Weird hanging bug with --kill-others

Hi,

I've found a superweird bug that I'd like some help to figure out what to do with.

When running concurrently in my node-alpine docker container, PIDs are (consistently) one off.

To test this, I modified concurrently to print the PID first when spawning a child process, and also when trying to kill them in killOtherProcesses. http-server is constantly one higher than the saved PID.

Example output:

bash-4.3# ps aux
PID   USER     TIME   COMMAND
    1 root       0:00 bash
 1293 root       0:00 ps aux

bash-4.3# concurrently --kill-others "http-server ." "npm --version" &
[1] 1294
1300
1307
[0] Starting up http-server, serving .
[0] Available on:
[0]   http://127.0.0.1:8080
[0]   http://172.17.0.3:8080
[0] Hit CTRL-C to stop the server
[1] 4.4.4
[1] npm --version exited with code 0
--> Sending SIGTERM to other processes..
1300
bash-4.3#  ps
PID   USER     TIME   COMMAND
    1 root       0:00 bash
 1294 root       0:00 node /usr/bin/concurrently --kill-others http-server . npm --version
 1301 root       0:00 node /usr/bin/http-server .
 1315 root       0:00 ps

So it's trying to kill http-server on 1300, but http-server is actually on 1301. If I change the code to kill child.pid+1 instead, all is fine...

Any ideas? I'm stumped for ideas where to look further.

Start by reproducing the provided concurrently --kill-others "http-server ." "npm --version" command in the node-alpine container and inspect the implementation around killOtherProcesses. Compare the spawned and terminated process IDs, then verify that the correct process receives SIGTERM without changing behavior outside this reproduction.