Want to kill the script B, which is initially started from script A, when using CTRL + C to kill script A.

Hongyi Zhao

unread,

Feb 18, 2016, 7:38:24 PM2/18/16

to

Hi all,

I have two shell bash scripts, say, A and B.

The script A is as follows:

---------&<---------
#!/bin/bash
# Here is the content of script-A

bla bla

./script-B &

bla bla
---------&<---------

When script A is running, I using CTRL + C to kill script A. But I find
that if the script B is running when I issue ``CTRL + C'', it will still
leave in the running status after the script A is killed.

So, is it possible to ensure the script B also be killed when using CTRL
+ C to kill the script A.

Regards
--
.: Hongyi Zhao [ hongyi.zhao AT gmail.com ] Free as in Freedom :.

Aragorn

unread,

Feb 18, 2016, 9:05:46 PM2/18/16

to

On Friday 19 Feb 2016 01:35, Hongyi Zhao conveyed the following to
comp.unix.shell...

> Hi all,
>
> I have two shell bash scripts, say, A and B.
>
> The script A is as follows:
>
> ---------&<---------
> #!/bin/bash
> # Here is the content of script-A
>
> bla bla
>
> ./script-B &
>
> bla bla
> ---------&<---------
>
> When script A is running, I using CTRL + C to kill script A. But I
> find that if the script B is running when I issue ``CTRL + C'', it
> will still leave in the running status after the script A is killed.
>
> So, is it possible to ensure the script B also be killed when using
> CTRL + C to kill the script A.

You are starting script B in a subshell, and you are daemonizing it ─
i.e. you are shoving it into the background ─ so the answer to your
question is "No", or at least, "Not as you would want it." Ctrl+C only
terminates the active process in the current shell. It doesn't touch
anything running in the background.

Is there a reason why you want it to run in the background ─ e.g. you
want the variables of script B to exist in their own address space? If
not, then instead of invoking it via a subshell, you could source it
into the running script, i.e.

. ./script

... or...

source ./script

If you do need script B to run as a separate process ─ i.e. in the
background, or simply in a separate address space ─ then you're going to
have to create a PID file for it, and then parse it, kill the process
and clean up the PID file.

Also note that relying on Ctrl+C to terminate a script is bad coding
practice. You could insert far more elegant ways of terminating the
script.

--
= Aragorn =

http://www.linuxcounter.net - registrant #223157

Icarus Sparry

unread,

Feb 18, 2016, 9:36:18 PM2/18/16

to

On Fri, 19 Feb 2016 00:35:24 +0000, Hongyi Zhao wrote:

> Hi all,
>
> I have two shell bash scripts, say, A and B.
>
> The script A is as follows:
>
> ---------&<---------
> #!/bin/bash # Here is the content of script-A
>
> bla bla
>
> ./script-B &
>
> bla bla ---------&<---------
>
> When script A is running, I using CTRL + C to kill script A. But I find
> that if the script B is running when I issue ``CTRL + C'', it will still
> leave in the running status after the script A is killed.
>
> So, is it possible to ensure the script B also be killed when using CTRL
> + C to kill the script A.
>
> Regards

You can capture the process id of script-B, using

./script-B &
pid=$!

and then you can arrange to trap the control-C and kill the script-B

trap "kill -2 $pid" 2

As always reading the manual helps.

Hongyi Zhao

unread,

Feb 18, 2016, 11:43:41 PM2/18/16

to

On Fri, 19 Feb 2016 02:36:13 +0000, Icarus Sparry wrote:

> You can capture the process id of script-B, using
>
> ./script-B &
> pid=$!
>
> and then you can arrange to trap the control-C and kill the script-B
>
> trap "kill -2 $pid" 2

Tried this in script-A but failed:

------- script-A ------
#!/bin/bash

./script-B &
pid=$!

bla

trap "kill -2 $pid" 2

-----------------

Then using control-C to kill script-A, and the script-B still be running.

Regards

>
> As always reading the manual helps.

