fix: price_monitor自愈式单例守卫(进程堆积事故)

- cron每2分钟拉起+外网抖动挂起+timeout仅SIGTERM杀不动 → 5并发烧155%CPU
- 新鲜实例在跑则退出(防堆积); 卡死>300s则SIGKILL接管(自愈)
- 符合进程生命周期铁律(防重复启动+卡死检测+自动清理)

deploy/profile-scripts/price_monitor.py

@@ -818,8 +818,46 @@ def run_once(round_label=""):
         pass
 
 
+def _singleton_guard(max_age_sec=300):
+    """自愈式单例守卫（2026-08-05 进程堆积事故修复）
+    背景：cron每2分钟拉起本脚本，外网抖动时实例挂起不退，timeout(仅SIGTERM)杀不动，
+    5个并发烧~155% CPU。规则：
+    - 有其他"新鲜"实例（存活<max_age）在跑 → 本实例立即退出（防堆积）
+    - 有"卡死"实例（存活>max_age）→ SIGKILL 接管（自愈）
+    """
+    import subprocess as _sp, os as _os, sys as _sys
+    my_pid = _os.getpid()
+    try:
+        out = _sp.run(["ps", "-C", "python3", "-o", "pid,etimes,cmd"],
+                      capture_output=True, text=True, timeout=10).stdout
+        for line in out.splitlines():
+            if "price_monitor.py" not in line:
+                continue
+            parts = line.split(None, 2)
+            if len(parts) < 3:
+                continue
+            try:
+                pid = int(parts[0]); age = int(parts[1])
+            except ValueError:
+                continue
+            if pid == my_pid:
+                continue
+            if age > max_age_sec:
+                try:
+                    _os.kill(pid, 9)
+                    print(f"[guard] SIGKILL卡死实例 pid={pid} age={age}s", flush=True)
+                except ProcessLookupError:
+                    pass
+            else:
+                print(f"[guard] 已有新鲜实例 pid={pid} age={age}s 在跑, 本实例退出", flush=True)
+                _sys.exit(0)
+    except Exception as _e:
+        print(f"[guard] 守卫异常(放行): {_e}", flush=True)
+
+
 def main():
     """每cron触发跑一轮"""
+    _singleton_guard(max_age_sec=300)
     run_once()