之前写了几篇关于Spring Cloud Stream使用中的常见问题,比如:
下面几天就集中来详细聊聊,当消息消费失败之后该如何处理的几种方式。不过不论哪种方式,都需要与具体业务结合,解决不同业务场景可能出现的问题。
今天第一节,介绍一下Spring Cloud Stream中默认就已经配置了的一个异常解决方案:重试!
依然要明确一点,任何解决方案都要结合具体的业务实现来确定,不要有了锤子看什么问题都是钉子。那么重试可以解决什么问题呢?由于重试的基础逻辑并不会改变,所以通常重试只能解决因环境不稳定等外在因素导致的失败情况,比如:当我们接收到某个消息之后,需要调用一个外部的Web Service做一些事情,这个时候如果与外部系统的网络出现了抖动,导致调用失败而抛出异常。这个时候,通过重试消息消费的具体逻辑,可能在下一次调用的时候,就能完成整合业务动作,从而解决刚才所述的问题。
先通过一个小例子来看看Spring Cloud Stream默认的重试机制是如何运作的。之前在 如何消费自己生产的消息 一文中的例子,我们可以继续沿用,或者也可以精简一些,都写到一个主类中,比如下面这样:
@EnableBinding(TestApplication.TestTopic.class) @SpringBootApplication public class TestApplication { public static void main(String[] args) { SpringApplication.run(TestApplication.class, args); } @RestController static class TestController { @Autowired private TestTopic testTopic; /** * 消息生产接口 * @param message * @return */ @GetMapping("/sendMessage") public String messageWithMQ(@RequestParam String message) { testTopic.output().send(MessageBuilder.withPayload(message).build()); return "ok"; } } /** * 消息消费逻辑 */ @Slf4j @Component static class TestListener { @StreamListener(TestTopic.INPUT) public void receive(String payload) { log.info("Received: " + payload); throw new RuntimeException("Message consumer failed!"); } } interface TestTopic { String OUTPUT = "example-topic-output"; String INPUT = "example-topic-input"; @Output(OUTPUT) MessageChannel output(); @Input(INPUT) SubscribableChannel input(); } }
内容很简单,既包含了消息的生产,也包含了消息消费。与之前例子不同的就是在消息消费逻辑中,主动的抛出了一个异常来模拟消息的消费失败。
在启动应用之前,还要记得配置一下输入输出通道对应的物理目标(exchange或topic名),比如:
spring.cloud.stream.bindings.example-topic-input.destination=test-topic spring.cloud.stream.bindings.example-topic-output.destination=test-topic
完成了上面配置之后,就可以启动应用,并尝试访问 localhost:8080/sendMessage?message=hello
接口来发送一个消息到MQ中了。此时可以看到类似下面的日志:
2018-12-10 11:20:21.345 INFO 30499 --- [w2p2yKethOsqg-1] c.d.stream.TestApplication$TestListener : Received: hello 2018-12-10 11:20:22.350 INFO 30499 --- [w2p2yKethOsqg-1] c.d.stream.TestApplication$TestListener : Received: hello 2018-12-10 11:20:24.354 INFO 30499 --- [w2p2yKethOsqg-1] c.d.stream.TestApplication$TestListener : Received: hello 2018-12-10 11:20:54.651 ERROR 30499 --- [w2p2yKethOsqg-1] o.s.integration.handler.LoggingHandler : org.springframework.messaging.MessagingException: Exception thrown while invoking com.didispace.stream.TestApplication$TestListener#receive[1 args]; nested exception is java.lang.RuntimeException: Message consumer failed!, failedMessage=GenericMessage [payload=byte[5], headers={amqp_receivedDeliveryMode=PERSISTENT, amqp_receivedRoutingKey=test-topic, amqp_receivedExchange=test-topic, amqp_deliveryTag=2, deliveryAttempt=3, amqp_consumerQueue=test-topic.anonymous.EuqBJu66Qw2p2yKethOsqg, amqp_redelivered=false, id=a89adf96-7de2-f29d-20b6-2fcb0c64cd8c, amqp_consumerTag=amq.ctag-XFy6vXU2w4RB_NRBzImWTA, contentType=application/json, timestamp=1544412051638}] at org.springframework.cloud.stream.binding.StreamListenerMessageHandler.handleRequestMessage(StreamListenerMessageHandler.java:63) at org.springframework.integration.handler.AbstractReplyProducingMessageHandler.handleMessageInternal(AbstractReplyProducingMessageHandler.java:109) at org.springframework.integration.handler.AbstractMessageHandler.handleMessage(AbstractMessageHandler.java:158) at org.springframework.integration.dispatcher.AbstractDispatcher.tryOptimizedDispatch(AbstractDispatcher.java:116) at org.springframework.integration.dispatcher.UnicastingDispatcher.doDispatch(UnicastingDispatcher.java:132) at org.springframework.integration.dispatcher.UnicastingDispatcher.dispatch(UnicastingDispatcher.java:105) at org.springframework.integration.channel.AbstractSubscribableChannel.doSend(AbstractSubscribableChannel.java:73) at org.springframework.integration.channel.AbstractMessageChannel.send(AbstractMessageChannel.java:445) at org.springframework.integration.channel.AbstractMessageChannel.send(AbstractMessageChannel.java:394) at org.springframework.messaging.core.GenericMessagingTemplate.doSend(GenericMessagingTemplate.java:181) at org.springframework.messaging.core.GenericMessagingTemplate.doSend(GenericMessagingTemplate.java:160) at