Celery 服务搭建

整个项目工程如下

 

__init__.py

"""
注意点：python3.7 需要执行 pip install --upgrade https://github.com/celery/celery/tarball/master
否则会报 from . import async, base

SyntaxError: invalid syntax

celery -A __init__  worker --concurrency=5 -l INFO -Q celery,save_redis
celery -A __init__ worker -l info -Q save_mongo
cd  /Users/admin/PycharmProjects/function_test/adminSyS/mq&&celery -A __init__  worker --concurrency=5


celery 启动  --autoscale=10,3 当worker不足时自动加3-10个
celery -A __init__  worker --concurrency=5 -l INFO -Q celery,save_redis2,save_redis --autoscale=10,3

supervisor 配置
[program:celery]
directory = /data/app/adminSyS/mq
command = /data/app/adminSyS/venv/bin/celery -A __init__  worker --concurrency=5 -l INFO -Q celery,save_redis
autostart = true
autorestart = true
startsecs = 5
startretries = 3

监控：
    文档 https://flower-docs-cn.readthedocs.io/zh/latest/install.html
    pip install flower
 celery flower --broker=redis://:z1234567@47.93.235.228:6379/5 --port=5555
 测试 http://47.93.235.228:9114
"""
import os
import sys
sys.path.append(os.path.abspath(os.path.dirname(__file__)) + "/..")
from celery import Celery

app = Celery(include=["mq.tasks"])
app.config_from_object("celeryconfig")  # 指定配置文件

 

celeryconfig.py

import os
import sys

sys.path.append(os.path.abspath(os.path.dirname(__file__)) + "/..")
from kombu import Queue, Exchange

BROKER_URL = "redis://:redis@127.0.0.1:6379/5"  # 消息代理
CELERY_RESULT_BACKEND = "redis://:redis@127.0.0.1:6379/6"  # 任务执行结果存放地址
CELERY_TASK_SERIALIZER = "json"  # 任务序列化和翻序列化的方案
CELERY_RESULT_SERIALIZER = "json"  # 读取任务结果序列化方式
CELERY_TASK_RESULT_EXPIRES = 60 * 60 * 24  # 任务过期时间
CELERY_ACCEPT_CONTENT = ["json", "msgpack"]  # 接受的任务类型
BROKER_TRANSPORT_OPTIONS = {'visibility_timeout': 86400}   # 任务重发时间
CELERYD_MAX_TASKS_PER_CHILD = 400  # 每个worker执行了多少任务就会自动重启防止内存泄漏

# 指定任务队列  save_redis save_redis2
CELERY_QUEUES = (
    Queue("default", Exchange("default"), routing_key="default"),
    Queue("save_redis", Exchange("save_redis"), routing_key="save_redis"),
    Queue("save_redis2", Exchange("save_redis2"), routing_key="save_redis2")
)

# mq.tasks.taskA  mq.tasks.taskB  具体函数名
CELERY_ROUTES = {
    'mq.tasks.taskA': {"queue": "save_redis", "routing_key": "save_redis"},
    'mq.tasks.taskB': {"queue": "save_redis2", "routing_key": "save_redis2"}
}

 

tasks.py

from mq import app
import redis
import time

rds = redis.Redis(
    host="localhost",
    port=6379,
    db=7,
    password="redis")


# 客户端

@app.task
def taskA(x, y):
    time.sleep(10)
    rds.setex(name="taskA", value=x + y, time=3600)
    return x + y


@app.task
def taskB(x, y, c):
    time.sleep(10)
    rds.setex(name="taskB{}".format(c), value=x + y, time=3600)
    return x + y

 

test.py

import sys
import os

sys.path.append(os.path.abspath(os.path.dirname(__file__)) + "/..")
from mq.tasks import taskA, taskB

#re1 = taskA.delay(100, 200)
# print(re1)
# print(re1.status)

# 服务端

for i in range(10):
    re2 = taskB.delay(100, i, i)
    print(re2)
    print(re2.status)

 

 

celery客户端启动  save_redis2,save_redis 代表函数名    --autoscale=10,3 当work不够用时自动起3-10个进程

#
celery -A __init__  worker --concurrency=5 -l INFO -Q celery,save_redis2,save_redis --autoscale=10,3

 #启动方式 -Q 启动指定队列

 ../venv/bin/celery -A __init__ worker --concurrency=5 -l INFO -Q celery,save_redis2,save_redis --autoscale=10,3

# 启动所有队列
../venv/bin/celery -A __init__ worker --concurrency=5 -l INFO  --autoscale=5,3


# 启动方式打印log
command=/home/op/saiyan_game_center/venv/bin/celery -A __init__ worker --concurrency=8 -l INFO -Q upload_box_task --autoscale=5,3 --logfile=/home/op/saiyan_game_center/logs/log.log

 

celery web监控

文档：https://flower-docs-cn.readthedocs.io/zh/latest/install.html

安装：pip install flower

启动：celery flower --broker=代理url --port=5555

celery flower --broker=redis://:redis@127.0.0.1:6379/5 --port=5555