0

I've seen suggestions how I can terminate my script if another instance of it is already running. My problem is the other way around: I would like to find a way to just kill all other instances of the same script -- if any -- and continue with executing my script. Is there any way to do that?

Zeitounator
  • 757
  • 4
  • 11
  • Does this answer your question? [Killing the previous instances of a script before running the same Unix script](https://unix.stackexchange.com/questions/213288/killing-the-previous-instances-of-a-script-before-running-the-same-unix-script) – αғsнιη Nov 06 '20 at 06:07
  • 1
    @αғsнιη: Kinf of, but in a very roundabout way, with loops and `if`s. The answer we have here is a one-liner! :-) –  Nov 06 '20 at 06:18
  • @Ardwena I doubt if the answer here could find all running instances. I don't know, maybe it does but never used `pidof`, btw, retracted my vote for duplicate. however what if same script was run with different name? or same script name was running but with different codes? – αғsнιη Nov 06 '20 at 08:56
  • 1
    @αғsнιη: So far it works correctly. Here's what I am doing: My script spins off a bunch of background processes, mostly monitoring, logging, reporting and stuff -- nothing too important. Every now and then, a new idea comes to mind, and I revise the script a bit, and then just run it. I don't want to specifically go hunting for the old instances, I just want the new revision to take effect. (Lazy, I know :) So, this thing does the job. Tested. –  Nov 06 '20 at 09:15

2 Answers2

2

You can do something like... PIDS=$(pidof -x nameofyourscript) to get all the pids, if any, including scripts. Then you can just kill $PIDS

EDIT: As the above would kill yout current script too, then revise the answer, adding option -o like this: PIDS=$(pidof -x -o $$ nameofyourscript) That -o $$ would make it omit the pid of your current script.

And better yet, instead of hardcoding nameofyourscript, just get its own name from $0:

kill $(pidof -x -o $$ $0)

So put that near the beginning of your script, maybe with a 2>/dev/null, in case it finds no other instances, and problem solved.

  • 2
    Thank you! That does it. :-) –  Nov 06 '20 at 05:39
  • 2
    This is both racy and unportable. On top of `pidof` being unreliable with finding the right process even the script is actually running. –  Nov 06 '20 at 06:41
0

In your script:

# store the Process ID here
pidfile=/var/run/mypid.local
# If there's a stored PID, terminate with SIGKILL
[[ -f $pidfile ]] && kill -s 9 $(cat $pidfile)
# Save our PID for the next run to kill
echo "$$" >$pidfile

and, to clean up, just before you exit,

rm $pidfile
waltinator
  • 4,439
  • 1
  • 16
  • 21