Python pika消费Rabbit MQ数据，慢消费引起的connection reset问题

曾彪彪2025-02-06 13:50

问题描述

使用python pika框架，从Rabbit MQ消费数据时，遇到了connection reset的错误，错误内容如下：

python 复制代码 
Traceback (most recent call last):
  File "/app/utils/rabbit.py", line 27, in message_callback
    channel.basic_ack(delivery_tag=method.delivery_tag)
  File "/usr/local/lib/python3.11/site-packages/pika/adapters/blocking_connection.py", line 2130, in basic_ack
    self._flush_output()
  File "/usr/local/lib/python3.11/site-packages/pika/adapters/blocking_connection.py", line 1353, in _flush_output
    self._connection._flush_output(lambda: self.is_closed, *waiters)
  File "/usr/local/lib/python3.11/site-packages/pika/adapters/blocking_connection.py", line 523, in _flush_output
    raise self._closed_result.value.error
pika.exceptions.StreamLostError: Stream connection lost: ConnectionResetError(104, 'Connection reset by peer')
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
  File "/app/app.py", line 14, in main
    rabbit.start_listen()
  File "/app/utils/rabbit.py", line 52, in start_listen
    channel.start_consuming()
  File "/usr/local/lib/python3.11/site-packages/pika/adapters/blocking_connection.py", line 1883, in start_consuming
    self._process_data_events(time_limit=None)
  File "/usr/local/lib/python3.11/site-packages/pika/adapters/blocking_connection.py", line 2044, in _process_data_events
    self.connection.process_data_events(time_limit=time_limit)
  File "/usr/local/lib/python3.11/site-packages/pika/adapters/blocking_connection.py", line 851, in process_data_events
    self._dispatch_channel_events()
  File "/usr/local/lib/python3.11/site-packages/pika/adapters/blocking_connection.py", line 567, in _dispatch_channel_events
    impl_channel._get_cookie()._dispatch_events()
  File "/usr/local/lib/python3.11/site-packages/pika/adapters/blocking_connection.py", line 1510, in _dispatch_events
    consumer_info.on_message_callback(self, evt.method,
  File "/app/utils/rabbit.py", line 34, in message_callback
    channel.basic_nack(delivery_tag=method.delivery_tag)
  File "/usr/local/lib/python3.11/site-packages/pika/adapters/blocking_connection.py", line 2151, in basic_nack
    self._impl.basic_nack(
  File "/usr/local/lib/python3.11/site-packages/pika/channel.py", line 401, in basic_nack
    self._raise_if_not_open()
  File "/usr/local/lib/python3.11/site-packages/pika/channel.py", line 1403, in _raise_if_not_open
    raise exceptions.ChannelWrongStateError('Channel is closed.')
pika.exceptions.ChannelWrongStateError: Channel is closed.

我的代码如下：

python 复制代码 
import pika
from utils.logger import logger
import traceback
import json
from utils.config import get_rabbit_config
from utils.mail import send_mail
import time
EXCHANGE = 'DEFAULT_EXCHANGE'
ROUTING_KEY = 'purchasing.system.contract'
QUEUE_NAME = 'PURCHASING_CONTRACT_CONSUMER'
class Rabbit():
    def __init__(self, callback=None) -> None:
        self.callback = callback
    def message_callback(self, channel, method, properties, body):
        try:
            logger.info(f'receive message: {body}')
            message = json.loads(body)
            if self.callback:
                result = self.callback(message)
                # time.sleep(5)
                # result = False
            else:
                logger.warn('self.callback is None')
            if result:
                channel.basic_ack(delivery_tag=method.delivery_tag)
            else:
                channel.basic_nack(delivery_tag=method.delivery_tag)
        except Exception as e:
            logger.error(traceback.format_exc())
            send_mail('purchasing-contract-consumer error',
                      traceback.format_exc())
            channel.basic_nack(delivery_tag=method.delivery_tag)
            time.sleep(3)
    def start_listen(self):
        config = get_rabbit_config()
        credentials = pika.PlainCredentials(
            username=config['username'], password=config['password'])
        parameters = pika.ConnectionParameters(
            # host=config['host'], port=config['port'], credentials=credentials, heartbeat=300, stack_timeout=300, socket_timeout=300, blocked_connection_timeout=300)
            host=config['host'], port=config['port'], credentials=credentials, heartbeat=120, socket_timeout=60)
        connection = pika.BlockingConnection(parameters=parameters)
        channel = connection.channel()
        # channel.exchange_declare(exchange=EXCHANGE,exchange_type='topic',durable=True)
        channel.queue_declare(queue=QUEUE_NAME, durable=True)
        channel.queue_bind(queue=QUEUE_NAME, exchange=EXCHANGE,
                           routing_key=ROUTING_KEY)
        channel.basic_consume(
            queue=QUEUE_NAME, on_message_callback=self.message_callback, auto_ack=False)
        logger.info('start listening...')
        channel.start_consuming()
        logger.warning('warning...')
        logger.warning('warning...')
        logger.warning('warning...')
        logger.warning('warning...')
        logger.warning('warning...')
        logger.warning('warning...')

