2022年2月23日
spring-kafka简单使用
1.启动kafka服务,或者使用spring-kafka-test来模拟启动
<dependency>
<groupId>org.springframework.kafka</groupId>
<artifactId>spring-kafka-test</artifactId>
<version>2.8.2</version>
<scope>compile</scope>
</dependency>
@RunWith(SpringRunner.class)
@SpringBootTest(classes = DemoApplicationTests.class)
@EmbeddedKafka(count = 4, ports = {9092,9093,9094,9095})
public class DemoApplicationTests {
@Test
public void contextLoads()throws IOException {
System.in.read();
}
}
2.可以先创建或者发送时候创建topic
@Configuration
public class KafkaInitialConfiguration {
// 创建一个名为testtopic的Topic并设置分区数为8,分区副本数为2
@Bean
public NewTopic initialTopic() {
return new NewTopic("topic1",8, (short) 2 );
}
// 如果要修改分区数,只需修改配置值重启项目即可
// 修改分区数并不会导致数据的丢失,但是分区数只能增大不能减小
@Bean
public NewTopic updateTopic() {
return new NewTopic("topic1",10, (short) 2 );
}
}
3.KafkaProducer
@RestController
public class KafkaProducer {
@Autowired
private KafkaTemplate<String, Object> kafkaTemplate;
// 发送消息
@GetMapping("/kafka/normal/{message}")
public void sendMessage1(@PathVariable("message") String normalMessage) {
kafkaTemplate.send("topic1", normalMessage);
}
/**
* 会回滚, 增加配置spring.kafka.producer.transaction-id-prefix=kafka_tx.
* @param input
*/
@GetMapping("/kafka/send/{input}")
@Transactional(rollbackFor = RuntimeException.class)
public void sendFoo(@PathVariable String input) {
kafkaTemplate.send("topic1", "kl");
if ("error".equals(input)) {
throw new RuntimeException("failed");
}
kafkaTemplate.send("topic1", "ckl");
}
}
4.KafkaConsumer
@Component
public class KafkaConsumer {
// 消费监听
@KafkaListener(id = "webGroup", topics = {"topic1"})
public void onMessage1(ConsumerRecord<?, ?> record){
// 消费的哪个topic、partition的消息,打印出消息内容
System.out.println("简单消费:"+record.topic()+"-"+record.partition()+"-"+record.value());
}
}
以上是基础的操作,还可以设置回调,设置定时器来启动或者停止consumer的监听等,参考:(25条消息) SpringBoot集成kafka全面实战_Felix-CSDN博客_springboot集成kafka
包括以下内容:
5.一些配置参数
###########【Kafka集群】########### spring.kafka.bootstrap-servers=112.126.74.249:9092,112.126.74.249:9093 ###########【初始化生产者配置】########### # 重试次数 spring.kafka.producer.retries=0 # 应答级别:多少个分区副本备份完成时向生产者发送ack确认(可选0、1、all/-1) spring.kafka.producer.acks=1 # 批量大小 spring.kafka.producer.batch-size=16384 # 提交延时 spring.kafka.producer.properties.linger.ms=0 # 当生产端积累的消息达到batch-size或接收到消息linger.ms后,生产者就会将消息提交给kafka # linger.ms为0表示每接收到一条消息就提交给kafka,这时候batch-size其实就没用了 # 生产端缓冲区大小 spring.kafka.producer.buffer-memory = 33554432 # Kafka提供的序列化和反序列化类 spring.kafka.producer.key-serializer=org.apache.kafka.common.serialization.StringSerializer spring.kafka.producer.value-serializer=org.apache.kafka.common.serialization.StringSerializer # 自定义分区器 # spring.kafka.producer.properties.partitioner.class=com.felix.kafka.producer.CustomizePartitioner ###########【初始化消费者配置】########### # 默认的消费组ID