kafka神坑

深藏阁楼爱情的钟 2022-10-31 13:34 88阅读 0赞

## Q1:生产者连接工厂，发送信息失败 ##

**原因：pom依赖有问题**

只有生产者时，依赖用：

<dependency>
        <groupId>org.apache.kafka</groupId>
        <artifactId>kafka-clients</artifactId>
        <version>0.9.0.0</version>
    </dependency>

生产者+消费者时，依赖用：

<dependency>
    	 <groupId>org.springframework.kafka</groupId>
    	 <artifactId>spring-kafka</artifactId>
    </dependency>

**排查方法：**  
1、利用kafkaTool查看，发现发送者的topic并未创建成功。  
因此，将发送者的topic修改到一个已有的topic，但是kafkaTool仍然收到发送的message  
2、将发送者用

kafkaProducer.send(new ProducerRecord<String, String>("test.topic", message)).get()

注意这个**get**！！！可以同步返回发送结果，然后我们发现生产者并未发送成功，显示连接超时，因此断定生产者连接有问题！！

[添加链接描述][Link 1]

## Q2 生产者实例 ##

//创建生产者
    public Producer<String, String> ProducerConfigs() { 
            Properties props = new Properties();
            props.put(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, "1.1.1.1:9029");
            props.put(ProducerConfig.ACKS_CONFIG, kafkaProducerAcks);
            props.put(ProducerConfig.RETRIES_CONFIG, kafkaProducerRetries);
            props.put(ProducerConfig.LINGER_MS_CONFIG, kafkaProducerLingerMs);
            props.put(ProducerConfig.BATCH_SIZE_CONFIG, kafkaProducerBatchSize);
            props.put(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG, StringSerializer.class);
            props.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, StringSerializer.class);
            KafkaProducer<String, String> producer = new KafkaProducer<>(props);
            this.kafkaProducer = producer;
            return producer;
        }
        
    //发送信息
     public void sendKafkaMessage(String message) { 
            if (kafkaProducer == null) { 
                ProducerConfigs();
            }
            kafkaProducer.send(new ProducerRecord<String, String>("test.topic", message));
        }

## Q3 消费者实例 ##

@Bean("defaultConsumerFactory")
        public DefaultKafkaConsumerFactory defaultConsumerFactory() { 
            try { 
                return new DefaultKafkaConsumerFactory(consumerProperties());
            } catch (BusinessException e) { 
                e.printStackTrace();
            } 
        }
    
        @Bean("batchListenerContainerFactory")
        public ConcurrentKafkaListenerContainerFactory BatchListenerContainerFactory(DefaultKafkaConsumerFactory
                                                                                                 consumerFactory) { 
            ConcurrentKafkaListenerContainerFactory factory = new ConcurrentKafkaListenerContainerFactory();
            //指定使用DefaultKafkaConsumerFactory
            factory.setConsumerFactory(consumerFactory);
            // 连接池中消费者数量
            factory.setConcurrency(evplanKafkaListenerConcurrency);
            // 开启并发消费
            factory.setBatchListener(Boolean.TRUE);
            HIK_LOG.info("Kafka batchListenerContainerFactory build successfully!!");
            return factory;
        }
    
        /** * 构造消费者属性map * * @return */
        private Map<String, Object> consumerProperties() { 
            Map<String, Object> props = new HashMap<>();
            //groupID:一个字符串用来指示一组consumer所在的组。相同的groupID表示在一个组里。相同的groupID消费记录offset时，记录的是同一个offset
            props.put(ConsumerConfig.GROUP_ID_CONFIG, KafkaConsumerGroupId);
            props.put(ConsumerConfig.ENABLE_AUTO_COMMIT_CONFIG, KafkaConsumerEnableAutoCommit);
            props.put(ConsumerConfig.AUTO_OFFSET_RESET_CONFIG, KafkaConsumerAutoOffsetReset);
            props.put(ConsumerConfig.AUTO_COMMIT_INTERVAL_MS_CONFIG, KafkaConsumerAutoCommitInterval);
            props.put(ConsumerConfig.SESSION_TIMEOUT_MS_CONFIG, KafkaConsumerSessionTimeoutMsConfig);
            props.put(ConsumerConfig.HEARTBEAT_INTERVAL_MS_CONFIG, KafkaConsumerHeartbeatIntervalMsConfig);
            props.put(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class);
            props.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class);
            //分区分配策略
            props.put(ConsumerConfig.PARTITION_ASSIGNMENT_STRATEGY_CONFIG, RoundRobinAssignor.class.getName());
            props.put(ConsumerConfig.MAX_POLL_RECORDS_CONFIG, KafkaConsumerMaxPollRecordsConfig);
            props.put(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, "1.1.1.1:9029");
            return props;
        }
    
    //消费者监听
     @KafkaListener(topics = "test.topic", containerFactory = "batchListenerContainerFactory")
        public void receiveCCTVSubSysQueue(List<ConsumerRecord<?, String>> records){ 
          
            records.forEach(message->{ 
                JSONObject jsonObject = JSONObject.parseObject(message.value());
                //CCTV子系统事件信息需要单独进行转换
                CCTVAlarmInfoDTO cctvAlarmInfoDTO = JSONObject.toJavaObject(jsonObject, CCTVAlarmInfoDTO.class);
                SubEventInfoDTO subEventInfoDTO = cctvAlarmInfoDTO2subEventInfoDTO(cctvAlarmInfoDTO);
                if (null != subEventInfoDTO){ 
                    kafkaMessageProcess(subEventInfoDTO);
                }
            });
        }

注意：  
由于消费者配置中开启了并发消费，因此在接收的时候，用`List<ConsumerRecord<?, String>> records`  
若没有开启并发消费，则接收直接用`ConsumerRecord<?, String> records`

[Link 1]: https://my.oschina.net/u/3247419/blog/1189241