From e9673cad8810052ecbb6d5e04842d9aae78404e7 Mon Sep 17 00:00:00 2001
From: weichen <weichen@home.local>
Date: Mon, 12 Oct 2026 14:56:00 +0000
Subject: [PATCH] Wait for port release after stopping a service

Why: on slow runners the orphaned http child of a stopped stub lingers
a few hundred milliseconds past the parent's exit (ppid-watch suicide),
and the immediately following start - notably the U2 rollback restart -
classified the unparented listener as a foreign occupant and refused to
start (CI run 3774).
Evidence: test_svc_u2/identity/svc suites 82 OK locally with the wait
in both stop paths.
Impact: adds <=3s bounded wait in do_stop/do_stop_quiet after SIGTERM
cleanup; no behavior change when the port is already free.
---
 scripts/svc.sh | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/scripts/svc.sh b/scripts/svc.sh
index 34c2a4dd..af9ff500 100755
--- a/scripts/svc.sh
+++ b/scripts/svc.sh
@@ -3999,6 +3999,7 @@ do_stop() {
       proc_stop "$pid"
       rm -f "$ENV_PID"
       svc_clear_autostart
+      svc_wait_port_released
       c_ok "[$ENV_NAME] 已停止"
       hist_event stop ok "$pid" "SIGTERM"
       ;;
@@ -4006,6 +4007,20 @@ do_stop() {
   return 0
 }
 
+
+# svc_wait_port_released：停机后等子进程释放端口（最多 3s）。桩/组合产物的 http
+# 子进程经 ppid-watch 自杀（≤0.5s 语义），慢机上略晚；不等会让紧跟的 start 把
+# 已无父的孤儿占口判成「非本服务」拒启（U2 回退实测，CI 慢机 run 3774）。
+svc_wait_port_released() {
+  local port="${ENV_ADDR##*:}" i=0
+  [ -n "$port" ] || return 0
+  while [ "$i" -lt 30 ] && port_listening "$port"; do
+    sleep 0.1
+    i=$((i + 1))
+  done
+  return 0
+}
+
 # do_stop_quiet：deploy 前置——未运行时静默返回（同时清除期望态：
 # deploy 尾端的 start 成功会重新置位；失败则保持停止等人工介入）。
 do_stop_quiet() {
@@ -4028,6 +4043,7 @@ do_stop_quiet() {
       proc_stop "$pid"
       rm -f "$ENV_PID"
       svc_clear_autostart
+      svc_wait_port_released
       c_ok "[$ENV_NAME] 已停止"
       hist_event stop ok "$pid" "SIGTERM"
       ;;