apache camel - parallel processor then join output

Question

I hope to do a parallel processing of two processors (fetching different info from different sources) then when both completed, i wanted to have access to both output for further processing (e.g. comparisons).

Something of sort:

from("direct:start)
            .processor("process1")
            .processor("process2")
      .to("direct:compare");

Except I need both output from process1 and process2 to be available in "compare" endpoint.

score 1 · Accepted Answer · answered Jul 15 '19 at 12:35

This is one way to achieve using multicast and aggregation strategy,

public class App {
  public static void main(String[] args) throws Exception {

    CamelContext context = new DefaultCamelContext();
    context.addRoutes(myRoute());
    context.startRoute("start");
    context.start();
    ProducerTemplate producerTemplate = context.createProducerTemplate();
    producerTemplate.sendBody("direct:start", null);
    Thread.sleep(10_000);
    context.stop();

  }

  private static RouteBuilder myRoute() {
    return new RouteBuilder() {
      @Override
      public void configure() throws Exception {
        from("direct:start").routeId("start")
                .multicast(new MyAggregationStrategy())
                .parallelProcessing()
                .to("direct:process1", "direct:process2", "direct:process3")
                .end()
        .to("direct:endgame");

        from("direct:process1")
                .process(e -> {
                  ArrayList<String> body = Lists.newArrayList("a", "b", "c");
                  e.getIn().setBody(body);
                });

        from("direct:process2")
                .process(e -> {
                  ArrayList<String> body = Lists.newArrayList("1", "2", "3");
                  e.getIn().setBody(body);
                });

        from("direct:process3")
                .process(e -> {
                  ArrayList<String> body = Lists.newArrayList("@", "#", "$");
                  e.getIn().setBody(body);
                });


        from("direct:endgame")
                .process(e -> {
                  log.info(" This final result : " + e.getIn().getBody());
                });
      }
    };
  }
}

//This is where we can aggregate results of the process which is running in parallel
class MyAggregationStrategy implements AggregationStrategy {

  @Override
  public Exchange aggregate(Exchange oldExchange, Exchange newExchange) {
    ArrayList<Object> objects = Lists.newArrayList();
    if (oldExchange == null) {
      return newExchange;
    }

    Object o = oldExchange.getIn().getBody();
    Object n = newExchange.getIn().getBody();

    objects.add(o);
    objects.add(n);

    newExchange.getIn().setBody(objects);

    return newExchange;
  }
}

You can use `GroupedBodyAggregationStrategy` instead implementing the custom one — Bedla, Jul 15 '19 at 13:02
@SunandPadmanabhan, so i have observed this is working fine with "direct", but when I changed to "seda" as it is expected to have multiple producer triggering start at the same time (and they should all run on each own thread), "endgame" endpoint is being called immediately (with zero list exchange) even though "process1/2/3" hasn't finished their job yet. any ideas? — lorraine batol, Jul 18 '19 at 03:30
I am not sure, I have to try it out myself. If you can paste the code snippet which you are trying to achieve that will be helpful. — Sunand Padmanabhan, Jul 20 '19 at 13:24
@SunandPadmanabhan here it is https://stackoverflow.com/questions/57087489/apache-camel-seda-endpoint-multicast — lorraine batol, Jul 20 '19 at 17:17

apache camel - parallel processor then join output

1 Answers1