We Keep Coding

How to merge multiple vertx web client responses

I am new to vertx and async programming.
I have 2 verticles communicating via an event bus as follows:
//API Verticle
public class SearchAPIVerticle extends AbstractVerticle {
public static final String GET_USEARCH_DOCS = "get.usearch.docs";
#Autowired
private Integer defaultPort;
private void sendSearchRequest(RoutingContext routingContext) {
final JsonObject requestMessage = routingContext.getBodyAsJson();
final EventBus eventBus = vertx.eventBus();
eventBus.request(GET_USEARCH_DOCS, requestMessage, reply -> {
if (reply.succeeded()) {
Logger.info("Search Result = " + reply.result().body());
routingContext.response()
.putHeader("content-type", "application/json")
.setStatusCode(200)
.end((String) reply.result().body());
} else {
Logger.info("Document Search Request cannot be processed");
routingContext.response()
.setStatusCode(500)
.end();
}
});
}
#Override
public void start() throws Exception {
Logger.info("Starting the Gateway service (Event Sender) verticle");
// Create a Router
Router router = Router.router(vertx);
//Added bodyhandler so we can process json messages via the event bus
router.route().handler(BodyHandler.create());
// Mount the handler for incoming requests
// Find documents
router.post("/api/search/docs/*").handler(this::sendSearchRequest);
// Create an HTTP Server using default options
HttpServer server = vertx.createHttpServer();
// Handle every request using the router
server.requestHandler(router)
//start listening on port 8083
.listen(config().getInteger("http.port", 8083)).onSuccess(msg -> {
Logger.info("*************** Search Gateway Server started on "
+ server.actualPort() + " *************");
});
}
#Override
public void stop(){
//house keeping
}
}
//Below is the target verticle should be making the multiple web client call and merging the responses
.
#Component
public class SolrCloudVerticle extends AbstractVerticle {
public static final String GET_USEARCH_DOCS = "get.usearch.docs";
#Autowired
private SearchRepository searchRepositoryService;
#Override
public void start() throws Exception {
Logger.info("Starting the Solr Cloud Search Service (Event Consumer) verticle");
super.start();
ConfigStoreOptions fileStore = new ConfigStoreOptions().setType("file")
.setConfig(new JsonObject().put("path", "conf/config.json"));
ConfigRetrieverOptions configRetrieverOptions = new ConfigRetrieverOptions()
.addStore(fileStore);
ConfigRetriever configRetriever = ConfigRetriever.create(vertx, configRetrieverOptions);
configRetriever.getConfig(ar -> {
if (ar.succeeded()) {
JsonObject configJson = ar.result();
EventBus eventBus = vertx.eventBus();
eventBus.<JsonObject>consumer(GET_USEARCH_DOCS).handler(getDocumentService(searchRepositoryService, configJson));
Logger.info("Completed search service event processing");
} else {
Logger.error("Failed to retrieve the config");
}
});
}
private Handler<Message<JsonObject>> getDocumentService(SearchRepository searchRepositoryService, JsonObject configJson) {
return requestMessage -> vertx.<String>executeBlocking(future -> {
try {
//I need to incorporate the logic here that adds futures to list and composes the compositefuture
/*
//Below is my logic to populate the future list
WebClient client = WebClient.create(vertx);
List<Future> futureList = new ArrayList<>();
for (Object collection : searchRepositoryService.findAllCollections(configJson).getJsonArray(SOLR_CLOUD_COLLECTION).getList()) {
Future<String> future1 = client.post(8983, "127.0.0.1", "/solr/" + collection + "/query")
.expect(ResponsePredicate.SC_OK)
.sendJsonObject(requestMessage.body())
.map(HttpResponse::bodyAsString).recover(error -> {
System.out.println(error.getMessage());
return Future.succeededFuture();
});
futureList.add(future1);
}
//Below is the CompositeFuture logic, but the logic and construct does not make sense to me. What goes as first and second argument of executeBlocking method
/*CompositeFuture.join(futureList)
.onSuccess(result -> {
result.list().forEach( x -> {
if(x != null){
requestMessage.reply(result.result());
}
}
);
})
.onFailure(error -> {
System.out.println("We should not fail");
})
*/
future.complete("DAO returns a Json String");
} catch (Exception e) {
future.fail(e);
}
}, result -> {
if (result.succeeded()) {
requestMessage.reply(result.result());
} else {
requestMessage.reply(result.cause()
.toString());
}
});
}
}
I was able to use the org.springframework.web.reactive.function.client.WebClient calls to compose my search result from multiple web client calls, as against using Future<io.vertx.ext.web.client.WebClient> with CompositeFuture.
I was trying to avoid mixing Springboot and Vertx, but unfortunately Vertx CompositeFuture did not work here:
//This method supplies the parameter for the future.complete(..) line in getDocumentService(SearchRepository,JsonObject)
private List<JsonObject> findByQueryParamsAndDataSources(SearchRepository searchRepositoryService,
JsonObject configJson,
JsonObject requestMessage)
throws SolrServerException, IOException {
List<JsonObject> searchResultList = new ArrayList<>();
for (Object collection : searchRepositoryService.findAllCollections(configJson).getJsonArray(SOLR_CLOUD_COLLECTION).getList()) {
searchResultList.add(new JsonObject(doSearchPerCollection(collection.toString(), requestMessage.toString())));
}
return aggregateMultiCollectionSearchResults(searchResultList);
}
public String doSearchPerCollection(String collection, String message) {
org.springframework.web.reactive.function.client.WebClient client =
org.springframework.web.reactive.function.client.WebClient.create();
return client.post()
.uri("http://127.0.0.1:8983/solr/" + collection + "/query")
.contentType(MediaType.APPLICATION_JSON)
.accept(MediaType.APPLICATION_JSON)
.body(BodyInserters.fromValue(message.toString()))
.retrieve()
.bodyToMono(String.class)
.block();
}
private List<JsonObject> aggregateMultiCollectionSearchResults(List<JsonObject> searchList){
//TODO: Search result aggregation
return searchList;
}
My use case is the second verticle should make multiple vertx web client calls and should combine the responses.
If an API call falls, I want to log the error and still continue processing and merging responses from other calls.
Please, any help on how my code above could be adaptable to handle the use case?
I am looking at vertx CompositeFuture, but no headway or useful example seen yet!

