我有一个带有n个消息监听器的事件队列。当消息到达时,一个消息侦听器接收它并执行状态机的新实例。 我面临的问题是,虽然并行处理多个消息,但状态机开始按顺序执行操作,尽管它们由不同的状态机实例调用,如下所示:
2017-10-18 16:11:03.740 INFO 30282 --- [lTaskExecutor-1] o.s.s.support.LifecycleObjectSupport : started org.springframework.statemachine.support.DefaultStateMachineExecutor@6ddb1ea6
2017-10-18 16:11:03.741 INFO 30282 --- [lTaskExecutor-1] o.s.s.support.LifecycleObjectSupport : started EVALUATE_IS_WALKTHROUGH SAVE END START / START / uuid=b922b6b1-a441-4924-8531-d45e0e0c9c40 / id=null
2017-10-18 16:11:03.740 INFO 30282 --- [TaskExecutor-10] o.s.s.support.LifecycleObjectSupport : started org.springframework.statemachine.support.DefaultStateMachineExecutor@13b6ace4
2017-10-18 16:11:03.741 INFO 30282 --- [TaskExecutor-10] o.s.s.support.LifecycleObjectSupport : started EVALUATE_IS_WALKTHROUGH SAVE END START / START / uuid=e06a8c1d-beed-41c6-bc63-d8c1a3a56169 / id=null
2017-10-18 16:11:03.759 INFO 30282 --- [pool-5-thread-1] i.b.b.e.processors.actions.SaveAction : [io.botbit.backend.events.processors.actions.SaveAction@607e4071] Saving event Event[id=null,
2017-10-18 16:11:24.046 INFO 30282 --- [pool-5-thread-1] i.b.b.e.processors.actions.SaveAction : [io.botbit.backend.events.processors.actions.SaveAction@607e4071] Saving event Event[id=null,
2017-10-18 16:11:44.058 INFO 30282 --- [pool-5-thread-1] i.b.b.e.p.a.EvaluateIsWalkthroughAction : Evaluation is WT,,,
2017-10-18 16:11:44.059 INFO 30282 --- [pool-5-thread-1] o.s.s.support.LifecycleObjectSupport : stopped org.springframework.statemachine.support.DefaultStateMachineExecutor@6ddb1ea6
2017-10-18 16:11:44.060 INFO 30282 --- [pool-5-thread-1] o.s.s.support.LifecycleObjectSupport : stopped EVALUATE_IS_WALKTHROUGH SAVE END START / / uuid=b922b6b1-a441-4924-8531-d45e0e0c9c40 / id=null
2017-10-18 16:11:44.060 INFO 30282 --- [pool-5-thread-1] i.b.b.e.p.a.EvaluateIsWalkthroughAction : Evaluation is WT,,,
2017-10-18 16:11:44.061 INFO 30282 --- [pool-5-thread-1] o.s.s.support.LifecycleObjectSupport : stopped org.springframework.statemachine.support.DefaultStateMachineExecutor@13b6ace4
2017-10-18 16:11:44.061 INFO 30282 --- [pool-5-thread-1] o.s.s.support.LifecycleObjectSupport : stopped EVALUATE_IS_WALKTHROUGH SAVE END START / / uuid=e06a8c1d-beed-41c6-bc63-d8c1a3a56169 / id=nul
我认为这是因为每个动作都在同一个线程中运行([pool-5-thread-1])。 我需要每个实例完全并行运行,这意味着状态机是并行执行的,但也是它的动作。
任何帮助将不胜感激,谢谢!
@Component
public class EventConsumer {
private final static Logger logger = Logger.getLogger(EventConsumer.class);
@Autowired
private StateMachineFactory<String, String> eventProcessorFactory;
public void consume(Event event) {
logger.info("Received <" + event + ">");
StateMachine<String, String> eventProcessor = eventProcessorFactory.getStateMachine();
eventProcessor.getExtendedState().getVariables().put("event", event);
eventProcessor.start();
eventProcessor.sendEvent(CommonEvents.SUCCESS);
}
public void consume(Collection<Event> events) {
for (Event event : events) {
this.consume(event);
}
}
}
这是状态机配置
@Configuration
@EnableStateMachineFactory
public class WiFiConnectEventProcessorConfig extends StateMachineConfigurerAdapter<String, String> {
@Autowired
SaveAction saveAction;
@Autowired
DeprecateAddVisitationAction addVisitation;
@Autowired
EvaluateIsWalkthroughAction isWTAction;
@Override
public void configure(StateMachineStateConfigurer<String, String> states) throws Exception {
states.withStates().initial(WiFiConnectStates.START).state(WiFiConnectStates.SAVE, saveAction)
.state(WiFiConnectStates.DEPRECATE_ADD_VISITATION, addVisitation)
.state(WiFiConnectStates.EVALUATE_IS_WALKTHROUGH, isWTAction).end(WiFiConnectStates.END);
}
@Override
public void configure(StateMachineTransitionConfigurer<String, String> transitions) throws Exception {
transitions.withExternal().source(WiFiConnectStates.START).target(WiFiConnectStates.SAVE)
.event(CommonEvents.SUCCESS)
.and().withExternal().source(WiFiConnectStates.SAVE)
.target(WiFiConnectStates.DEPRECATE_ADD_VISITATION).event(CommonEvents.SUCCESS)
.and().withExternal()
.source(WiFiConnectStates.DEPRECATE_ADD_VISITATION).target(WiFiConnectStates.EVALUATE_IS_WALKTHROUGH)
.event(CommonEvents.SUCCESS)
.and().withExternal().source(WiFiConnectStates.EVALUATE_IS_WALKTHROUGH)
.target(WiFiConnectStates.END).event(CommonEvents.SUCCESS);
}
}
答案 0 :(得分:1)
您需要创建自己的TaskScheduler,并在StateMachineConfig @Configuration文件中进行如下配置。选择适合您需要的PoolSize。
@Override
public void configure(StateMachineConfigurationConfigurer<TaskState, TaskEvent> config) throws Exception {
config.withConfiguration()
.taskScheduler(myTaskScheduler());
}
@Bean
public TaskScheduler myTaskScheduler() {
final ThreadPoolTaskScheduler scheduler = new ThreadPoolTaskScheduler();
scheduler.setPoolSize(10);
return scheduler;
}
答案 1 :(得分:0)
如评论中所述,您的状态机将共享相同的资源,包括TaskExecutor和TaskScheduler。
的默认行为是在调度程序中使用单线程执行程序,这就是为什么您在应用程序中看到瓶颈的原因。
这说明了可以使用的各种类型的执行程序和调度程序: https://docs.spring.io/spring/docs/4.2.x/spring-framework-reference/html/scheduling.html
然后您可以升级配置:
@Override
public void configure(StateMachineConfigurationConfigurer<RfqState, RfqEvent> config)
throws Exception {
SyncTaskExecutor executor = new SyncTaskExecutor();
config.withConfiguration()
.taskExecutor( my executor )
.taskScheduler( my scheduler );
}