BleakScanner配合asyncio.wait_for时超时异常问题求助

环境信息 #

• bleak版本: 0.22.2
• Python版本: 3.8.18
• 操作系统: Linux
• BlueZ版本 (bluetoothctl -v): 5.64

问题描述 #

使用带超时设置的BleakScanner时，asyncio.wait_for函数返回后过20秒仍触发超时错误，后续扫描会被DBus的Operation already in progress错误阻断。

操作场景 #

网关以轮询方式持续连接4台同类型设备，运行下方代码后，尽管asyncio.wait_for(_scan_with_timeout(filter, device_id), timeout)已返回，但还是触发了asyncio.TimeoutError日志，后续扫描抛出BleakDBusError('org.bluez.Error.InProgress', 'Operation already in progress')。

代码实现：

async def scan(
    device_id, timeout=SCAN_TIMEOUT_S
) -> Optional[Tuple[BLEDevice, Set[str]]]:
    try:
        filter = get_device_filter(device_id)
        return await asyncio.wait_for(_scan_with_timeout(filter, device_id), timeout)
    except asyncio.TimeoutError:
        logger.info(
            f"Timeout reached. Device {device_id} not found after {timeout} seconds"
        )

    return None


async def _scan_with_timeout(filter, device_id):
    seen = set()
    async with BleakScanner() as scanner:
        async for device, advertisement in scanner.advertisement_data():
            found_device_id = get_device_id(device)
            if found_device_id and found_device_id.startswith(filter):
                logger.info(f"Found {device!r}")
                logger.debug(f"{advertisement!r}")
                logger.debug(f"platform_data: {advertisement.platform_data}")
                logger.debug(f"service_data: {advertisement.service_data}")
                seen.add(found_device_id)

            if found_device_id == device_id:
                logger.info(f"Connecting")
                return (device, seen)

关键日志 #

2024-08-13 15:25:10,427 | INFO | xtel | Connecting
2024-08-13 15:25:30,122 | INFO | xtel | Timeout reached. Device D0:83:D4:00:A2:D1 not found after 30 seconds

解决方案 #

问题根源 #

当_scan_with_timeout执行return时，async with BleakScanner()的上下文管理器未及时终止BlueZ端的扫描操作：

1. async for循环退出后，BleakScanner的后台扫描任务仍在运行，导致DBus上的扫描进程未终止；
2. asyncio.wait_for的超时定时器因事件循环调度延迟，在函数返回后才触发回调；
3. 未终止的扫描会导致后续启动新扫描时，BlueZ抛出Operation already in progress错误。

修复步骤 #

在return前手动调用scanner.stop()，确保扫描操作立即终止，再退出上下文管理器：

修改后的_scan_with_timeout函数：

async def _scan_with_timeout(filter, device_id):
    seen = set()
    async with BleakScanner() as scanner:
        async for device, advertisement in scanner.advertisement_data():
            found_device_id = get_device_id(device)
            if found_device_id and found_device_id.startswith(filter):
                logger.info(f"Found {device!r}")
                logger.debug(f"{advertisement!r}")
                logger.debug(f"platform_data: {advertisement.platform_data}")
                logger.debug(f"service_data: {advertisement.service_data}")
                seen.add(found_device_id)

            if found_device_id == device_id:
                logger.info(f"Connecting")
                # 手动停止扫描，确保BlueZ端的扫描进程终止
                await scanner.stop()
                return (device, seen)

额外建议 #

1. 可在scan函数的finally块中添加扫描状态检查，确保每次扫描后资源都被正确释放；
2. 轮询场景下，建议在两次扫描之间添加短暂延迟，避免频繁操作导致BlueZ资源占用异常。

内容的提问来源于stack exchange，提问作者icc97