What you are looking for can done with Future coordination with a little bit of additional handling:
CompositeFuture.join(future1, future2, future3).onComplete(ar -> {
if (ar.succeeded()) {
// All succeeded
} else {
// All completed and at least one failed
}
});
The join composition waits until all futures are completed, either with a success or a failure.
CompositeFuture.join
takes several futures arguments (up to 6) and returns a future that is succeeded when all the futures are succeeded, and failed when all the futures are completed and at least one of them is failed
Using join you will wait for all Futures to complete, the issue is that if one of them fails you will not be able to obtain response from others as CompositeFuture will be failed. To avoid this you should add Future<T> recover(Function<Throwable, Future<T>> mapper) on each of your Futures in which you should log the error and pass an empty response so that the future does not fail.
Here is short example:
Future<String> response1 = client.post(8887, "localhost", "work").expect(ResponsePredicate.SC_OK).send()
.map(HttpResponse::bodyAsString).recover(error -> {
System.out.println(error.getMessage());
return Future.succeededFuture();
});
Future<String> response2 = client.post(8887, "localhost", "error").expect(ResponsePredicate.SC_OK).send()
map(HttpResponse::bodyAsString).recover(error -> {
System.out.println(error.getMessage());
return Future.succeededFuture();
});
CompositeFuture.join(response2, response1)
.onSuccess(result -> {
result.list().forEach(x -> {
if(x != null) {
System.out.println(x);
}
});
})
.onFailure(error -> {
System.out.println("We should not fail");
});
Edit 1:
Limit for CompositeFuture.join(Future...) is 6 Futures, in the case you need more you can use: CompositeFuture.join(Arrays.asList(future1, future2, future3)); where you can pass unlimited number of futures.

Short circuiting the chain of CompletionStage

I am using Java 8 and I have a chain of CompletionStage that I am trying to run.
I don't want to use join() or get(), I want to explicity complete the CompletionStage.
I am trying to run two database queries, the second has dependency on the result of the first query. I am starting a database transaction using session, running write query1, write query2 and only if both are successful I want to commit the transaction or else roll it back.
The transaction and session are part of Neo4j java API https://neo4j.com/docs/api/java-driver/current/org/neo4j/driver/async/AsyncSession.html#writeTransactionAsync-org.neo4j.driver.async.AsyncTransactionWork-
After running both queries success/failure I want to close the session(a standard database practice)
Here is psuedo code -
DB Session starts transaction
run Write Query1
run Write Query2
if both are successful
commit transaction
else
rollback transaction
close session
What I want to achieve is if query1/query2 fails then it should just rollback transaction and close session.
Query 1 can also throw a CustomException if the result from Query1 is incorrect(less than some threshold). In this case it should rollback transaction. I am rolling back transaction in the exceptionally block for each query.
The happy path works fine in the code below, but when I want to throw CustomException, the Query2 block is not called and even the Completable.allOf is never called.
CompletableFuture<String> firstFuture = new CompletableFuture();
CompletableFuture<String> secondFuture = new CompletableFuture();
CompletableFuture<String> lastFuture = new CompletableFuture();
//Lambda that executes transaction
TransactionWork<CompletionStage<String>> runTransactionWork = transaction -> {
//Write Query1
transaction.runAsync("DB WRITE QUERY1") //Running Write Query 1
.thenCompose(someFunctionThatReturnsCompletionStage)
.thenApply(val -> {
//throw CustomException if value less then threshold
if(val < threshold){
throw new CustomException("Incorrect value found");
}else{
//if value is correct then complete future
firstFuture.complete(val);
}
firstQuery.complete(val);
}).exceptionally(error -> {
//Since failure occured in Query1 want to roll back
transaction.rollbackAsync();
firstFuture.completeExceptionally(error);
throw new RuntimeException("There has been an error in first query " + error.getMessage());
});
//after the first write query is done then run the second write query
firstFuture.thenCompose(val -> transaction.runAsync("DB Write QUERY 2"))
.thenCompose(someFunctionThatReturnsCompletionStage)
.thenApply(val -> {
//if value is correct then complete
secondFuture.complete(val);
}
}).exceptionally(error -> {
//Incase of failure in Query2 want to roll back
transaction.rollbackAsync();
secondFuture.completeExceptionally(error);
throw new RuntimeException("There has been an error in second query " + error.getMessage());
});
//wait for both to complete and then complete the last future
CompletableFuture.allOf(firstFuture, secondFuture)
.handle((empty, ex) -> {
if(ex != null){
lastFuture.completeExceptionally(ex);
}else{
//commit the transaction
transaction.commitAsync();
lastFuture.complete("OK");
}
return lastFuture;
});
return lastFuture;
}
//Create a database session
Session session = driver.session();
//runTransactionWork is lambda that has access to transaction
session.writeTransactionAsync(runTransactionWork)
.handle((val, err) -> {
if(val != null){
session.closeAsync();
//send message to some broker about success
}else{
//fail logic
}
});
How can I achieve short circuiting the exception to ensure the transaction is rolled back and it directly goes to exception block on session.
These are my observations about the code blocks that are called based on different use cases, note these are based on debug points that I have placed in the code -
Happy path - firstFuture(success) -> secondFuture(success) -> LastFuture (success) -> session block success called (works fine)
First Future fail - firstFuture(failed due to exception) -> secondFuture(never called) -> LastFuture(never called) -> session block failure(never called)
Second Future fail - firstFuture(success) -> secondFuture(failed due to exception) -> LastFuture(never called) -> session block failure(never called)
I want #2 and #3 to work as well and the respective transaction should be rolled back and session should be closed.
My question is if is why does the exeption part from handle of allOf does not get called when one of the future completesExceptionally ?

