GPT on Question 14, Sep 19

The idiomatic Bash approach is to run the work in a dedicated process group/session and kill that group from a trap:

#!/usr/bin/env bash
set -Eeuo pipefail

worker=

cleanup() {
    local status=$?
    trap - EXIT INT TERM HUP

    if [[ -n ${worker:-} ]]; then
        # Negative PID means “the process group whose PGID is $worker”.
        kill -TERM -- "-$worker" 2>/dev/null || true

        # Optional grace period, then force remaining descendants down.
        sleep 2
        kill -KILL -- "-$worker" 2>/dev/null || true

        wait "$worker" 2>/dev/null || true
    fi

    exit "$status"
}
trap cleanup EXIT INT TERM HUP

# setsid makes this command the leader of a new session/process group.
setsid bash -c '
    child_one &
    child_two &
    wait
' &
worker=$!

wait "$worker"

Why a process group rather than pkill -P "$$":

• •pkill -P only finds direct children.
• •Grandchildren can be reparented as soon as an intermediate parent exits.
• •Signaling -$worker targets all processes remaining in that process group.

Important limitation: traps do not run if the parent receives SIGKILL, crashes, or the machine goes down. If cleanup must also work in those cases, use an OS supervisor such as a systemd service/scope with KillMode=control-group. On Linux, a small wrapper using prctl(PR_SET_PDEATHSIG, ...) is another option, but that property must be set by each relevant child and has race/descendant considerations.