?其实,启动celery的app和我们web服务所用app是两个独立的app,celery无法通过.env中的环境变量取到相应的值,这里有三种解决办法:
不使用环境变量的方式,直接将相关信息写在配置文件中例如: SQLALCHEMY_DATABASE_URI = "mysql+pymysql://xxx:xxx@127.0.0.1:3306/test?charset=utf8"
相比之下,方案三是采纳比较多的,于是我们在test_api/settings.py文件中加入如下代码:
from dotenv import find_dotenv, load_dotenv
load_dotenv(find_dotenv())
find_dotenv函数会在当前以及父目录中搜寻.env文件,load_dotenv函数则负责加载环境变量。如此,大功告成。我们可以继续愉快撸代码啦。
附:项目源码
免责声明:本站发布的内容(图片、视频和文字)以原创、转载和分享为主,文章观点不代表本网站立场,如果涉及侵权请联系站长邮箱:is@yisu.com进行举报,并提供相关证据,一经查实,将立刻删除涉嫌侵权内容。