Apache Camel:聚合多播 - AggregationStrategy 调用过于频繁 [英] Apache Camel: multicast with aggregation - AggregationStrategy called too often

问题描述

对于多播 + 聚合，我有以下奇怪的(或至少我不清楚)行为.考虑以下路线:

I have the following strange (or at least unclear to me) behaviour for a multi-cast + aggregation. Consider the following route:

    from("direct:multicaster")
                .multicast()
                .to("direct:A", "direct:B")
                .aggregationStrategy(new AggregationStrategy() {
                    @Override
                    public Exchange aggregate(Exchange oldExchange, Exchange newExchange) {
                        if (oldExchange == null) {
                            List firstResult = newExchange.getIn().getBody(List.class);
                            newExchange.getIn().setBody(ImmutableList.copyOf(firstResult));
                            return newExchange;
                        } else {
                            List oldResults = oldExchange.getIn().getBody(List.class);
                            List newResults = newExchange.getIn().getBody(List.class);
                            ImmutableList aggResult = ImmutableList.copyOf(Iterables.concat(oldResults, newResults));
                            oldExchange.getIn().setBody(aggResult);
                            return oldExchange;
                        }
                    }
                })
                .end()
//                .to("log:bla")

本质上，这个路由接受一个输入，将它发送到 direct:A 和 direct:B，期望来自这两个端点的列表并将它们连接起来(最后一行是有原因的，我稍后会解释).

Essentially, this route takes an input, sends it to direct:A and direct:B, expects lists from these two endpoints and concatenates them (the comment in the last line is there for a reason I will explain later).

现在假设这两个端点返回"分别列出 [A] 和 [B].如果我将消息 M 发送到 direct:multicaster，则使用 oldExchange = null 和 newExchange.in 调用一次聚合器.body=[A]，然后使用 oldExchange.in.body=[A] 和 newExchange.out.body=[B](因为它应该要做).

Now assume that these two endpoints "return" the lists [A] and [B], respectively. If I send the message M to direct:multicaster, then the aggregator is called once with oldExchange = null and newExchange.in.body=[A], then with oldExchange.in.body=[A] and newExchange.out.body=[B] (as it is supposed to do).

到目前为止一切都很好.但是使用 oldExchange.in.body=[A,B] 和 newExchange.in=M 再次调用聚合器(M 是初始消息).这看起来类似于包含的丰富模式.

All good up to this point. But the aggregator is called once more with oldExchange.in.body=[A,B] and newExchange.in=M (M is the initial message). This looks similar to an included enrichment pattern.

您可以通过删除最后一行中的注释来获得预期的行为，即简单地添加一个虚拟的 to("log:bla").有了这个，一切都按预期运行.

You can get the expected behaviour by removing the comment in the last line, i.e. simply adding a dummy to("log:bla"). With this everthing behaves as expected.

            .multicast()
            .aggregationStrategy(aggStrategy)
            .to("direct:A", "direct:B")
            .end()

和

            .multicast(aggStrategy)
            .to("direct:A", "direct:B")
            .end()

两者都导致相同的行为.

both result in the same behaviour.

这里发生了什么 - 我做错了什么?

What is happening here - what did I get wrong?

提前致谢马克斯

推荐答案

我曾尝试重现该问题，但没有成功.这就是我所做的:

I have tried to reproduce the problem, but without success. This is what I did:

路线:

public class MulticastRoute extends RouteBuilder {
    @Override
    public void configure() throws Exception {
        AggregationStrategy myAggregationStrategy = new MyAggregationStrategy();
        List<String> listA = Lists.newArrayList("A");
        List<String> listB = Lists.newArrayList("B");
        from("direct:multicast").routeId("multicastRoute").multicast(myAggregationStrategy).to("direct:A", "direct:B").end();

        from("direct:A").setBody(constant(listA));
        from("direct:B").setBody(constant(listB));
    }

    class MyAggregationStrategy implements AggregationStrategy {
        @Override
        public org.apache.camel.Exchange aggregate(Exchange oldExchange, Exchange newExchange) {
            System.out.println("Aggregate called with oldExchange = " + (oldExchange == null ? "null" :
                    oldExchange.getIn().getBody().toString()) + ", newExchange = " +
                    newExchange.getIn().getBody().toString());
            return newExchange;
        }
    }
}

创建了一个简单的测试来运行路由.

Created a simple test just to run the route.

测试:

public class MulticastRouteTest extends CamelTestSupport {
  @Test
    public void testMulticastRoute() throws Exception {
        context.addRoutes(new MulticastRoute());
        template.sendBody("direct:multicast", null);
    }
}

打印:

Aggregate called with oldExchange = null, newExchange = [A]
Aggregate called with oldExchange = [A], newExchange = [B]

这是我们所期望的.希望这会帮助你.我看不出我做事的方式有什么不同，但希望你能发现.

This is what we would expect. Hope this will help you. I can not see any difference in the way I do things, but hopefully you will spot it.

这篇关于Apache Camel:聚合多播 - AggregationStrategy 调用过于频繁的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！