GPT on Question 14, Oct 3

A Bash process cannot reliably clean up children after it has died. The usual pattern is to track child PIDs and kill/reap them from an EXIT trap:

#!/usr/bin/env bash

children=()

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

    if ((${#children[@]})); then
        kill -TERM "${children[@]}" 2>/dev/null || true
        wait "${children[@]}" 2>/dev/null || true
    fi

    exit "$status"
}

trap cleanup EXIT
trap 'exit 129' HUP
trap 'exit 130' INT
trap 'exit 143' TERM

some_command &
children+=("$!")

another_command &
children+=("$!")

wait

Important limitations:

• •This handles normal exit and catchable signals, but not SIGKILL, power loss, or a hard crash.
• •It kills only the recorded direct children. If they spawn descendants, use a dedicated process group/session and terminate the group:

setsid some_command &
group_leader=$!

cleanup() {
    status=$?
    trap - EXIT INT TERM HUP
    kill -TERM -- "-$group_leader" 2>/dev/null || true
    wait "$group_leader" 2>/dev/null || true
    exit "$status"
}

Avoid blindly using kill 0; it can kill unrelated processes in the script’s process group.

For guaranteed cleanup of an entire process tree, including when the script gets SIGKILL, use an external supervisor such as systemd with a dedicated cgroup (KillMode=control-group), a container runtime, or—on Linux—a small wrapper using prctl(PR_SET_PDEATHSIG, ...).