Saga handler not executed if Event Handling Component and Saga have identical Processing Group specified

In order to be able to use Saga with axon-kafka I have set the following configuration

public class MySagaConfig {

  @Bean
  public SagaStore mySagaStore() {
    return new InMemorySagaStore(); //JdbcSagaStore, InMemorySagaStore, JpaSagaStore and MongoSagaStore.
  }

  @Bean
  public SagaConfiguration<MySaga> mySagaConfiguration(@Autowired KafkaMessageSource<String, Object> kafkaMessageSource) {
    return SagaConfiguration.trackingSagaManager(MySaga.class, "MyProcessor", configuration -> kafkaMessageSource);
  }
}

You can find the full example here https://github.com/marinkobabic/axon-kafka-example/

It seems if there is already a processor “MyProcessor” registered using the SimpleMethodInvoker then the MultiEventHandlerInvoker is not anymore used. When I remove the processor name in the code above for the Saga and define a different name in the annotation of the Saga “MyProcessor2”, then the methods are called. I have done this hack just to see if there is any difference, but I think Kafka is not supporting multiple processors. When the hack is used then the normal event handler are not executed anymore.

[marinkobabic]

I will do a test and let you know. I have seen once this exception

java.util.ConcurrentModificationException: KafkaConsumer is not safe for multi-threaded access

Hope you have a new KafkaConsumer instance per Trackingprocessor.

On Mon, Sep 3, 2018, 12:47 Steven van Beelen [e-mail] wrote:

Axon Kafka shouldn’t make a difference in this scenario! As long as you configure both your Processing Groups to have Kafka as the message source, you should be good.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/AxonFramework/AxonFramework/issues/734#issuecomment-418077275, or mute the thread https://github.com/notifications/unsubscribe-auth/ADHSOBiMkXgt2LOpoyKMNg9tUGj2p59-ks5uXQjVgaJpZM4WOoSx .

[smcvb]

Closing this issue in assumption that #750 has resolved the problem. This is part of 3.3.6. If you’ve got time checking it out @marinkobabic, that would be nice! If issues still arise, then we very likely will reopen this issue.