org.springframework.messaging.core.GenericMessagingTemplate.doSend(GenericMessagingTemplate.java:47) at org.springframework.messaging.core.AbstractMessageSendingTemplate.send(AbstractMessageSendingTemplate.java:108) at org.springframework.integration.endpoint.MessageProducerSupport.sendMessage(MessageProducerSupport.java:203) at org.springframework.integration.amqp.inbound.AmqpInboundChannelAdapter.access$1100(AmqpInboundChannelAdapter.java:60) at org.springframework.integration.amqp.inbound.AmqpInboundChannelAdapter$Listener.lambda$onMessage$0(AmqpInboundChannelAdapter.java:214) at org.springframework.retry.support.RetryTemplate.doExecute(RetryTemplate.java:287) at org.springframework.retry.support.RetryTemplate.execute(RetryTemplate.java:180) at org.springframework.integration.amqp.inbound.AmqpInboundChannelAdapter$Listener.onMessage(AmqpInboundChannelAdapter.java:211) at org.springframework.amqp.rabbit.listener.AbstractMessageListenerContainer.doInvokeListener(AbstractMessageListenerContainer.java:1414) at org.springframework.amqp.rabbit.listener.AbstractMessageListenerContainer.actualInvokeListener(AbstractMessageListenerContainer.java:1337) at org.springframework.amqp.rabbit.listener.AbstractMessageListenerContainer.invokeListener(AbstractMessageListenerContainer.java:1324) at org.springframework.amqp.rabbit.listener.AbstractMessageListenerContainer.executeListener(AbstractMessageListenerContainer.java:1303) at org.springframework.amqp.rabbit.listener.SimpleMessageListenerContainer.doReceiveAndExecute(SimpleMessageListenerContainer.java:817) at org.springframework.amqp.rabbit.listener.SimpleMessageListenerContainer.receiveAndExecute(SimpleMessageListenerContainer.java:801) at org.springframework.amqp.rabbit.listener.SimpleMessageListenerContainer.access$700(SimpleMessageListenerContainer.java:77) at org.springframework.amqp.rabbit.listener.SimpleMessageListenerContainer$AsyncMessageProcessingConsumer.run(SimpleMessageListenerContainer.java:1042) at java.lang.Thread.run(Thread.java:748) Caused by: java.lang.RuntimeException: Message consumer failed! at com.didispace.stream.TestApplication$TestListener.receive(TestApplication.java:65) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) at java.lang.reflect.Method.invoke(Method.java:498) at org.springframework.messaging.handler.invocation.InvocableHandlerMethod.doInvoke(InvocableHandlerMethod.java:181) at org.springframework.messaging.handler.invocation.InvocableHandlerMethod.invoke(InvocableHandlerMethod.java:114) at org.springframework.cloud.stream.binding.StreamListenerMessageHandler.handleRequestMessage(StreamListenerMessageHandler.java:55) ... 27 more
从日志中可以看到,一共输出了三次 Received: hello
,也就是说消息消费逻辑执行了3次,然后抛出了最终执行失败的异常。
默认情况下Spring Cloud Stream会重试3次,我们也可以通过配置的方式修改这个默认配置,比如下面的配置可以将重试次数调整为1次:
spring.cloud.stream.bindings.example-topic-input.consumer.max-attempts=1
对于一些纯内部计算逻辑,不需要依赖外部环境,如果出错通常是代码逻辑错误的情况下,不论我们如何重试都会继续错误的业务逻辑可以将该参数设置为0,避免不必要的重试影响消息处理的速度。
完成了上面的基础尝试之后,再思考下面两个问题:
答案是不会。因为重试过程是消息处理的一个整体,如果某一次重试成功了,会任务对所收到消息的消费成功了。
这个问题可以在上述例子中做一些小改动来验证,比如:
@Slf4j @Component static class TestListener { int counter = 1; @StreamListener(TestTopic.INPUT) public void receive(String payload) { log.info("Received: " + payload + ", " + counter); if (counter == 3) { counter = 1; return; } else { counter++; throw new RuntimeException("Message consumer failed!"); } } }
通过加入一个计数器,当重试是第3次的时候,不抛出异常来模拟消费逻辑处理成功了。此时重新运行程序,并调用接口 localhost:8080/sendMessage?message=hello
,可以获得如下日志结果,并没有异常打印出来。
2018-12-10 16:07:38.390 INFO 66468 --- [L6MGAj-MAj7QA-1] c.d.stream.TestApplication$TestListener : Received: hello, 1 2018-12-10 16:07:39.398 INFO 66468 --- [L6MGAj-MAj7QA-1] c.d.stream.TestApplication$TestListener : Received: hello, 2 2018-12-10 16:07:41.402 INFO 66468 --- [L6MGAj-MAj7QA-1] c.d.stream.TestApplication$TestListener : Received: hello, 3
也就是,虽然前两次消费抛出了异常,但是并不影响最终的结果,也不会打印中间过程的异常,避免了对日志告警产生误报等问题。
如果消息在重试了还是失败之后,目前的配置唯一能做的就是将异常信息记录下来,进行告警。由于日志中有消息的消息信息描述,所以应用维护者可以根据这些信息来做一些补救措施。
当然,这样的做法显然不是最好的,因为太过麻烦。那么怎么做才好呢?且听下回分解!
本文示例读者可以通过查看下面仓库的中的 stream-exception-handler-1
项目: