为什么 python 带有线程库的脚本吃掉 100% CPU？

Question

我有一些客户端脚本，连接套接字并等待数据。在服务器端有带有 json 包装器的 ZeroMQ 库。

所以，实际上，它是一个 json 数据格式的套接字连接。它是来自 metatrader 的股市数据。所以我每分钟等待一条数据线。因为我的时间表是 M1.

问题是为什么这个脚本吃100%cpu？

大部分时间它应该等待数据而不做任何事情。我认为脚本中可能存在一些错误。但是我还没有threding技能。请告诉我应该在哪里寻找问题。

这是一个脚本：



    import zmq
    import threading
    class MTraderAPI:
        def __init__(self, host=None):
            self.HOST = host or 'localhost'
            self.SYS_PORT = 15555       # REP/REQ port
            self.DATA_PORT = 15556      # PUSH/PULL port
            self.LIVE_PORT = 15557      # PUSH/PULL port
            self.EVENTS_PORT = 15558    # PUSH/PULL port
    
            # ZeroMQ timeout in seconds
            sys_timeout = 1
            data_timeout = 10
    
            # initialise ZMQ context
            context = zmq.Context()
    
            # connect to server sockets
            try:
                self.sys_socket = context.socket(zmq.REQ)
                # set port timeout
                self.sys_socket.RCVTIMEO = sys_timeout * 1000
                self.sys_socket.connect('tcp://{}:{}'.format(self.HOST, self.SYS_PORT))
    
                self.data_socket = context.socket(zmq.PULL)
                # set port timeout
                self.data_socket.RCVTIMEO = data_timeout * 1000
                self.data_socket.connect('tcp://{}:{}'.format(self.HOST, self.DATA_PORT))
            except zmq.ZMQError:
                raise zmq.ZMQBindError("Binding ports ERROR")
    
        def _send_request(self, data: dict) -> None:
            """Send request to server via ZeroMQ System socket"""
            try:
                self.sys_socket.send_json(data)
                msg = self.sys_socket.recv_string()
                # terminal received the request
                assert msg == 'OK', 'Something wrong on server side'
            except AssertionError as err:
                raise zmq.NotDone(err)
            except zmq.ZMQError:
                raise zmq.NotDone("Sending request ERROR")
    
        def _pull_reply(self):
            """Get reply from server via Data socket with timeout"""
            try:
                msg = self.data_socket.recv_json()
            except zmq.ZMQError:
                raise zmq.NotDone('Data socket timeout ERROR')
            return msg
    
        def live_socket(self, context=None):
            """Connect to socket in a ZMQ context"""
            try:
                context = context or zmq.Context.instance()
                socket = context.socket(zmq.PULL)
                socket.connect('tcp://{}:{}'.format(self.HOST, self.LIVE_PORT))
            except zmq.ZMQError:
                raise zmq.ZMQBindError("Live port connection ERROR")
            return socket
    
        def streaming_socket(self, context=None):
            """Connect to socket in a ZMQ context"""
            try:
                context = context or zmq.Context.instance()
                socket = context.socket(zmq.PULL)
                socket.connect('tcp://{}:{}'.format(self.HOST, self.EVENTS_PORT))
            except zmq.ZMQError:
                raise zmq.ZMQBindError("Data port connection ERROR")
            return socket
    
        def construct_and_send(self, **kwargs) -> dict:
            """Construct a request dictionary from default and send it to server"""
    
            # default dictionary
            request = {
                "action": None,
                "actionType": None,
                "symbol": None,
                "chartTF": None,
                "fromDate": None,
                "toDate": None,
                "id": None,
                "magic": None,
                "volume": None,
                "price": None,
                "stoploss": None,
                "takeprofit": None,
                "expiration": None,
                "deviation": None,
                "comment": None
            }
    
            # update dict values if exist
            for key, value in kwargs.items():
                if key in request:
                    request[key] = value
                else:
                    raise KeyError('Unknown key in **kwargs ERROR')
    
            # send dict to server
            self._send_request(request)
    
            # return server reply
            return self._pull_reply()
    
    api = MTraderAPI()
    print(api.construct_and_send(action="CONFIG", symbol="XBRUSD.c", chartTF="TICK"))
    
    def _t_livedata():
        socket = api.live_socket()
        while True:
            try:
                last_candle = socket.recv_json()
            except zmq.ZMQError:
                raise zmq.NotDone("Live data ERROR")
            print(last_candle)
    
    
    def _t_streaming_events():
        socket = api.streaming_socket()
        while True:
            try:
                trans = socket.recv_json()
                request, reply = trans.values()
            except zmq.ZMQError:
                raise zmq.NotDone("Streaming data ERROR")
            print(request)
            print(reply)
    
    
    
    t = threading.Thread(target=_t_livedata, daemon=True)
    t.start()
    
    t = threading.Thread(target=_t_streaming_events, daemon=True)
    t.start()
    
    while True:
        pass

Answer 1

是你的：

while True:
    pass

最后 - 重点是消耗 CPU。使用 time.sleep 添加一个小睡眠，CPU 的使用将减少。或者如果你只是等待线程完成，就加入它们，它会阻塞直到它们执行完成。

例如：

from time import sleep

while True:
    sleep(0.1)

或者如果您将线程对象收集在一个列表中，您应该：

for t in threads:
    t.join()

有关详细信息，请参阅。

为什么 python 带有线程库的脚本吃掉 100% CPU？

Why python script with threading lib eats 100% CPU?

python

json

multithreading

zeromq