Rakesh Sharma

unread,

Feb 19, 2016, 1:00:26 AM2/19/16

to

On Friday, 19 February 2016 10:13:41 UTC+5:30, Hongyi Zhao wrote:

>
> ------- script-A ------
> #!/bin/bash
>
> ./script-B &
> pid=$!
>
> bla
>
> trap "kill -2 $pid" 2
> -----------------
>
> Then using control-C to kill script-A, and the script-B still be running.
>

You are not able to slay the script-B since there is no interrupt catching mechanism in place there. For unconditional killings, you have to bring in
the heavy artillery, (kill -9).

So either do this in case you are unwilling to change script-b:
trap 'kill -9 "$pid"' 2
or better yet, you could put in place an interrup-catching mechanism in script-b
by placing this line in there:

trap 'echo "Caught SIGINT, exiting...";exit 1' 2
and then you could retain the line
trap 'kill -2 "$pid"' 2 in script-A

Hongyi Zhao

unread,

Feb 19, 2016, 1:41:50 AM2/19/16

to

On Thu, 18 Feb 2016 22:00:11 -0800, Rakesh Sharma wrote:

> You are not able to slay the script-B since there is no interrupt
> catching mechanism in place there. For unconditional killings, you have
> to bring in the heavy artillery, (kill -9).

Really? But I've found the following method will also do the job:

script-A:
---
#!/bin/bash

control_c()
# run if user hits control-c
{
echo -en "\n*** Ouch! Exiting ***\n"
if [ "x"$pid != "x" ]; then
kill $pid
fi
exit $?
}

# trap keyboard interrupt (control-c) and call control_c()
# trap control_c SIGINT
#or
trap control_c INT

bla

./script-B &
pid=$!
---

>
> So either do this in case you are unwilling to change script-b:
> trap 'kill -9 "$pid"' 2
> or better yet, you could put in place an interrup-catching mechanism in
> script-b by placing this line in there:
>
> trap 'echo "Caught SIGINT, exiting...";exit 1' 2
> and then you could retain the line
> trap 'kill -2 "$pid"' 2 in script-A

Thanks a lot for your above methods.

Regards

Hongyi Zhao

unread,

Feb 19, 2016, 2:00:48 AM2/19/16

to

On Fri, 19 Feb 2016 06:38:49 +0000, Hongyi Zhao wrote:

> if [ "x"$pid != "x" ]; then

More elegant way:

if [ "x"$pid != "x" ] && kill -0 $pid ; then

Hongyi Zhao

unread,

Feb 19, 2016, 3:00:41 AM2/19/16

to

On Thu, 18 Feb 2016 22:00:11 -0800, Rakesh Sharma wrote:

[snip]

> So either do this in case you are unwilling to change script-b:
> trap 'kill -9 "$pid"' 2

I want to the script A exit too. So the above code should be changed as
follows:

trap 'kill -9 "$pid" ; exit ' 2

> or better yet, you could put in place an interrup-catching mechanism in
> script-b by placing this line in there:
>
> trap 'echo "Caught SIGINT, exiting...";exit 1' 2
> and then you could retain the line
> trap 'kill -2 "$pid"' 2 in script-A

Also, this should be changed into the following for my case:

trap 'kill -2 "$pid" ; exit ' 2

Regards

Hongyi Zhao

unread,

Feb 19, 2016, 3:02:01 AM2/19/16

to

On Fri, 19 Feb 2016 06:57:47 +0000, Hongyi Zhao wrote:

>> if [ "x"$pid != "x" ]; then
>
> More elegant way:
>
> if [ "x"$pid != "x" ] && kill -0 $pid ; then

Using the following more elegant way:

if [ "x"$pid != "x" ] && kill -0 $pid &> /dev/null ; then

Michael Paoli

unread,

Sep 7, 2016, 7:46:03 AM9/7/16

to

$ cat A
#!/bin/sh
./B &
B_PID="$!"
trap "kill -9 $B_PID $$" 2
wait
$ cat B
#!/bin/sh
while :
do
sleep 1
done
$ ./A &
[1] 10180
$ ps
PID TTY TIME CMD
10180 pts/19 00:00:00 A
10181 pts/19 00:00:00 B
10222 pts/19 00:00:00 sleep
10223 pts/19 00:00:00 ps
28249 pts/19 00:00:00 bash
$ kill -2 10180
$
[1]+ Killed ./A
$ ps
PID TTY TIME CMD
10275 pts/19 00:00:00 ps
28249 pts/19 00:00:00 bash
$