An example of how to do a fire and forget pattern in java react

AsyncCacheUpdate.java

  public Flux<Integer> testCacheAsync() {
    // flux of 1-10 integers with a delay of 100millis between each element
    Flux<Integer> integerFlux = Flux.range(1, 10).delayElements(Duration.ofMillis(200));

    /**
     * A forget and fire pattern event here. which is executed after collectList() i.e. after the
     * flux stream has it's last event collected. After which you can run a blocking task on
     * collected list. This is essentially another subscriber on integerFlux
     */
    integerFlux
        .collectList()
        .doOnNext(
            data ->
                Mono.fromRunnable(() -> blockingTask(data)).subscribeOn(TEST_SCHEDULER).subscribe())
        .subscribe();

    /**
     * Showcasing buffer which is similar to collectList(), but collect with max size
     */
    integerFlux
        .buffer(2)
        .doOnNext(
            data ->
                Mono.fromRunnable(() -> blockingBufferTask(data))
                    .subscribeOn(TEST_SCHEDULER)
                    .subscribe())
        .subscribe();

    // return integer flux which can have other subscribers in the response path
    return integerFlux;
  }

Replace this:
Flux<Integer> integerFlux = Flux.range(1, 10).delayElements(Duration.ofMillis(200));
with this:
Flux<Integer> integerFlux = Flux.range(1, 10).doOnNext(System.out::println).delayElements(Duration.ofMillis(200));

You'll see that each number is printed twice (or three times if the returned Flux is also subscribed) as you have two subscribe calls.

That's why you might need caching.