Akka "Best Effort Retry" with RxJava - akka

Question

I am curious about using RxJava to implement best effort retry in Akka, without Persistent Actors. The idea is to use Rx's retry method to keep asking until a response is received from the destination actor.
Other examples of this is hard to find. Are there any Akka gurus out there that could verify this implementation, or point me to a better solution?
Example:
public class RxWithAkka {
private final Logger LOGGER = LoggerFactory.getLogger(getClass());
public static final Timeout TIMEOUT = Timeout.apply(10, TimeUnit.MILLISECONDS);
private final ActorRef actor;
private final ActorSystem actorSystem;
public RxWithAkka(ActorSystem actorSystem) {
this.actorSystem = actorSystem;
this.actor = actorSystem.actorOf(Props.create(MyActor.class));
}
public Observable<Object> ping() {
return createObservable()
.doOnError(t -> LOGGER.warn(t.getMessage()))
.retry();
}
Observable<Object> createObservable() {
return Observable.create(subscriber -> {
LOGGER.info("Send ping");
Patterns.ask(actor, "ping", TIMEOUT)
.onComplete(new OnComplete<Object>() {
#Override
public void onComplete(Throwable failure, Object success) throws Throwable {
if (success != null) {
subscriber.onNext(success);
subscriber.onCompleted();
} else {
subscriber.onError(failure);
}
}
}, actorSystem.dispatcher());
});
}
}
Example actor to demonstrate message not received and timeout:
public class MyActor extends UntypedActor {
private int counter = 0;
#Override
public void onReceive(Object message) throws Exception {
switch (counter++) {
case 0:
// ignore message
break;
case 1:
// timeout
Thread.sleep(200);
break;
default:
getSender().tell("pong", getSelf());
}
}
}
}
Test:
public class RxWithAkkaTest {
#Test
public void testIt() throws Exception {
ActorSystem system = ActorSystem.create("system");
RxWithAkka example = new RxWithAkka(system);
String res = (String) example.ping().toBlocking().first();
assertThat(res).isEqualTo("pong");
}
}

Answer 1

In RxJava, you can use the timeout operator in conjunction with retry.