Question

我遇到的情况是不同的行为。共有3种不同的服务

第一个服务将从Solace队列中侦听并将其生成给kafka topic-1（启用交易的地方）
第二个服务将从kafka topic-1之上侦听并将其写入另一个kafka topic-2（在这里我们没有手动提交，交易能够产生其他主题，自动提交偏移量为false和 Isolation.level设置为read_commited）之前删除
第三服务将从kafka topic-2侦听并将其写回到Solace队列（在这里我们没有手动提交，自动提交偏移为 false和Isolation.level设置为read_commited。

现在，如果我在第二个服务中启用了事务和隔离级别，则该问题无法读取任何消息，如果我在第二个服务中禁用了该事务，则可以读取所有消息。

我们可以在一项服务中启用事务和隔离级别吗？
如果我的服务只是生产者或消费者，它如何工作（如何保证这些服务的EoS）

编辑：以下是我的yml的外观

 - kafka:
   - binder:
     - transaction:
         - transaction-id-prefix:
       - brokers: 
         - configuration: 
               all my consumer properties (ssl, sasl)

已更新（带有Spring Cloud的yml）：

spring: 
  cloud.stream:
      bindings:
        input:
          destination: test_input
          content-type: application/json
          group: test_group
        output:
          destination: test_output
          content-type: application/json
      kafka.binder: 
          configuration: 
            isolation.level: read_committed
            security.protocol: SASL_SSL
            sasl.mechanism: GSSAPI
            sasl.kerberos.service.name: kafka
            ssl.truststore.location: jks
            ssl.truststore.password: 
            ssl.endpoint.identification.algorithm: null            
          brokers: broker1:9092,broker2:9092,broker3:9092
          auto-create-topics: false
          transaction:
            transaction-id-prefix: trans-2
            producer:
              configuration:
                retries: 2000
                acks: all
                security.protocol: SASL_SSL
                sasl.mechanism: GSSAPI
                sasl.kerberos.service.name: kafka
                ssl.truststore.location: jks
                ssl.truststore.password: 
                ssl.endpoint.identification.algorithm: null

已更新（带有Spring Kafka的yml）：

spring:
  kafka:
    bootstrap-servers: broker1:9092,broker2:9092,broker3:9092
    consumer:
      properties:
        isolation.level: read_committed
        ssl.truststore.location: truststore.jks
        ssl.truststore.password: 
        security.protocol: SASL_SSL
        sasl.mechanism: GSSAPI
        sasl.kerberos.service.name: kafka
    producer:
      transaction-id-prefix: trans-2
      retries: 2000
      acks: all
      properties:
        ssl.truststore.location: truststore.jks
        ssl.truststore.password: 
        security.protocol: SASL_SSL
        sasl.mechanism: GSSAPI
        sasl.kerberos.service.name: kafka
    admin:
      properties:
        ssl.truststore.location: truststore.jks
        ssl.truststore.password: 
        security.protocol: SASL_SSL
        sasl.mechanism: GSSAPI
        sasl.kerberos.service.name: kafka

已更新为动态目的地

Caused by: java.lang.IllegalStateException: Cannot perform operation after producer has been closed
    at org.apache.kafka.clients.producer.KafkaProducer.throwIfProducerClosed(KafkaProducer.java:810) ~[kafka-clients-2.0.0.jar:na]
    at org.apache.kafka.clients.producer.KafkaProducer.doSend(KafkaProducer.java:819) ~[kafka-clients-2.0.0.jar:na]
    at org.apache.kafka.clients.producer.KafkaProducer.send(KafkaProducer.java:803) ~[kafka-clients-2.0.0.jar:na]
    at org.springframework.kafka.core.DefaultKafkaProducerFactory$CloseSafeProducer.send(DefaultKafkaProducerFactory.java:423) ~[spring-kafka-2.2.0.RELEASE.jar:2.2.0.RELEASE]
    at org.springframework.kafka.core.KafkaTemplate.doSend(KafkaTemplate.java:351) ~[spring-kafka-2.2.0.RELEASE.jar:2.2.0.RELEASE]
    at org.springframework.kafka.core.KafkaTemplate.send(KafkaTemplate.java:209) ~[spring-kafka-2.2.0.RELEASE.jar:2.2.0.RELEASE]
    at org.springframework.integration.kafka.outbound.KafkaProducerMessageHandler.handleRequestMessage(KafkaProducerMessageHandler.java:382) ~[spring-integration-kafka-3.1.0.RELEASE.jar:3.1.0.RELEASE]
    at org.springframework.integration.handler.AbstractReplyProducingMessageHandler.handleMessageInternal(AbstractReplyProducingMessageHandler.java:123) [spring-integration-core-5.1.0.RELEASE.jar:5.1.0.RELEASE]
    at org.springframework.integration.handler.AbstractMessageHandler.handleMessage(AbstractMessageHandler.java:169) [spring-integration-core-5.1.0.RELEASE.jar:5.1.0.RELEASE]