When you throw that CustomException, firstFuture is not completed. As a matter of fact, nothing happens to it. Because it is not completed (successfully), this:
firstFuture.thenCompose...
will not be executed. The documentation of thenCompose says:
When this stage completes normally, the given function is invoked with this stage's result as the argument...
Since this is not the case, that code is obviously not going to be triggered. Because of that, nothing in turn happens to secondFuture either, so CompletableFuture::allOf has to do exactly zero. May be a simplified example will help:
public class CF {
public static void main(String[] args) {
CompletableFuture<Void> one = CompletableFuture.runAsync(CF::db1);
LockSupport.parkNanos(TimeUnit.MILLISECONDS.toNanos(500));
System.out.println(one.isCompletedExceptionally());
CompletableFuture<Void> two = one.thenRun(CF::db2);
System.out.println("first is done : " + FIRST_FUTURE.isDone());
System.out.println("second is done : " + SECOND_FUTURE.isDone());
CompletableFuture.allOf(FIRST_FUTURE, SECOND_FUTURE).thenRun(() -> {
System.out.println("allOf");
});
LockSupport.parkNanos(TimeUnit.MILLISECONDS.toNanos(500));
}
private static final boolean FAIL = true;
private static final CompletableFuture<String> FIRST_FUTURE = new CompletableFuture<>();
private static final CompletableFuture<String> SECOND_FUTURE = new CompletableFuture<>();
private static void db1() {
if(FAIL) {
throw new RuntimeException("failed one");
} else {
FIRST_FUTURE.complete("42");
}
}
private static void db2() {
System.out.println("Running");
SECOND_FUTURE.complete("42");
}
}
If you run this, you will notice that nothing gets printed...
Unfortunately I am not familiar with Neo4j, but you can most probably adjust this example to your needs:
public class CF {
public static void main(String[] args) {
CompletableFuture<Void> one = CompletableFuture.runAsync(CF::db1);
CompletableFuture<Void> terminal =
one.whenComplete((ok, th) -> {
if(th != null || FIRST_FUTURE.isCompletedExceptionally()) {
// no need to schedule the second one, need to rollback whatever the first one did
// transaction.rollbackAsync();
System.out.println("rollback because first one failed");
LAST_FUTURE.completeExceptionally(new RuntimeException("because first one failed"));
} else {
CompletableFuture<Void> two = CompletableFuture.runAsync(CF::db2);
two.whenComplete((ok2, th2) -> {
if(th2 != null || SECOND_FUTURE.isCompletedExceptionally()) {
System.out.println("rollback because second one failed");
// transaction.rollbackAsync();
LAST_FUTURE.completeExceptionally(new RuntimeException("because second one failed"));
} else {
LAST_FUTURE.complete("OK");
}
});
}
});
// simulate that someone will call this
terminal.join();
System.out.println(LAST_FUTURE.join());
}
private static final boolean FAIL_ONE = false;
private static final boolean FAIL_TWO = true;
private static final CompletableFuture<String> FIRST_FUTURE = new CompletableFuture<>();
private static final CompletableFuture<String> SECOND_FUTURE = new CompletableFuture<>();
private static final CompletableFuture<String> LAST_FUTURE = new CompletableFuture<>();
private static void db1() {
if(FAIL_ONE) {
LockSupport.parkNanos(TimeUnit.MILLISECONDS.toNanos(500));
RuntimeException ex = new RuntimeException("failed one");;
FIRST_FUTURE.completeExceptionally(ex);
} else {
FIRST_FUTURE.complete("42");
}
}
private static void db2() {
if(FAIL_TWO) {
LockSupport.parkNanos(TimeUnit.MILLISECONDS.toNanos(500));
RuntimeException ex = new RuntimeException("failed one");;
SECOND_FUTURE.completeExceptionally(ex);
} else {
SECOND_FUTURE.complete("42");
}
}
}

Manually acknowledge Kafka Event A consuming after producing event B

I have a case where I have to consume event A and do some processing, then produce the event B. So my problem is what would happen is the processing crashed and the application couldn't produce B while it consumed already A. My approach is to acknowledge after successfully publishing B, am I correct or should implement another solution for this case?
#KafkaListener(
id = TOPIC_ID,
topics = TOPIC_ID,
groupId = GROUP_ID,
containerFactory = LISTENER_CONTAINER_FACTORY
)
public void listen(List<Message<A>> messages, Acknowledgment acknowledgment) {
try {
final AEvent aEvent = messages.stream()
.filter(message -> null != message.getPayload())
.map(Message::getPayload)
.findFirst()
.get();
processDao.doSomeProcessing() // returns a Mono<Example> by calling an externe API
.subscribe(
response -> {
ProducerRecord<String, BEvent> BEventRecord = new ProducerRecord<>(TOPIC_ID, null, BEvent);
ListenableFuture<SendResult<String, BEvent>> future = kafkaProducerTemplate.send(buildBEvent());
future.addCallback(new ListenableFutureCallback<SendResult<String, BEvent>>() {
#Override
public void onSuccess(SendResult<String, BEvent> BEventSendResult) {
//TODO: do when event published successfully
}
#Override
public void onFailure(Throwable exception) {
exception.printStackTrace();
throw new ExampleException();
}
});
},
error -> {
error.printStackTrace();
throw new ExampleException();
}
);
acknowledgment.acknowledge(); // ??
} catch (ExampleException) {
exception.printStackTrace();
}
}

You can't manage kafka "acknowledgments" when using async code such as reactor.
Kafka does not manage discrete acks for each topic/partition, just the last committed offset for the partition.
If you process two records asynchronously, you will have a race as to which offset will be committed first.
You need to perform the sends on the listener container thread to maintain proper ordering.

Vert.x request handler and blocking db queries

How can i translate this piece of code correctly in vert.x?
normally, in spring or in a simple sevlet with a template engine to output an html response i'll do like this
function test(request, response) {
templatecontext tc = getContext();
init conditions
if (condition1) {
retrieve data from db ({
asyncresult -> {
tc.put("data1", data1)
})
} else if (condition2) {
other code
if (condition 2.1) {
retrieve data from db ({
asyncresult -> {
tc.put("data2", data2)
})
}
}
get other data from db and put in context
template.eval("templatefile", tc)
write to response
}
the problem is, retrieving data from database is an handler of asyncresult, so i cannot grant that template evaluation is made with data1 or data2, because retrieving of async without falling in a callback hell.
I've not really understood rxjava2, but i feel i'm trying to kill a bean with a spoon.

You can use futures and composition. See the ComposeExample in vertx-examples repo:
public class ComposeExample extends AbstractVerticle {
#Override
public void start() throws Exception {
Future<String> future = anAsyncAction();
future.compose(this::anotherAsyncAction)
.setHandler(ar -> {
if (ar.failed()) {
System.out.println("Something bad happened");
ar.cause().printStackTrace();
} else {
System.out.println("Result: " + ar.result());
}
});
}
private Future<String> anAsyncAction() {
Future<String> future = Future.future();
// mimic something that take times
vertx.setTimer(100, l -> future.complete("world"));
return future;
}
private Future<String> anotherAsyncAction(String name) {
Future<String> future = Future.future();
// mimic something that take times
vertx.setTimer(100, l -> future.complete("hello " + name));
return future;
}
}

Retry logic with CompletableFuture

I need to submit a task in an async framework I'm working on, but I need to catch for exceptions, and retry the same task multiple times before "aborting".
The code I'm working with is:
int retries = 0;
public CompletableFuture<Result> executeActionAsync() {
// Execute the action async and get the future
CompletableFuture<Result> f = executeMycustomActionHere();
// If the future completes with exception:
f.exceptionally(ex -> {
retries++; // Increment the retry count
if (retries < MAX_RETRIES)
return executeActionAsync(); // <--- Submit one more time
// Abort with a null value
return null;
});
// Return the future
return f;
}
This currently doesn't compile because the return type of the lambda is wrong: it expects a Result, but the executeActionAsync returns a CompletableFuture<Result>.
How can I implement this fully async retry logic?

Chaining subsequent retries can be straight-forward:
public CompletableFuture<Result> executeActionAsync() {
CompletableFuture<Result> f=executeMycustomActionHere();
for(int i=0; i<MAX_RETRIES; i++) {
f=f.exceptionally(t -> executeMycustomActionHere().join());
}
return f;
}
Read about the drawbacks below
This simply chains as many retries as intended, as these subsequent stages won’t do anything in the non-exceptional case.
One drawback is that if the first attempt fails immediately, so that f is already completed exceptionally when the first exceptionally handler is chained, the action will be invoked by the calling thread, removing the asynchronous nature of the request entirely. And generally, join() may block a thread (the default executor will start a new compensation thread then, but still, it’s discouraged). Unfortunately, there is neither, an exceptionallyAsync or an exceptionallyCompose method.
A solution not invoking join() would be
public CompletableFuture<Result> executeActionAsync() {
CompletableFuture<Result> f=executeMycustomActionHere();
for(int i=0; i<MAX_RETRIES; i++) {
f=f.thenApply(CompletableFuture::completedFuture)
.exceptionally(t -> executeMycustomActionHere())
.thenCompose(Function.identity());
}
return f;
}
demonstrating how involved combining “compose” and an “exceptionally” handler is.
Further, only the last exception will be reported, if all retries failed. A better solution should report the first exception, with subsequent exceptions of the retries added as suppressed exceptions. Such a solution can be build by chaining a recursive call, as hinted by Gili’s answer, however, in order to use this idea for exception handling, we have to use the steps to combine “compose” and “exceptionally” shown above:
public CompletableFuture<Result> executeActionAsync() {
return executeMycustomActionHere()
.thenApply(CompletableFuture::completedFuture)
.exceptionally(t -> retry(t, 0))
.thenCompose(Function.identity());
}
private CompletableFuture<Result> retry(Throwable first, int retry) {
if(retry >= MAX_RETRIES) return CompletableFuture.failedFuture(first);
return executeMycustomActionHere()
.thenApply(CompletableFuture::completedFuture)
.exceptionally(t -> { first.addSuppressed(t); return retry(first, retry+1); })
.thenCompose(Function.identity());
}
CompletableFuture.failedFuture is a Java 9 method, but it would be trivial to add a Java 8 compatible backport to your code if needed:
public static <T> CompletableFuture<T> failedFuture(Throwable t) {
final CompletableFuture<T> cf = new CompletableFuture<>();
cf.completeExceptionally(t);
return cf;
}

Instead of implementing your own retry logic, I recommend using a proven library like failsafe, which has built-in support for futures (and seems more popular than guava-retrying). For your example, it would look something like:
private static RetryPolicy retryPolicy = new RetryPolicy()
.withMaxRetries(MAX_RETRIES);
public CompletableFuture<Result> executeActionAsync() {
return Failsafe.with(retryPolicy)
.with(executor)
.withFallback(null)
.future(this::executeMycustomActionHere);
}
Probably you should avoid .withFallback(null) and just have let the returned future's .get() method throw the resulting exception so the caller of your method can handle it specifically, but that's a design decision you'll have to make.
Other things to think about include whether you should retry immediately or wait some period of time between attempts, any sort of recursive backoff (useful when you're calling a web service that might be down), and whether there are specific exceptions that aren't worth retrying (e.g. if the parameters to the method are invalid).