spring.kafka.consumer.properties.group.id=defaultConsumerGroup # 是否自动提交offset spring.kafka.consumer.enable-auto-commit=true # 提交offset延时(接收到消息后多久提交offset) spring.kafka.consumer.auto.commit.interval.ms=1000 # 当kafka中没有初始offset或offset超出范围时将自动重置offset # earliest:重置为分区中最小的offset; # latest:重置为分区中最新的offset(消费分区中新产生的数据); # none:只要有一个分区不存在已提交的offset,就抛出异常; spring.kafka.consumer.auto-offset-reset=latest # 消费会话超时时间(超过这个时间consumer没有发送心跳,就会触发rebalance操作) spring.kafka.consumer.properties.session.timeout.ms=120000 # 消费请求超时时间 spring.kafka.consumer.properties.request.timeout.ms=180000 # Kafka提供的序列化和反序列化类 spring.kafka.consumer.key-deserializer=org.apache.kafka.common.serialization.StringDeserializer spring.kafka.consumer.value-deserializer=org.apache.kafka.common.serialization.StringDeserializer # 消费端监听的topic不存在时,项目启动会报错(关掉) spring.kafka.listener.missing-topics-fatal=false # 设置批量消费 # spring.kafka.listener.type=batch # 批量消费每次最多消费多少条消息 # spring.kafka.consumer.max-poll-records=50
6.带回调的生产者
kafkaTemplate提供了一个回调方法addCallback,我们可以在回调方法中监控消息是否发送成功 或 失败时做补偿处理,有两种写法,
@GetMapping("/kafka/callbackOne/{message}")
public void sendMessage2(@PathVariable("message") String callbackMessage) {
kafkaTemplate.send("topic1", callbackMessage).addCallback(success -> {
// 消息发送到的topic
String topic = success.getRecordMetadata().topic();
// 消息发送到的分区
int partition = success.getRecordMetadata().partition();
// 消息在分区内的offset
long offset = success.getRecordMetadata().offset();
System.out.println("发送消息成功:" + topic + "-" + partition + "-" + offset);
}, failure -> {
System.out.println("发送消息失败:" + failure.getMessage());
});
}
@GetMapping("/kafka/callbackTwo/{message}")
public void sendMessage3(@PathVariable("message") String callbackMessage) {
kafkaTemplate.send("topic1", callbackMessage).addCallback(new ListenableFutureCallback<SendResult<String, Object>>() {
@Override
public void onFailure(Throwable ex) {
System.out.println("发送消息失败:"+ex.getMessage());
}
@Override
public void onSuccess(SendResult<String, Object> result) {
System.out.println("发送消息成功:" + result.getRecordMetadata().topic() + "-"
+ result.getRecordMetadata().partition() + "-" + result.getRecordMetadata().offset());
}
});
}
7.自定义分区器
我们知道,kafka中每个topic被划分为多个分区,那么生产者将消息发送到topic时,具体追加到哪个分区呢?这就是所谓的分区策略,Kafka 为我们提供了默认的分区策略,同时它也支持自定义分区策略。其路由机制为:
① 若发送消息时指定了分区(即自定义分区策略),则直接将消息append到指定分区;
② 若发送消息时未指定 patition,但指定了 key(kafka允许为每条消息设置一个key),则对key值进行hash计算,根据计算结果路由到指定分区,这种情况下可以保证同一个 Key 的所有消息都进入到相同的分区;
③ patition 和 key 都未指定,则使用kafka默认的分区策略,轮询选出一个 patition;
※ 我们来自定义一个分区策略,将消息发送到我们指定的partition,首先新建一个分区器类实现Partitioner接口,重写方法,其中partition方法的返回值就表示将消息发送到几号分区,
public class CustomizePartitioner implements Partitioner {
@Override
public int partition(String topic, Object key, byte[] keyBytes, Object value, byte[] valueBytes, Cluster cluster) {
// 自定义分区规则(这里假设全部发到0号分区)
// ......