尝试了两种解决动态目标解析器问题的方法： dynamic destination resolver

Answer 1

对我来说很好；这些都在同一个应用程序中，但这没什么关系...

@SpringBootApplication
@EnableBinding(Channels.class)
public class So55419549Application {

    public static void main(String[] args) {
        SpringApplication.run(So55419549Application.class, args);
    }

    @Bean
    public IntegrationFlow service1(MessageChannel out1) {
        return IntegrationFlows.from(() -> "foo", e -> e
                    .poller(Pollers.fixedDelay(Duration.ofSeconds(5))))
                .log(Level.INFO, m -> "s1 " + m.getPayload())
                .channel(out1)
                .get();
    }

    @StreamListener("in2")
    @SendTo("out2")
    public String service2(String in) {
        System.out.println("s2 " + in);
        return in.toUpperCase();
    }

    @StreamListener("in3")
    public void service3(String in) {
        System.out.println("s3 " + in);
    }

}

interface Channels {

    @Output
    MessageChannel out1();

    @Input
    MessageChannel in2();

    @Output
    MessageChannel out2();

    @Input
    MessageChannel in3();

}

和

spring:
  cloud:
    stream:
      bindings:
        out1:
          destination: topic1
        in2:
          group: s2
          destination: topic1
        out2:
          destination: topic2
        in3:
          group: s3
          destination: topic2
      kafka:
        binder:
          transaction:
            transaction-id-prefix: tx
        bindings:
          in2:
            consumer:
              configuration:
                isolation:
                  level: read_committed
          in3:
            consumer:
              configuration:
                isolation:
                  level: read_committed
  kafka:
    producer:
      # needed again here so boot declares a TM for us
      transaction-id-prefix: tx
      retries: 10
      acks: all
logging:
  level:
    org.springframework.kafka.transaction: debug

和

2019-03-29 12:57:08.345  INFO 75700 --- [ask-scheduler-1] o.s.integration.handler.LoggingHandler   
    : s1 foo
2019-03-29 12:57:08.353 DEBUG 75700 --- [container-0-C-1] o.s.k.t.KafkaTransactionManager          : Creating new transaction with name [null]: PROPAGATION_REQUIRED,ISOLATION_DEFAULT
2019-03-29 12:57:08.353 DEBUG 75700 --- [container-0-C-1] o.s.k.t.KafkaTransactionManager          : Created Kafka transaction on producer [CloseSafeProducer [delegate=org.apache.kafka.clients.producer.KafkaProducer@6790c874, txId=txs2.topic1.0]]
s2 foo
2019-03-29 12:57:08.357 DEBUG 75700 --- [container-0-C-1] o.s.k.t.KafkaTransactionManager          : Creating new transaction with name [null]: PROPAGATION_REQUIRED,ISOLATION_DEFAULT
2019-03-29 12:57:08.358 DEBUG 75700 --- [container-0-C-1] o.s.k.t.KafkaTransactionManager          : Created Kafka transaction on producer [CloseSafeProducer [delegate=org.apache.kafka.clients.producer.KafkaProducer@820ef3d, txId=txs3.topic2.0]]
s3 FOO

编辑

活页夹未在事务管理器上启用事务同步。解决方法是，添加

TransactionSynchronizationManager.setActualTransactionActive(true);

发送到您的@StreamListener。

我对着活页夹打开了a bug。

带有消费者/生产者API的Kafka的Spring Cloud Stream恰好在带有transaction-id-prefix的语义无法按预期运行时

1 个答案: