Harden local debug stale listener cleanup

openspec/changes/split-run-into-independent-repository/design.md

@@ -16,6 +16,8 @@ git clone git@git.npc0.com:admin343/run.git run
 
 If `RUN_REPO_DIR` is missing or does not contain a run `go.mod`, local debug scripts fail with a clear setup message.
 
+Local debug start/stop scripts treat pid files as hints, not the only source of truth. If a previous local debug process still owns the configured platform or web port after pid files are removed, the scripts identify it by cwd/log ownership under this repository and stop it before starting a new stack.
+
 ## Docker Compose
 
 The compose file keeps the run service for all-in-one local deployment, but its build context points at `${RUN_REPO_DIR:-./run}` and uses the external repository's `Dockerfile`.

openspec/changes/split-run-into-independent-repository/tasks.md

@@ -3,6 +3,7 @@
 - [x] Add OpenSpec proposal/design/spec for the repository split.
 - [x] Remove `run/` from browser-owned project roots and structure checks.
 - [x] Update README, local debug scripts, and Docker Compose to use `RUN_REPO_DIR`.
+- [x] Harden local debug start/stop against stale owned listeners when pid files are missing.
 - [x] Remove tracked `run/` source files from `browser.git`.
 - [x] Run `scripts/check-structure.sh`.
 - [x] Run `openspec validate split-run-into-independent-repository --strict`.

scripts/local-debug-start.sh

@@ -23,6 +23,24 @@ port_listener_pid() {
   lsof -tiTCP:"$port" -sTCP:LISTEN 2>/dev/null | head -n 1 || true
 }
 
+listener_owned_by_local_debug() {
+  local pid="$1"
+  local details
+  details="$(lsof -nP -p "$pid" 2>/dev/null || true)"
+  [[ "$details" == *"$ROOT_DIR"* || "$details" == *"$LOCAL_DEBUG_ROOT"* ]]
+}
+
+wait_for_pid_exit() {
+  local pid="$1"
+  for _ in $(seq 1 20); do
+    if ! kill -0 "$pid" 2>/dev/null; then
+      return 0
+    fi
+    sleep 0.25
+  done
+  return 1
+}
+
 ensure_port_available() {
   local name="$1"
   local port="$2"
@@ -35,6 +53,17 @@ ensure_port_available() {
   if managed_pid_running "$pid_file" && [[ "$listener_pid" == "$(cat "$pid_file")" ]]; then
     return 0
   fi
+  if listener_owned_by_local_debug "$listener_pid"; then
+    printf 'stopping stale %s listener pid %s on port %s\n' "$name" "$listener_pid" "$port"
+    kill "$listener_pid" 2>/dev/null || true
+    if ! wait_for_pid_exit "$listener_pid"; then
+      printf 'stale %s listener pid %s did not stop after SIGTERM; sending SIGKILL\n' "$name" "$listener_pid"
+      kill -9 "$listener_pid" 2>/dev/null || true
+      wait_for_pid_exit "$listener_pid" || true
+    fi
+    rm -f "$pid_file"
+    return 0
+  fi
   printf '%s port %s is already used by pid %s; choose another LOCAL_DEBUG_*_PORT or stop the existing listener\n' "$name" "$port" "$listener_pid" >&2
   return 1
 }

scripts/local-debug-stop.sh

@@ -7,20 +7,63 @@ source "$ROOT_DIR/scripts/local-debug-env.sh"
 
 if [[ ! -d "$LOCAL_DEBUG_PID_DIR" ]]; then
   printf 'no local debug pid directory at %s\n' "$LOCAL_DEBUG_PID_DIR"
-  exit 0
+else
+  for name in platform_web run platform; do
+    pid_file="$LOCAL_DEBUG_PID_DIR/$name.pid"
+    if [[ ! -f "$pid_file" ]]; then
+      continue
+    fi
+    pid="$(cat "$pid_file")"
+    if kill -0 "$pid" 2>/dev/null; then
+      printf 'stopping %s pid %s\n' "$name" "$pid"
+      kill "$pid" 2>/dev/null || true
+    else
+      printf '%s pid %s is not running\n' "$name" "$pid"
+    fi
+    rm -f "$pid_file"
+  done
 fi
 
-for name in platform_web run platform; do
+wait_for_pid_exit() {
-  pid_file="$LOCAL_DEBUG_PID_DIR/$name.pid"
+  local pid="$1"
-  if [[ ! -f "$pid_file" ]]; then
+  for _ in $(seq 1 20); do
-    continue
+    if ! kill -0 "$pid" 2>/dev/null; then
+      return 0
+    fi
+    sleep 0.25
+  done
+  return 1
+}
+
+listener_owned_by_local_debug() {
+  local pid="$1"
+  local details
+  details="$(lsof -nP -p "$pid" 2>/dev/null || true)"
+  [[ "$details" == *"$ROOT_DIR"* || "$details" == *"$LOCAL_DEBUG_ROOT"* ]]
+}
+
+stop_listener_on_port() {
+  local name="$1"
+  local port="$2"
+  local pid
+  pid="$(lsof -tiTCP:"$port" -sTCP:LISTEN 2>/dev/null | head -n 1 || true)"
+  if [[ -z "$pid" ]]; then
+    return 0
   fi
-  pid="$(cat "$pid_file")"
+
-  if kill -0 "$pid" 2>/dev/null; then
+  if listener_owned_by_local_debug "$pid"; then
-    printf 'stopping %s pid %s\n' "$name" "$pid"
+    printf 'stopping %s listener pid %s on port %s\n' "$name" "$pid" "$port"
     kill "$pid" 2>/dev/null || true
+    if ! wait_for_pid_exit "$pid"; then
+      printf '%s listener pid %s did not stop after SIGTERM; sending SIGKILL\n' "$name" "$pid"
+      kill -9 "$pid" 2>/dev/null || true
+      wait_for_pid_exit "$pid" || true
+    fi
+    return 0
   else
-    printf '%s pid %s is not running\n' "$name" "$pid"
+    printf '%s port %s is still used by pid %s outside local debug ownership\n' "$name" "$port" "$pid" >&2
   fi
-  rm -f "$pid_file"
+}
-done
+
+stop_listener_on_port platform "$LOCAL_DEBUG_PLATFORM_PORT"
+stop_listener_on_port platform_web "$LOCAL_DEBUG_WEB_PORT"