return 0;
}
@Override
public void close() {
}
@Override
public void configure(Map<String, ?> configs) {
}
}
在application.propertise中配置自定义分区器,配置的值就是分区器类的全路径名,
# 自定义分区器
spring.kafka.producer.properties.partitioner.class=com.felix.kafka.producer.CustomizePartitioner
8.kafka事务提交
如果在发送消息时需要创建事务,可以使用 KafkaTemplate 的 executeInTransaction 方法来声明事务,
@GetMapping("/kafka/transaction")
public void sendMessage7(){
// 声明事务:后面报错消息不会发出去
kafkaTemplate.executeInTransaction(operations -> {
operations.send("topic1","test executeInTransaction");
throw new RuntimeException("fail");
});
// 不声明事务:后面报错但前面消息已经发送成功了
kafkaTemplate.send("topic1","test executeInTransaction");
throw new RuntimeException("fail");
}
9.指定topic、partition、offset消费
/**
* @Title 指定topic、partition、offset消费
* @Description 同时监听topic1和topic2,监听topic1的0号分区、topic2的 "0号和1号" 分区,指向1号分区的offset初始值为8
* @Author long.yuan
* @Date 2020/3/22 13:38
* @Param [record]
* @return void
**/
@KafkaListener(id = "consumer1",groupId = "felix-group",topicPartitions = {
@TopicPartition(topic = "topic1", partitions = { "0" }),
@TopicPartition(topic = "topic2", partitions = "0", partitionOffsets = @PartitionOffset(partition = "1", initialOffset = "8"))
})
public void onMessage2(ConsumerRecord<?, ?> record) {
System.out.println("topic:"+record.topic()+"|partition:"+record.partition()+"|offset:"+record.offset()+"|value:"+record.value());
}
topics和topicPartitions不能同时使用;
10.批量消费
# 设置批量消费
spring.kafka.listener.type=batch
# 批量消费每次最多消费多少条消息
spring.kafka.consumer.max-poll-records=50
@KafkaListener(id = "consumer2",groupId = "felix-group", topics = "topic1")
public void onMessage3(List<ConsumerRecord<?, ?>> records) {
System.out.println(">>>批量消费一次,records.size()="+records.size());
for (ConsumerRecord<?, ?> record : records) {
System.out.println(record.value());
}
}
11.自定义异常处理
// 新建一个异常处理器,用@Bean注入
@Bean
public ConsumerAwareListenerErrorHandler consumerAwareErrorHandler() {
return (message, exception, consumer) -> {
System.out.println("消费异常:"+message.getPayload());
return null;
};
}
// 将这个异常处理器的BeanName放到@KafkaListener注解的errorHandler属性里面
@KafkaListener(topics = {"topic1"},errorHandler = "consumerAwareErrorHandler")
public void onMessage4(ConsumerRecord<?, ?> record) throws Exception {
throw new Exception("简单消费-模拟异常");
}
// 批量消费也一样,异常处理器的message.getPayload()也可以拿到各条消息的信息
@KafkaListener(topics = "topic1",errorHandler="consumerAwareErrorHandler")
public void onMessage5(List<ConsumerRecord<?, ?>> records) throws Exception {
System.out.println("批量消费一次...");
throw new Exception("批量消费-模拟异常");
}
12.消息过滤器
@Component
public class KafkaConsumer {
@Autowired
ConsumerFactory consumerFactory;
// 消息过滤器
@Bean
public ConcurrentKafkaListenerContainerFactory filterContainerFactory() {
ConcurrentKafkaListenerContainerFactory factory = new ConcurrentKafkaListenerContainerFactory();
factory.setConsumerFactory(consumerFactory);
// 被过滤的消息将被丢弃
factory.setAckDiscarded(true);
// 消息过滤策略
factory.setRecordFilterStrategy(consumerRecord -> {
if (Integer.parseInt(consumerRecord.value().toString()) % 2 == 0) {
return false;
}
//返回true消息则被过滤
return true;
});
return factory;
}
// 消息过滤监听
@KafkaListener(topics = {"topic1"},containerFactory = "filterContainerFactory")
public void onMessage6(ConsumerRecord<?, ?> record) {
System.out.println(record.value());
}
}
13.消息转发
/**
* @Title 消息转发
* @Description 从topic1接收到的消息经过处理后转发到topic2
* @Author long.yuan
* @Date 2020/3/23 22:15
* @Param [record]
* @return void
**/
@KafkaListener(topics = {"topic1"})
@SendTo("topic2")
public String onMessage7(ConsumerRecord<?, ?> record) {
return record.value()+"-forward message";
}
14.定时启动、停止监听器
默认情况下,当消费者项目启动的时候,监听器就开始工作,监听消费发送到指定topic的消息,那如果我们不想让监听器立即工作,想让它在我们指定的时间点开始工作,或者在我们指定的时间点停止工作,该怎么处理呢——使用KafkaListenerEndpointRegistry,下面我们就来实现:
① 禁止监听器自启动;
② 创建两个定时任务,一个用来在指定时间点启动定时器,另一个在指定时间点停止定时器;
新建一个定时任务类,用注解@EnableScheduling声明,KafkaListenerEndpointRegistry 在SpringIO中已经被注册为Bean,直接注入,设置禁止KafkaListener自启动,
@EnableScheduling
