3

I want to run some script of python every 24 hours, at every 23:59:58 i want to kill the the job. I want to run python immediately. Thus I use this syntax bellow

* */24 * * * cd /ftp/ftp1 && timeout -s 9 86398 python2.6 lpr_10.10.252.121.py 10.10.252.121 450 >> res_10.10.252.121.txt

But on cron log, their script are not running. I've try to /etc/rc.d/init.d/crond restart

Here are the log cron

May 18 10:13:45 cisco-cumc crontab[31368]: (root) REPLACE (root)
May 18 10:13:45 cisco-cumc crontab[31368]: (root) END EDIT (root)
May 18 10:14:01 cisco-cumc crond[5090]: (root) RELOAD (/var/spool/cron/root)
May 18 10:15:24 cisco-cumc crontab[31449]: (root) BEGIN EDIT (root)
May 18 10:15:42 cisco-cumc crontab[31449]: (root) REPLACE (root)
May 18 10:15:42 cisco-cumc crontab[31449]: (root) END EDIT (root)
May 18 10:15:46 cisco-cumc crontab[31451]: (root) LIST (root)
May 18 10:15:57 cisco-cumc crond[5090]: (CRON) INFO (Shutting down)
May 18 10:15:57 cisco-cumc crond[31471]: (CRON) STARTUP (1.4.4)
May 18 10:15:57 cisco-cumc crond[31471]: (CRON) INFO (RANDOM_DELAY will be scaled with factor 73% if used.)
May 18 10:15:57 cisco-cumc crond[31471]: (CRON) INFO (running with inotify support)
May 18 10:15:57 cisco-cumc crond[31471]: (CRON) INFO (@reboot jobs will be run at computer's startup.)

Tried ps -ax | grep python

no python script running

So how to running script every 24 hours and start immediately? My machine running on centos 6.4 thanks

Yohanim
  • 157
  • 1
  • 3
  • 10
  • It's difficult to control seconds in a cronjob directly. Your cronjob seem to be set to run At every minute past every 24th hour. Is that what you need? Note that it will stop running the command once clock strikes 01:00:00. Also, is the command given correctly, including paths? You can make a script file and set proper environment paths and then execute it in a neat cron command. – Nikhil_CV May 18 '18 at 05:08

1 Answers1

10

You're going about this wrong. cron is designed to start jobs unattended at known future times; it's not designed to start them now, nor is it designed to stop them. When you torture a tool to do something it wasn't supposed to do, you can get into a mess, as you are now.

Let the job take care of shutting itself down. Consider something like

#!/bin/bash
cd /ftp/ftp1
let pause=86400+`date +%s -d 0000`-`date +%s`-10
python2.6 lpr_10.10.252.121.py 10.10.252.121 450 >> res_10.10.252.121.txt &
sleep $pause
kill -15 %1
exit 0

This tiny shellscript calculates the number of seconds left between now and next midnight, minus ten seconds (I could almost certainly have done that more elegantly, but c'est la vie). It then starts your python job in the background and goes to sleep for that many seconds, so waking up at 23:59:50, then kills the backgrounded job and exits. For a more professional approach the job should also check at startup whether another copy is running, and terminate (with an error to your monitoring system) if it is.

You can start that safely from cron once a day at midnight:

0 0 * * *  /usr/local/bin/noddy-printer-script

You can also start it immediately from the shell, because it will adjust the sleep gap accordingly.

Note that Nikhil_CV makes a good point above; */24 doesn't mean every hour until 25 but instead means when hour is 0 or 24, which since the latter never happens means only when hour=0, which is why your job's not running at 10am. Come next midnight, cron is going to spend the following hour starting a copy of the job every single minute, which is probably not what you want. Note also that your CentOS 6 box is hopelessly out-of-patch and vulnerable and you should bring it up to C6.9 immediately.

MadHatter
  • 78,442
  • 20
  • 178
  • 229