我的应用场景是：从rabbit 消费数据，并通过restful API调用web服务(self.callback是一个web API调用)，上报消费的数据。但是web服务响应非常慢，每次调用花费1min多钟，并且rabbit MQ中积累了大量消息。

原因分析

刚开始以为是rabbit MQ connection的超时问题引起的，于是设置连接超时时间为300s，代码如下：

host=config['host'], port=config['port'], credentials=credentials, heartbeat=300, stack_timeout=300, socket_timeout=300, blocked_connection_timeout=300)

connection reset问题依然存在，于是又怀疑是result = self.callback(message)调用时间过长，超过了1min，于是将其注释掉，改成

python 复制代码 
result = self.callback(message)
time.sleep(5)
result = False

错误依然存在，也排除了不是self.callback(message)长时间阻塞引起的问题。

查阅相关资料，发现是因为未设置prefetch_count，如果不设置这个参数，框架默认设置为0，意味着无限消费数据，此时，如果消费端处理消息的速度非常慢，并且Rabbit MQ中有大量消息堆积，那么socket的缓存区就会塞满，此时客户端socket就会告诉服务端socket，将滑动窗口设置为0。由于客户端socket的缓存区一直被沾满，服务端长时间无法发送数据，甚至连socket的心跳消息也无法发出，就会导致connection reset异常；之后如果客户端调用restful API完成数据上传，尝试调用channel.basic_ack(delivery_tag=method.delivery_tag)，此时，连接已经被重置，那么就抛出connection reset异常。

解决方案

在代码中增加prefetch_count，代码如下：

python 复制代码 
import pika
from utils.logger import logger
import traceback
import json
from utils.config import get_rabbit_config
from utils.mail import send_mail
import time
EXCHANGE = 'DEFAULT_EXCHANGE'
ROUTING_KEY = 'purchasing.system.contract'
QUEUE_NAME = 'PURCHASING_CONTRACT_CONSUMER'
class Rabbit():
    def __init__(self, callback=None) -> None:
        self.callback = callback
    def message_callback(self, channel, method, properties, body):
        try:
            logger.info(f'receive message: {body}')
            message = json.loads(body)
            if self.callback:
                result = self.callback(message)
                # time.sleep(5)
                # result = False
            else:
                logger.warn('self.callback is None')
            if result:
                channel.basic_ack(delivery_tag=method.delivery_tag)
            else:
                channel.basic_nack(delivery_tag=method.delivery_tag)
        except Exception as e:
            logger.error(traceback.format_exc())
            send_mail('purchasing-contract-consumer error',
                      traceback.format_exc())
            channel.basic_nack(delivery_tag=method.delivery_tag)
            time.sleep(3)
    def start_listen(self):
        config = get_rabbit_config()
        credentials = pika.PlainCredentials(
            username=config['username'], password=config['password'])
        parameters = pika.ConnectionParameters(
            # host=config['host'], port=config['port'], credentials=credentials, heartbeat=300, stack_timeout=300, socket_timeout=300, blocked_connection_timeout=300)
            host=config['host'], port=config['port'], credentials=credentials, heartbeat=120, socket_timeout=60)
        connection = pika.BlockingConnection(parameters=parameters)
        channel = connection.channel()
        channel.basic_qos(prefetch_count=1) # 限制消费端消费的数据，防止缓存区被占满。
        # channel.exchange_declare(exchange=EXCHANGE,exchange_type='topic',durable=True)
        channel.queue_declare(queue=QUEUE_NAME, durable=True)
        channel.queue_bind(queue=QUEUE_NAME, exchange=EXCHANGE,
                           routing_key=ROUTING_KEY)
        channel.basic_consume(
            queue=QUEUE_NAME, on_message_callback=self.message_callback, auto_ack=False)
        logger.info('start listening...')
        channel.start_consuming()
        logger.warning('warning...')
        logger.warning('warning...')
        logger.warning('warning...')
        logger.warning('warning...')
        logger.warning('warning...')
        logger.warning('warning...')

设置完成后，可以正常消费数据，connection reset的异常消息。

所有文章在Github上同步，你也可以访问曾彪彪的个人博客点击查看

相关推荐
酷飞飞4 小时前
Python网络与多任务编程：TCP/UDP实战指南
网络·python·tcp/ip
数字化顾问5 小时前
Python：OpenCV 教程——从传统视觉到深度学习：YOLOv8 与 OpenCV DNN 模块协同实现工业缺陷检测
python
学生信的大叔6 小时前
【Python自动化】Ubuntu24.04配置Selenium并测试
python·selenium·自动化
诗句藏于尽头7 小时前
Django模型与数据库表映射的两种方式
数据库·python·django
智数研析社8 小时前
9120 部 TMDb 高分电影数据集 | 7 列全维度指标 (评分 / 热度 / 剧情)+API 权威源 | 电影趋势分析 / 推荐系统 / NLP 建模用
大数据·人工智能·python·深度学习·数据分析·数据集·数据清洗
扯淡的闲人8 小时前
多语言编码Agent解决方案（5）-IntelliJ插件实现
开发语言·python
moxiaoran57538 小时前
Flask学习笔记(一)
后端·python·flask
秋氘渔8 小时前
迭代器和生成器的区别与联系
python·迭代器·生成器·可迭代对象
Gu_shiwww9 小时前
数据结构8——双向链表
c语言·数据结构·python·链表·小白初步
Dxy123931021610 小时前
python把文件从一个文件复制到另一个文件夹
开发语言·python
热门推荐
01GitHub 镜像站点02UV 工具安装与国内镜像源配置指南03UV安装并设置国内源04A股预测还能更准？开源大模型Kronos带你跑通预测+回测全流程05KGG转MP3工具|非KGM文件|解密音频0646个Nano-banana 精选提示词，持续更新中07conda中设置镜像地址（附所有可换的地址）08突破百度网盘的下载限速，两种方法教会你【超详细】09Spec-Kit 使用指南10保姆级教程：手把手教你用Dify实现完美多轮对话（附Chatflow和提示词）