在Ubuntu下后台持续运行Python程序

川长思鸟来 2022-05-14 10:59 449阅读 0赞

由于项目的需要，现在需要通过一种方式，让Python程序能够在会话退出后继续在后台运行，并将屏幕的输出重定向到文件中去。

本文的示例代码已经上传到[http://download.csdn.net/detail/mrbcy/9789067][http_download.csdn.net_detail_mrbcy_9789067]

## 参考资料 ##

[http://blog.csdn.net/loonger\_leon/article/details/5764552][http_blog.csdn.net_loonger_leon_article_details_5764552]

[http://www.cnblogs.com/hester/p/5575658.html][http_www.cnblogs.com_hester_p_5575658.html]

## 测试程序 ##

先写一个测试程序，用于输出日志和打印到控制台。

#-*- coding: utf-8 -*-
    import logging
    import time
    from logging.handlers import RotatingFileHandler
    
    
    def func():
        init_log()
        while True:
            print "output to the console"
            logging.debug("output the debug log")
            logging.info("output the info log")
            time.sleep(3);
    
    
    def init_log():
        logging.getLogger().setLevel(logging.DEBUG)
        console = logging.StreamHandler()
        console.setLevel(logging.DEBUG)
        formatter = logging.Formatter('%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
        console.setFormatter(formatter)
        logging.getLogger().addHandler(console)
    
        # add log ratate
        Rthandler = RotatingFileHandler("backend_run.log", maxBytes=10 * 1024 * 1024, backupCount=100,
                                        encoding="gbk")
        Rthandler.setLevel(logging.INFO)
        # formatter = logging.Formatter('%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
        Rthandler.setFormatter(formatter)
        logging.getLogger().addHandler(Rthandler)
    
    
    
    if __name__ == '__main__':
        func()

## 后台启动Python脚本 ##

可以使用下面的命令来启动上面的脚本，让Python在后台运行。

nohup python -u main.py > test.out 2>&1 &

来解释一下这几个命令的参数。这一段来自[http://blog.csdn.net/tswisdom/article/details/8548710][http_blog.csdn.net_tswisdom_article_details_8548710]

其中 0、1、2分别代表如下含义：
     0 – stdin (standard input)
     1 – stdout (standard output)
     2 – stderr (standard error)
    nohup python -u main.py > test.out 2>&1 &
    nohup+最后面的& 是让命令在后台执行
    >out.log 是将信息输出到out.log日志中
    2>&1 是将标准错误信息转变成标准输出，这样就可以将错误信息输出到out.log 日志里面来。

运行命令后，会返回一个pid。像下面这样：

[1] 9208

后续可以学习Hadoop它们，把pid存起来，到时候stop的时候就把它杀掉。

## **在后台利用GPU跑程序：** ##

CUDA_VISIBLE_DEVICES=2,3 nohup python -u search.py > out.log 2>&1 &

## 跟踪输出文件变化 ##

为了验证脚本可以在后台继续运行，我们退出当前会话。然后重新连接一个Session，然后输入下面的命令来跟踪文件的输出：

tail -f test.out

输出内容如下：

output to the console
    2017-03-21 20:15:02,632 main.py[line:11] DEBUG output the debug log
    2017-03-21 20:15:02,632 main.py[line:12] INFO output the info log
    output to the console
    2017-03-21 20:15:05,635 main.py[line:11] DEBUG output the debug log
    2017-03-21 20:15:05,636 main.py[line:12] INFO output the info log
    output to the console
    2017-03-21 20:15:08,637 main.py[line:11] DEBUG output the debug log
    2017-03-21 20:15:08,638 main.py[line:12] INFO output the info log
    output to the console
    2017-03-21 20:15:11,640 main.py[line:11] DEBUG output the debug log
    2017-03-21 20:15:11,642 main.py[line:12] INFO output the info log
    output to the console
    2017-03-21 20:15:14,647 main.py[line:11] DEBUG output the debug log
    2017-03-21 20:15:14,647 main.py[line:12] INFO output the info log
    output to the console
    2017-03-21 20:15:17,653 main.py[line:11] DEBUG output the debug log
    2017-03-21 20:15:17,654 main.py[line:12] INFO output the info log
    output to the console
    2017-03-21 20:15:20,655 main.py[line:11] DEBUG output the debug log
    2017-03-21 20:15:20,656 main.py[line:12] INFO output the info log
    output to the console
    2017-03-21 20:15:23,661 main.py[line:11] DEBUG output the debug log
    2017-03-21 20:15:23,661 main.py[line:12] INFO output the info log
    output to the console
    2017-03-21 20:15:26,665 main.py[line:11] DEBUG output the debug log
    2017-03-21 20:15:26,666 main.py[line:12] INFO output the info log
    output to the console
    2017-03-21 20:15:29,670 main.py[line:11] DEBUG output the debug log
    2017-03-21 20:15:29,671 main.py[line:12] INFO output the info log

说明我们的脚本确实在后台持续运行。

## 结束程序 ##

可以直接通过之前的那个pid杀掉脚本，或者可以通过下面的命令查找pid。

ps -ef | grep python

输出的内容如下：

root      1659     1  0 17:40 ?        00:00:00 /usr/bin/python /usr/lib/python2.6/site-packages/ambari_agent/AmbariAgent.py start
    root      1921  1659  0 17:40 ?        00:00:48 /usr/bin/python /usr/lib/python2.6/site-packages/ambari_agent/main.py start
    user      8866  8187  0 20:03 ?        00:00:06 /usr/bin/python3 /usr/bin/update-manager --no-update --no-focus-on-map
    root      9208  8132  0 20:12 pts/16   00:00:00 python -u main.py
    root      9358  8132  0 20:17 pts/16   00:00:00 grep --color=auto python

可以看到我们的pid是9208，调用kill杀掉就可以了。

kill -9 9208

## 编写启动及停止脚本 ##

### 启动脚本 ###

#!/bin/sh
    
    pid=`ps -ef|grep "python -u main.py"| grep -v "grep"|awk '{print $2}'`
    
    if [ "$pid" != "" ]
    then
            echo "main.py already run, stop it first"
            kill -9 ${pid}
    fi
    
    echo "starting now..."
    
    nohup python -u main.py > test.out 2>&1 &
    
    pid=`ps -ef|grep "python -u main.py"| grep -v "grep"|awk '{print $2}'`
    
    echo ${pid} > pid.out
    echo "main.py started at pid: "${pid}

### 停止脚本 ###

#!/bin/sh
    
    pid=`ps -ef|grep "python -u main.py"| grep -v "grep"|awk '{print $2}'`
    
    if [ "$pid" != "" ]
    then
            kill -9 ${pid}
            echo "stop main.py complete"
    else
            echo "main.py is not run, there's no need to stop it"
    fi

稍后我会把实例代码上传到资料共享里面。

[http_download.csdn.net_detail_mrbcy_9789067]: http://download.csdn.net/detail/mrbcy/9789067
[http_blog.csdn.net_loonger_leon_article_details_5764552]: http://blog.csdn.net/loonger_leon/article/details/5764552
[http_www.cnblogs.com_hester_p_5575658.html]: http://www.cnblogs.com/hester/p/5575658.html
[http_blog.csdn.net_tswisdom_article_details_8548710]: http://blog.csdn.net/tswisdom/article/details/8548710