基于websocket的celery任务状态监控

1. 目的
曾经想向前台实时返回Celery任务的状态监控,也查看了很多博客,但是好多也没能如愿,因此基于网上已有的博客已经自己的尝试,写了一个小的demo,实现前台实时获取后台传输的任务状态。

2. 准备
本篇文章使用的是Flask框架,安装celery,celery采用redis作为存储。同时用到了Flask-SocketIO建立websocket。同时还用到了协程库eventlet(这个是Flask-SocketIO文档建议的,链接文档)。

3. 实现
demo仿照其他例子实现了一个简单的后台任务监控。我们直接上代码吧,下面是server端代码:

# -*- utf-8 -*-
# app.py
import time
import uuid
from flask import Flask, render_template, request, make_response, jsonify
from flask_socketio import SocketIO
from celery import Celery
import eventlet
from flask_redis import FlaskRedis
eventlet.monkey_patch()

app = Flask(__name__)

app.config['BROKER_URL'] = 'redis://localhost:6379/0'
app.config['CELERY_RESULT_BACKEND'] = 'redis://localhost:6379/0'
app.config['CELERY_ACCEPT_CONTENT'] = ['json', 'pickle']
app.config['REDIS_URL'] = 'redis://localhost:6379/0'

socketio = SocketIO(app, async_mode='eventlet',message_queue=app.config['CELERY_RESULT_BACKEND'])
redis = FlaskRedis(app)

celery = Celery(app.name)
celery.conf.update(app.config)

#模拟后台耗时任务
@celery.task
def background_task(uid):
    sid = redis.get(uid)
    socketio.emit('info', {'data': 'Task starting ...', 'time': time.time() * 1000 },room=sid, namespace='/task')
    socketio.sleep(4)
    socketio.emit('info', {'data': 'Task running!', 'time': time.time() * 1000 }, room=sid, namespace='/task')
    socketio.sleep(5)
    socketio.emit('info', {'data': 'Task complete!', 'time': time.time()*1000 }, room=sid, namespace='/task')

#建立链接时把sid传到浏览器端保存。
@socketio.on('connect', namespace='/task')
def connect_host():
    sid = request.sid
    socketio.emit('hostadd', {'sid': sid}, room=sid, namespace='/task')

#将每一个客户端生成一个uuid存放在cookie中
@app.route('/')
def index():
    if not request.cookies.get('host_uid', None):
        uid = uuid.uuid1().get_hex()
        response = make_response(render_template('index.html'))
        response.set_cookie('host_uid', uid)
        return response
    return render_template('index.html')

@app.route('/task')
def start_background_task():
    uid = request.cookies.get('host_uid')
    background_task.delay(uid)
    return 'Started'

#设置sid建立链接后浏览器将sid传送到server,并将uid与sid映射存放在redis里面,默认保留12小时
@app.route('/setsid', methods=['POST'])
def set_uid():
    data = request.json
    uid = request.cookies.get('host_uid')
    redis.set(uid, data['sid'])
    redis.expire(uid, 3600 * 12)
    return jsonify({'success': True})

if __name__ == '__main__':
    socketio.run(app, host='0.0.0.0', port=5000, debug=True)

如果不想使用debug模式的话,可以用gunicorn运行,命令如下所示:

gunicorn --worker-class eventlet -w 1 app:app

使用上述命令需要注意,由于gunicorn负载均衡算法的限制,文档建议worker数量为1,我测试过大于1,确实会出问题。
前端代码如下,index.html:

<!DOCTYPE html>
<html>
<head>
    <title>test</title>
    <script type="text/javascript" src="https://cdnjs.cloudflare.com/ajax/libs/jquery/2.2.4/jquery.min.js"></script>
    <script type="text/javascript" src="https://cdnjs.cloudflare.com/ajax/libs/socket.io/2.1.0/socket.io.js"></script>
</head>
<body>
    <h3>Logging</h3>
    <p id="log"></p>
    <button id="background">Execute</button>
    <script type="text/javascript">
        $(document).ready(function () {
            namespace = '/task';
            socket = io.connect('http://' + document.domain + ':' + location.port + namespace);
            socket.on('hostadd', function(msg){
                console.log(msg.sid);
                $.ajax({
                    url: "{{ url_for('set_uid') }}",
                    data: JSON.stringify({ sid: msg.sid }),
                    type: 'post',
                    dataType: 'json',
                    contentType: "application/json; charset=utf-8"
                })
            });
            socket.on('info', function (msg) {
                console.log('Recived: ' + msg.data);
                var t = new Date(msg.time);
                $('#log').append('Recived: ' + t.toLocaleTimeString() + '->' + msg.data + '<br>');
            });
            $('#background').on('click', function(){
                $.get("{{ url_for('start_background_task') }}");
            });
        });
    </script>
</body>
</html>

GitHub地址:https://github.com/junfenggoo...

相关推荐