Processes not terminating correctly (or at all?)

I've started using concurrently quite heavily.

For example, I might have tsc, webpack and node-sass all running with --watch during development.

I started noticing weird behavior, including on several occasions, my source files randomly getting deleted when I run git commands! Fortunately these are usually under source-control, so I can get them back with a git reset, but it's quite confusing.

The problem, it seems, is when I try to shut down concurrently, I can sometimes wait for several minutes and nothing happens - often this happens after e.g. webpack has crashed or stalled, which is tends to do if I remove one of the source-files it's watching.

So eventually I press CTRL+C again, which terminates concurrently, but appears to leave running processes hanging in the background.

Here's some terminal output:

$ ps -a
  PID TTY          TIME CMD
 1541 tty1     00:00:00 node
 1685 tty1     00:00:26 node-sass
 1723 tty1     00:00:00 node
 1867 tty1     00:00:06 node-sass
 5220 tty2     00:00:00 sudo
 5221 tty2     00:00:03 caddy
 5504 tty1     00:00:00 node
 5648 tty1     00:02:16 node-sass
 6041 tty1     00:00:00 ps
$ kill 1685
$ ps -a
  PID TTY          TIME CMD
 1541 tty1     00:00:00 node
 1685 tty1     00:00:26 node-sass
 1723 tty1     00:00:00 node
 1867 tty1     00:00:06 node-sass
 5220 tty2     00:00:00 sudo
 5221 tty2     00:00:03 caddy
 5504 tty1     00:00:00 node
 5648 tty1     00:02:16 node-sass
 6042 tty1     00:00:00 ps
$ kill -9 1685
$ [1] npm run watch-sass exited with code null
^C
$ ps -a
  PID TTY          TIME CMD
 1723 tty1     00:00:00 node
 1867 tty1     00:00:06 node-sass
 5220 tty2     00:00:00 sudo
 5221 tty2     00:00:03 caddy
 5504 tty1     00:00:00 node
 5648 tty1     00:02:16 node-sass
 6043 tty1     00:00:00 ps

As you can see, my first attempt to kill 1685 fails - I'm betting this is why concurrently hangs as well, it isn't able to shut down the hanging node-sass process the normal way.

Forcing node-sass to stop with kill -9 1685 also apparently causes the hanging node process (which is running concurrently) to report that node-sass has finally stopped - and as you can see from the last ps -a, both the node process that was running concurrently, and the node-sass process it was running, have finally stopped.

Is it possible to set a timeout for concurrently, e.g. forcing child processes to stop after a set maximum period of waiting? I didn't see a command-line option for that.

Any idea how to resolve this?

Start by reproducing the shutdown behavior with the watch commands described in the issue, then trace how concurrently handles CTRL+C and child-process termination. Done means a documented or implemented timeout behavior that stops hanging child processes and leaves no background processes after shutdown.