@Component
public class CronTimer {
/**
* @KafkaListener注解所标注的方法并不会在IOC容器中被注册为Bean,
* 而是会被注册在KafkaListenerEndpointRegistry中,
* 而KafkaListenerEndpointRegistry在SpringIOC中已经被注册为Bean
**/
@Autowired
private KafkaListenerEndpointRegistry registry;
@Autowired
private ConsumerFactory consumerFactory;
// 监听器容器工厂(设置禁止KafkaListener自启动)
@Bean
public ConcurrentKafkaListenerContainerFactory delayContainerFactory() {
ConcurrentKafkaListenerContainerFactory container = new ConcurrentKafkaListenerContainerFactory();
container.setConsumerFactory(consumerFactory);
//禁止KafkaListener自启动
container.setAutoStartup(false);
return container;
}
// 监听器
@KafkaListener(id="timingConsumer",topics = "topic1",containerFactory = "delayContainerFactory")
public void onMessage1(ConsumerRecord<?, ?> record){
System.out.println("消费成功:"+record.topic()+"-"+record.partition()+"-"+record.value());
}
// 定时启动监听器
@Scheduled(cron = "0 42 11 * * ? ")
public void startListener() {
System.out.println("启动监听器...");
// "timingConsumer"是@KafkaListener注解后面设置的监听器ID,标识这个监听器
if (!registry.getListenerContainer("timingConsumer").isRunning()) {
registry.getListenerContainer("timingConsumer").start();
}
//registry.getListenerContainer("timingConsumer").resume();
}
// 定时停止监听器
@Scheduled(cron = "0 45 11 * * ? ")
public void shutDownListener() {
System.out.println("关闭监听器...");
registry.getListenerContainer("timingConsumer").pause();
}
}
15.同组不同组的消费者设置
// 消费监听
@KafkaListener(id = "KafkaConsumer2", groupId = "webGroup", topics = {"topic1"})
public void onMessage(ConsumerRecord<?, ?> record){
// 消费的哪个topic、partition的消息,打印出消息内容
log.info("KafkaConsumer2_webGroup:"+record.topic()+"-"+record.partition()+"_"+record.offset()+"-"+record.value());
}
// 消费监听
@KafkaListener(id = "KafkaConsumer2_1", groupId = "webGroup", topics = {"topic1"})
public void onMessage1(ConsumerRecord<?, ?> record){
// 消费的哪个topic、partition的消息,打印出消息内容
log.info("KafkaConsumer2_1_webGroup:"+record.topic()+"-"+record.partition()+"_"+record.offset()+"-"+record.value());
}
// 消费监听
@KafkaListener(id = "KafkaConsumer2_1", groupId = "webGroup2", topics = {"topic1"})
public void onMessage1(ConsumerRecord<?, ?> record){
// 消费的哪个topic、partition的消息,打印出消息内容
log.info("KafkaConsumer2_1_webGroup2:"+record.topic()+"-"+record.partition()+"_"+record.offset()+"-"+record.value());
}
同一个消息在一个组里只能有一个消费,不同组可以多次消费。如果需要保证一个组消费顺序,那就要发送到同一个分区。
配置错误: 不过如果要确保幂等性功能正常,还需要确保生产者客户端的 retries、acks、max.in.flight.requests.per.connection 这几个参数不被配置错。实际上在使用幂等性功能的时候,用户完全可以不用配置(也不建议配置)这几个参数。 如果用户显式地指定了 retries 参数,那么这个参数的值必须大于0,否则会报出 ConfigException: org.apache.kafka.common.config.ConfigException: Must set retries to non-zero when using the idempotent producer. 如果用户没有显式地指定 retries 参数,那么 KafkaProducer 会将它置为 Integer.MAX_VALUE。同时还需要保证 max.in.flight.requests.per.connection 参数的值不能大于5(这个参数的值默认为5,在 2.2.1 节中有相关的介绍),否则也会报出 ConfigException: org.apache.kafka.common.config.ConfigException: Must set max.in.flight. requests.per.connection to at most 5 to use the idempotent producer. 如果用户还显式地指定了 acks 参数,那么还需要保证这个参数的值为 -1(all),如果不为 -1(这个参数的值默认为1),那么也会报出 ConfigException: org.apache.kafka.common.config.ConfigException: Must set acks to all in order to use the idempotent producer. Otherwise we cannot guarantee idempotence. 如果用户没有显式地指定这个参数,那么 KafkaProducer 会将它置为-1。开启幂等性功能之后,生产者就可以如同未开启幂等时一样发送消息了。
还有使用代码设置批量消费的,参考:https://blog.csdn.net/weixin_43215250/article/details/103241644
import org.apache.kafka.clients.CommonClientConfigs;
import org.apache.kafka.clients.consumer.ConsumerConfig;
import org.apache.kafka.common.config.SaslConfigs;
import org.apache.kafka.common.config.SslConfigs;
import org.apache.kafka.common.serialization.StringDeserializer;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.kafka.annotation.EnableKafka;
import org.springframework.kafka.config.ConcurrentKafkaListenerContainerFactory;
import org.springframework.kafka.config.KafkaListenerContainerFactory;
import org.springframework.kafka.core.ConsumerFactory;