I think I was successfully. Here's an example class I created and the test code:
RetriableTask.java
public class RetriableTask
{
protected static final int MAX_RETRIES = 10;
protected int retries = 0;
protected int n = 0;
protected CompletableFuture<Integer> future = new CompletableFuture<Integer>();
public RetriableTask(int number) {
n = number;
}
public CompletableFuture<Integer> executeAsync() {
// Create a failure within variable timeout
Duration timeoutInMilliseconds = Duration.ofMillis(1*(int)Math.pow(2, retries));
CompletableFuture<Integer> timeoutFuture = Utils.failAfter(timeoutInMilliseconds);
// Create a dummy future and complete only if (n > 5 && retries > 5) so we can test for both completion and timeouts.
// In real application this should be a real future
final CompletableFuture<Integer> taskFuture = new CompletableFuture<>();
if (n > 5 && retries > 5)
taskFuture.complete(retries * n);
// Attach the failure future to the task future, and perform a check on completion
taskFuture.applyToEither(timeoutFuture, Function.identity())
.whenCompleteAsync((result, exception) -> {
if (exception == null) {
future.complete(result);
} else {
retries++;
if (retries >= MAX_RETRIES) {
future.completeExceptionally(exception);
} else {
executeAsync();
}
}
});
// Return the future
return future;
}
}
Usage
int size = 10;
System.out.println("generating...");
List<RetriableTask> tasks = new ArrayList<>();
for (int i = 0; i < size; i++) {
tasks.add(new RetriableTask(i));
}
System.out.println("issuing...");
List<CompletableFuture<Integer>> futures = new ArrayList<>();
for (int i = 0; i < size; i++) {
futures.add(tasks.get(i).executeAsync());
}
System.out.println("Waiting...");
for (int i = 0; i < size; i++) {
try {
CompletableFuture<Integer> future = futures.get(i);
int result = future.get();
System.out.println(i + " result is " + result);
} catch (Exception ex) {
System.out.println(i + " I got exception!");
}
}
System.out.println("Done waiting...");
Output
generating...
issuing...
Waiting...
0 I got exception!
1 I got exception!
2 I got exception!
3 I got exception!
4 I got exception!
5 I got exception!
6 result is 36
7 result is 42
8 result is 48
9 result is 54
Done waiting...
Main idea and some glue code (failAfter function) come from here.
Any other suggestions or improvement are welcome.

util class:
public class RetryUtil {
public static <R> CompletableFuture<R> retry(Supplier<CompletableFuture<R>> supplier, int maxRetries) {
CompletableFuture<R> f = supplier.get();
for(int i=0; i<maxRetries; i++) {
f=f.thenApply(CompletableFuture::completedFuture)
.exceptionally(t -> {
System.out.println("retry for: "+t.getMessage());
return supplier.get();
})
.thenCompose(Function.identity());
}
return f;
}
}
usage:
public CompletableFuture<String> lucky(){
return CompletableFuture.supplyAsync(()->{
double luckNum = Math.random();
double luckEnough = 0.6;
if(luckNum < luckEnough){
throw new RuntimeException("not luck enough: " + luckNum);
}
return "I'm lucky: "+luckNum;
});
}
#Test
public void testRetry(){
CompletableFuture<String> retry = RetryUtil.retry(this::lucky, 10);
System.out.println("async check");
String join = retry.join();
System.out.println("lucky? "+join);
}
output
async check
retry for: java.lang.RuntimeException: not luck enough: 0.412296354211683
retry for: java.lang.RuntimeException: not luck enough: 0.4099777199676573
lucky? I'm lucky: 0.8059089479049389

I recently solved a similar problem using the guava-retrying library.
Callable<Result> callable = new Callable<Result>() {
public Result call() throws Exception {
return executeMycustomActionHere();
}
};
Retryer<Boolean> retryer = RetryerBuilder.<Result>newBuilder()
.retryIfResult(Predicates.<Result>isNull())
.retryIfExceptionOfType(IOException.class)
.retryIfRuntimeException()
.withStopStrategy(StopStrategies.stopAfterAttempt(MAX_RETRIES))
.build();
CompletableFuture.supplyAsync( () -> {
try {
retryer.call(callable);
} catch (RetryException e) {
e.printStackTrace();
} catch (ExecutionException e) {
e.printStackTrace();
}
});

