RabbitTemplate convertAndSend上的Spring Amqp内部NullPointerExcep

Question

在尝试使用RabbitTemplate.covertAndSend 发布消息时，我随机遇到NullPointerException

我尝试了spring-amqp:1.7.6和spring-amqp:1.7.8，这应该是我的依赖树的相关部分：

[INFO] | +- org.springframework.amqp:spring-rabbit:jar:1.7.4.RELEASE:compile
[INFO] | | +- com.rabbitmq:http-client:jar:1.1.1.RELEASE:compile
[INFO] | | - com.rabbitmq:amqp-client:jar:4.0.3:compile

这是异常的堆栈

org.springframework.amqp.UncategorizedAmqpException: java.lang.NullPointerException
at org.springframework.amqp.rabbit.support.RabbitExceptionTranslator.convertRabbitAccessException(RabbitExceptionTranslator.java:83)
at org.springframework.amqp.rabbit.connection.RabbitAccessor.convertRabbitAccessException(RabbitAccessor.java:113)
at org.springframework.amqp.rabbit.core.RabbitTemplate.doExecute(RabbitTemplate.java:1461)
at org.springframework.amqp.rabbit.core.RabbitTemplate.execute(RabbitTemplate.java:1411)
at org.springframework.amqp.rabbit.core.RabbitTemplate.send(RabbitTemplate.java:712)
at org.springframework.amqp.rabbit.core.RabbitTemplate.convertAndSend(RabbitTemplate.java:813)
at org.springframework.amqp.rabbit.core.RabbitTemplate.convertAndSend(RabbitTemplate.java:791)
at io.reactivex.internal.operators.completable.CompletableFromAction.subscribeActual(CompletableFromAction.java:34)
at io.reactivex.Completable.subscribe(Completable.java:1635)
at io.reactivex.internal.operators.completable.CompletableCache.subscribeActual(CompletableCache.java:59)
at io.reactivex.Completable.subscribe(Completable.java:1635)
at io.reactivex.internal.operators.completable.CompletableSubscribeOn$SubscribeOnObserver.run(CompletableSubscribeOn.java:64)
at io.reactivex.Scheduler$DisposeTask.run(Scheduler.java:452)
at io.reactivex.internal.schedulers.ScheduledRunnable.run(ScheduledRunnable.java:61)
at io.reactivex.internal.schedulers.ScheduledRunnable.call(ScheduledRunnable.java:52)
at java.util.concurrent.FutureTask.run(FutureTask.java:266)
at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(ScheduledThreadPoolExecutor.java:180)
at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)
Caused by: java.lang.NullPointerException: null
at sun.reflect.GeneratedMethodAccessor487.invoke(Unknown Source)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:498)
at org.springframework.amqp.rabbit.connection.CachingConnectionFactory$CachedChannelInvocationHandler.invoke(CachingConnectionFactory.java:980)
at com.sun.proxy.$Proxy174.basicPublish(Unknown Source)
at org.springframework.amqp.rabbit.core.RabbitTemplate.doSend(RabbitTemplate.java:1532)
at org.springframework.amqp.rabbit.core.RabbitTemplate$3.doInRabbit(RabbitTemplate.java:716)
at org.springframework.amqp.rabbit.core.RabbitTemplate.doExecute(RabbitTemplate.java:1455)
... 19 common frames omitted 

有什么线索可以说明是什么原因造成的吗？

编辑：附加信息

起初我没有注意到spring-rabbit和spring-amqp之间的区别，但从jar内部看，我有以下依赖项：

128356 Tue May 29 17:50:46 CEST 2018 lib/spring-amqp-1.7.8.RELEASE.jar
469867 Mon Sep 11 15:28:04 CEST 2017 lib/spring-rabbit-1.7.4.RELEASE.jar
481919 Wed Jun 07 15:25:06 CEST 2017 lib/amqp-client-4.0.3.jar

Answer 1

我想说问题出在这样的块中：

synchronized (this.targetMonitor) {
if (this.target == null) {
this.target = createBareChannel(this.theConnection, this.transactional);
}
Object result = method.invoke(this.target, args);
if (this.transactional) {
if (txStarts.contains(methodName)) {
this.txStarted = true;
}
else if (txEnds.contains(methodName)) {
this.txStarted = false;
}
}
return result;
}

注意if (this.target == null) {和method.invoke(this.target, args);。在这些操作之间，this.target可以变为null。这就是下面catch (InvocationTargetException ex) {块的事实：我们在synchronized (this.targetMonitor) {之外执行this.target = null;。因此，某些其他线程可能会导致此target属性出现竞争条件。

这就是我对NPE问题的意思，这是一个如何解决的建议。

现在我们需要了解，不同的线程如何访问同一个ChannelProxy，从而导致这样的竞争条件。。。也许通道的缓存仍然存在一些缺陷。

当然，如果您使用最新的spring-rabbit-1.7.8尝试您的解决方案，那将是非常好的，因为您的依赖关系仍然显示它是1.7.4。

Answer 2

我在实时环境中随机发布时遇到了同样的错误，这是由于创建了通道而没有关闭它们。Rabbitmq的通道限制为5000个。这在测试环境中不容易再现，因为它需要打开5k通道。