import org.springframework.kafka.core.DefaultKafkaConsumerFactory;
import org.springframework.kafka.listener.AbstractMessageListenerContainer;
import org.springframework.kafka.listener.ConcurrentMessageListenerContainer;
import java.util.HashMap;
import java.util.Map;
@Configuration
@EnableKafka
public class KafkaConsumerConfig {
@Value("${kafka.bootstrap-servers}")
private String servers;
@Value("${kafka.consumer.enable-auto-commit}")
private boolean enableAutoCommit;
@Value("${kafka.consumer.session-timeout}")
private String sessionTimeout;
@Value("${kafka.consumer.auto-commit-interval}")
private String autoCommitInterval;
@Value("${kafka.consumer.group-id}")
private String groupId;
@Value("${kafka.consumer.auto-offset-reset}")
private String autoOffsetReset;
@Value("${kafka.consumer.concurrency}")
private int concurrency;
@Value("${kafka.consumer.max-poll-records}")
private int maxPollRecords;
@Value("${kafka.sasl-mechanism}")
private String saslMechanism;
@Value("${kafka.ssl.truststore.location}")
private String truststoreLocation;
@Value("${kafka.ssl.truststore.password}")
private String truststorePassword;
@Value("${kafka.ssl.security-protocol}")
private String securityProtocol;
@Value("${java.security.auth.login.config}")
private String authLoginConfig;
@Bean
public KafkaListenerContainerFactory<ConcurrentMessageListenerContainer<String, String>> kafkaListenerContainerFactory() {
ConcurrentKafkaListenerContainerFactory<String, String> factory = new ConcurrentKafkaListenerContainerFactory<>();
factory.setConsumerFactory(consumerFactory());
// 并发创建的消费者数量
factory.setConcurrency(concurrency);
// 开启批处理
factory.setBatchListener(true);
factory.getContainerProperties().setPollTimeout(1500);
if (!enableAutoCommit) {
factory.getContainerProperties().setAckMode((AbstractMessageListenerContainer.AckMode.MANUAL));
}
return factory;
}
@Bean
public ConsumerFactory<String, String> consumerFactory() {
return new DefaultKafkaConsumerFactory<>(getCommonPropertis(groupId));
}
private Map<String, Object> getCommonPropertis(String groupId) {
Map<String, Object> props = new HashMap<>();
props.put(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, servers);
props.put(ConsumerConfig.ENABLE_AUTO_COMMIT_CONFIG, enableAutoCommit);
props.put(ConsumerConfig.AUTO_COMMIT_INTERVAL_MS_CONFIG, autoCommitInterval);
props.put(ConsumerConfig.SESSION_TIMEOUT_MS_CONFIG, sessionTimeout);
props.put(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class);
props.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class);
props.put(ConsumerConfig.GROUP_ID_CONFIG, groupId);
props.put(ConsumerConfig.AUTO_OFFSET_RESET_CONFIG, autoOffsetReset);
props.put(ConsumerConfig.MAX_POLL_RECORDS_CONFIG, maxPollRecords);
props.put(ConsumerConfig.FETCH_MAX_BYTES_CONFIG, 10485760);
props.put(ConsumerConfig.RECEIVE_BUFFER_CONFIG, 10485760);
props.put(CommonClientConfigs.SECURITY_PROTOCOL_CONFIG, securityProtocol);
props.put(SslConfigs.SSL_TRUSTSTORE_LOCATION_CONFIG, truststoreLocation);
props.put(SslConfigs.SSL_TRUSTSTORE_PASSWORD_CONFIG, truststorePassword);
props.put(SaslConfigs.SASL_MECHANISM, saslMechanism);
return props;
}
}
@KafkaListener(id = "test", topics = "#{'${kafka.consumer.topics}'.split(',')}", containerFactory = "kafkaListenerContainerFactory")
public void consumeList(List<ConsumerRecord<String, String>> records, Acknowledgment ack) {
logger.info("=============== Total " + records.size() + " events in this batch ..");
try {
List<String> list = new ArrayList<String>();
for (ConsumerRecord<String, String> record : records) {
Optional<?> kafkaMessage = Optional.ofNullable(record.value());
if (kafkaMessage.isPresent()) {
String message = record.value();
list.add(message);
String topic = record.topic();
logger.info("message = {}", message);
}
}
boolean handle = this.batchHandle(list);
if (handle) {
// 直接提交 offset
ack.acknowledge();
}
} catch (Exception e) {
logger.error(e.getMessage(), e);
}
}