Here is an approach that will work for any CompletionStage subclass and does not return a dummy CompletableFuture that does nothing more than wait to get updated by other futures.
/**
* Sends a request that may run as many times as necessary.
*
* #param request a supplier initiates an HTTP request
* #param executor the Executor used to run the request
* #return the server response
*/
public CompletionStage<Response> asyncRequest(Supplier<CompletionStage<Response>> request, Executor executor)
{
return retry(request, executor, 0);
}
/**
* Sends a request that may run as many times as necessary.
*
* #param request a supplier initiates an HTTP request
* #param executor the Executor used to run the request
* #param tries the number of times the operation has been retried
* #return the server response
*/
private CompletionStage<Response> retry(Supplier<CompletionStage<Response>> request, Executor executor, int tries)
{
if (tries >= MAX_RETRIES)
throw new CompletionException(new IOException("Request failed after " + MAX_RETRIES + " tries"));
return request.get().thenComposeAsync(response ->
{
if (response.getStatusInfo().getFamily() != Response.Status.Family.SUCCESSFUL)
return retry(request, executor, tries + 1);
return CompletableFuture.completedFuture(response);
}, executor);
}

maybe it's late but hopes someone might find this useful, I recently solved this problem for retrying rest API call on failure. In my case, I have to retry on 500 HTTP status code, below is my rest client code (we are using WSClient from play framework) you can change it to whatever rest-client as per requirement.
int MAX_RETRY = 3;
CompletableFuture<WSResponse> future = new CompletableFuture<>();
private CompletionStage<WSResponse> getWS(Object request,String url, int retry, CompletableFuture future) throws JsonProcessingException {
ws.url(url)
.post(Json.parse(mapper.writeValueAsString(request)))
.whenCompleteAsync((wsResponse, exception) -> {
if(wsResponse.getStatus() == 500 && retry < MAX_RETRY) {
try {
getWS(request, retry+1, future);
} catch (IOException e) {
throw new Exception(e);
}
}else {
future.complete(wsResponse);
}
});
return future;
}
This code will return immediately if the status code is 200 or other than 500 whereas if HTTP status is 500 it will retry 3 times.

Inspired by theazureshadow's answer. His or her answer was great but doesn't work with new version of FailSafe. The below code works with
<dependency>
<groupId>dev.failsafe</groupId>
<artifactId>failsafe</artifactId>
<version>3.3.0</version>
</dependency>
solution:
RetryPolicy<Object> retryPolicy = RetryPolicy.builder()
.withMaxRetries(MAX_RETRY)
.withBackoff(INITIAL_DELAY, MAX_DELAY, ChronoUnit.SECONDS)
.build();
Fallback<Object> fallback = Fallback.of((AuditEvent) null);
public CompletableFuture<Object> executeAsync(Runnable asyncTask) {
return Failsafe.with(fallback)
.compose(retryPolicy)
.with(executorService)
.onFailure(e -> LOG.error(e.getException().getMessage()))
.getAsync(() -> asyncTask());
}

We needed to retry a task based on an error condition.
public static <T> CompletableFuture<T> retryOnCondition(Supplier<CompletableFuture<T>> supplier,
Predicate<Throwable> retryPredicate, int maxAttempts) {
if (maxAttempts <= 0) {
throw new IllegalArgumentException("maxAttempts can't be <= 0");
}
return retryOnCondition(supplier, retryPredicate, null, maxAttempts);
}
private static <T> CompletableFuture<T> retryOnCondition(
Supplier<CompletableFuture<T>> supplier, Predicate<Throwable> retryPredicate,
Throwable lastError, int attemptsLeft) {
if (attemptsLeft == 0) {
return CompletableFuture.failedFuture(lastError);
}
return supplier.get()
.thenApply(CompletableFuture::completedFuture)
.exceptionally(error -> {
boolean doRetry = retryPredicate.test(error);
int attempts = doRetry ? attemptsLeft - 1 : 0;
return retryOnCondition(supplier, retryPredicate, error, attempts);
})
.thenCompose(Function.identity());
}
Usage:
public static void main(String[] args) {
retryOnCondition(() -> myTask(), e -> {
//log exception
return e instanceof MyException;
}, 3).join();
}

I would suggest using resilience4j for this use case. It's very handy!!
Introduction: resilience4j-retry and its Javadoc: Retry
They have method to decorate completionStage directly as below:
default <T> java.util.concurrent.CompletionStage<T> executeCompletionStage​(java.util.concurrent.ScheduledExecutorService scheduler,
java.util.function.Supplier<java.util.concurrent.CompletionStage<T>> supplier)