celery简介:
celery是一个异步任务队列/基于分布式消息传递的作业队列。它侧重于实时操作,但对调度支持也很好。 celery是用Python编写的,但该协议可以在任何语言实现。更多简介的请自己在网上搜索
本文目的是用python使用celery做异步任务,在centos 6.4上安装celery,并用supervisor来管理celery进程,celery采用redis做中间件的消息传输。现实中可以celery做异步请求,如发送邮箱,发送消息,请求URL等场景
环境部署
环境
系统:centos 6.4 64位
python版本:2.6.6
创建相关的目录
mkdir -p /data/{redis,logs,www}
(我会把python脚本都放在/data/www中)
安装celery
yum install python-devel python-setuptools
easy_install pip
pip install celery
pip install redis #(python中的redis模块)
安装supervisor
pip install supervisor
echo_supervisord_conf > /etc/supervisord.conf
如果在执行echo_supervisord_conf > /etc/supervisord.conf时报pkg_resources.DistributionNotFound: meld3>=0.6.5错误的话,找到supervisor-3.1.3-py2.6.egg-info/requires.txt,把文件里面meld3 >= 0.6.5注释掉,然后再执行echo_supervisord_conf > /etc/supervisord.conf就好了
查找方法:
find / | grep requires.txt
配置
vim /etc/supervisord.conf
在配置后面添加以下参数
[program:celery]
command=/usr/bin/celery worker -A tasks
directory=/data/www
stdout_logfile=/data/logs/celery.log
autostart=true
autorestart=true
redirect_stderr=true
stopsignal=QUIT
注释
;program:celery 要管理的进程名,你自己随便定义,我这定义了叫celery
;command 是启动celery的命令
;directory 是程序目录 ,因为我要启动celery,需要进入/data/www目录中才能生效的,所以这里在启动命令时,会切换到这个目录里
;autorstart 自动重启celery
;stdout_logfile 存放celery的日志路径
以上命令的大概意思就是:
进行到/data/www目录,然后执行/usr/bin/celery worker -A tasks,并把输出的日志保存到/data/logs/celery.log中,这是指定了worker模式,如果不指定,默认为prefork模式,一般你机器有几核,系统就开启几个worker进程,如果有异常,记得查看日志/data/logs/celery.log
安装redis
cd /usr/local/src
wget http://download.redis.io/releases/redis-3.0.5.tar.gz
tar xf redis-3.0.5.tar.gz
cd redis-3.0.5
make
make install
#(可用make PREFIX=/usr/local/redis install 安装到指定的路径下面 )
cp utils/redis_init_script /etc/init.d/redis
chmod a+x /etc/init.d/redis
mkdir /etc/redis
cp redis.conf /etc/redis/
简单修改redis.conf(前面数字是行号)
42 daemonize yes #进程转入后台运行
50 port 22222 #修改端口,不用默认的6379
69 bind 127.0.0.1 #绑定IP,只能本地连接
192 dir /data/redis #修改redis文件存放路径
396 requirepass DILSi3_dflka_f3i_LFKDkdf!idf #设置密码
453 maxmemory 256M #redis最大使用256M内存,我的是虚拟机,所以内存设置小
注释:
requirepass DILSi3_dflka_f3i_LFKDkdf!idf 是配置redis密码,客户端连接和关闭redis时需要此密码
修改/etc/init.d/redis(前面数字是行号)
6 REDISPORT=22222
7 EXEC=/usr/local/bin/redis-server
8 CLIEXEC=/usr/local/bin/redis-cli
9 PASSWD="DILSi3_dflka_f3i_LFKDkdf!idf" #增加行
10 PIDFILE=/var/run/redis.pid
11 CONF="/etc/redis/redis.conf"
30 $CLIEXEC -p $REDISPORT -a $PASSWD shutdown
注释:
关闭redis时,需要带上redis密码,否则失败,如果没有设置密码,则不需要带密码关闭
启动redis
service redis start
查看redis进程
ps -ef |grep redis
root 28631 1 0 04:30 ? 00:00:00 /usr/local/bin/redis-server 127.0.0.1:22222
root 28635 1542 0 04:31 pts/0 00:00:00 grep redis
编写celery脚本
cd /data/www
vim tasks.py
#!/usr/bin/env python
#coding:utf-8
import time
#导入celery相关模块、方法
from celery import Celery
from celery import platforms
#因为supervisord默认是用root运行的,必须设置以下参数为True才能允许celery用root运行
platforms.C_FORCE_ROOT = True
#配置celery,连接redis,DILSi3_dflka_f3i_LFKDkdf!idf是redis的密码,22222是端口 10是库(redis有16个库,这取第11个,即10)
config={}
config['CELERY_BROKER_URL'] = 'redis://:[email protected]:22222/10'
config['CELERY_RESULT_BACKEND'] = 'redis://:[email protected]:22222/10'
#不需要返回任务状态,即设置以下参数为True
config['CELERY_IGNORE_RESULT'] = True
app = Celery("tasks", broker=config['CELERY_BROKER_URL'])
app.conf.update(config)
@app.task
def output(num):
"""输出传进来的数字,在输出前等待2秒"""
time.sleep(2)
print num
启动supervisord
/usr/bin/supervisord
查看是否启动成功
[root@drfdai www]# ps -ef |grep supervisor
root 28656 1 0 04:52 ? 00:00:00 /usr/bin/python /usr/bin/supervisord
root 28692 1542 0 04:53 pts/0 00:00:00 grep supervisor
启动celery:
supervisorctl start celery
查看celery是否连接成功redis
tailf /data/logs/celery.log
能看到[2015-11-16 05:05:27,938: WARNING/MainProcess] celery@drfdai ready.说明连接成功
测试
在终端【终端1】中打开实时日志:
[root@drfdai www]# tailf /data/logs/celery.log
打开另一终端【终端2】测试:
[root@drfdai ~]# cd /data/www
[root@drfdai www]# python
Python 2.6.6 (r266:84292, Jul 23 2015, 15:22:56)
[GCC 4.4.7 20120313 (Red Hat 4.4.7-11)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> from tasks import output
>>> for i in xrange(20):
... output.apply_async(args=[i])
...
切换到【终端1】可以看到,每隔2秒就会打印一行日志
warnings.warn(CDeprecationWarning(W_PICKLE_DEPRECATED))
[2015-11-16 05:58:09,687: WARNING/MainProcess] celery@drfdai ready.
[2015-11-16 05:59:18,816: WARNING/Worker-1] 0
[2015-11-16 05:59:19,826: WARNING/Worker-1] 1
[2015-11-16 05:59:20,830: WARNING/Worker-1] 2
[2015-11-16 05:59:21,836: WARNING/Worker-1] 3
[2015-11-16 05:59:22,841: WARNING/Worker-1] 4
[2015-11-16 05:59:23,848: WARNING/Worker-1] 5
[2015-11-16 05:59:24,853: WARNING/Worker-1] 6
[2015-11-16 05:59:25,858: WARNING/Worker-1] 7
[2015-11-16 05:59:26,864: WARNING/Worker-1] 8
[2015-11-16 05:59:27,873: WARNING/Worker-1] 9
[2015-11-16 05:59:28,884: WARNING/Worker-1] 10
[2015-11-16 05:59:29,892: WARNING/Worker-1] 11
[2015-11-16 05:59:30,899: WARNING/Worker-1] 12
[2015-11-16 05:59:31,908: WARNING/Worker-1] 13
[2015-11-16 05:59:32,912: WARNING/Worker-1] 14
[2015-11-16 05:59:33,920: WARNING/Worker-1] 15
[2015-11-16 05:59:34,926: WARNING/Worker-1] 16
[2015-11-16 05:59:35,933: WARNING/Worker-1] 17
[2015-11-16 05:59:36,939: WARNING/Worker-1] 18
[2015-11-16 05:59:37,944: WARNING/Worker-1] 19
当我们在【终端2】中调用此函数后,执行并没有受到time.sleep(2)的影响,执行后马上就结束了,而【终端1】每隔2秒就输出1行日志,这日志是函数output输出来的。
实用代码片断:
vim common/tasks.py
#导入发送邮件模块
from email.mime.text import MIMEText
import smtplib
import json
import datetime
from celery import Celery
from celery import platforms
import requests
#导入配置文件
from config import setting
#发送邮箱配置
smtp = setting.smtp
user = setting.user
passwd = setting.passwd
此处省略……
@celery.task
def sendMail(mail):
'''
to_list 接收者邮件,每个邮件地址用","分隔,str格式
subuect 邮件主题,str格式
context 邮件内容,str格式
'''
to_list = mail.get('to_list')
subject = mail.get('subject')
context = mail.get('context')
msg = MIMEText(context,'html','utf-8')
msg['Subject'] = subject
msg['From'] = "linux系统运维" % user
to_list = list(to_list.split(','))
msg['To'] = ",".join((to_list))
try:
s = smtplib.SMTP()
s.connect(smtp)
s.login(user,passwd)
s.sendmail(user,to_list,msg.as_string())
s.close()
return json.dumps({"redid":0,"redmsg":"发送成功"})
except Exception,e:
print 'Error: %s' % e
return json.dumps({"redid":-1,"redmsg":"发送失败"})
在实际项目中调用此方法,做异步邮箱发送,发送后不需要等待返回的结果
#/usr/bin/env python
#coding:utf-8
from common.tasks import sendMail
sendMail.delay(dict=(to_list ='[email protected]',subject ='邮件发送测试',context ='