Summary of what I want to achieve:
I want to execute N tasks in parallel such that no individual task should run for more than two seconds (we can mark such tasks as failed). As an output I want to return the output of successful tasks and status of failed tasks as failed. Also a timeout of one task should not lead to a circuit break, i.e., other tasks execution should not stop.
Note: I am restricted to use Java 8.
I referenced this article for parallel processing. I am doing a similar kind of parallel processing as given in the example in this article:
public void parallelProcessing() {
try {
ExecutorService executorService = Executors.newWorkStealingPool(10);
List<CompletableFuture<Integer>> futuresList = new ArrayList<CompletableFuture<Integer>>();
futuresList.add(CompletableFuture.supplyAsync(()->(addFun1(10, 5)), executorService));
futuresList.add(CompletableFuture.supplyAsync(()->(subFun1(10, 5)), executorService));
futuresList.add(CompletableFuture.supplyAsync(()->(mulFun1(10, 5)), executorService));
CompletableFuture<Void> allFutures = CompletableFuture.allOf(futuresList.toArray(new CompletableFuture[futuresList.size()]));
CompletableFuture<List<Integer>> allCompletableFuture = allFutures.thenApply(future -> futuresList.stream().map(completableFuture -> completableFuture.join())
.collect(Collectors.toList()));
CompletableFuture<List<Integer>> completableFuture = allCompletableFuture.toCompletableFuture();
List<Integer> finalList = (List<Integer>) completableFuture.get();
} catch (Exception ex) {
}
}
public static Integer addFun1(int a, int b) {
System.out.println(Thread.currentThread().getName());
for (int i = 0; i < 10; i++) {
System.out.print(Thread.currentThread().getName() + i);
}
return a + b;
}
public static Integer subFun1(int a, int b) {
System.out.println(Thread.currentThread().getName());
for (int i = 0; i < 10; i++) {
System.out.print(Thread.currentThread().getName() + i);
}
return a - b;
}
public static Integer mulFun1(int a, int b) {
System.out.println(Thread.currentThread().getName());
for (int i = 0; i < 10; i++) {
System.out.print(Thread.currentThread().getName() + i);
}
return a * b;
}
This works fine. But I want to set a timeout for an individual thread. I know I can use an overloaded get function in the last line. But that would set the timeout for combined futures, right? E.g., if I want no individual thread should be blocked for more than 2 seconds, and if I set a 2 seconds timeout in the last line, it will be combined timeout, right?
get(long timeout, TimeUnit unit)
Here's what I want to achieve as a final outcome:
Suppose there are five threads and four complete on time, one timeout (due to running more than two seconds). In this case, I want to send the output of four threads and send the error for the fifth thread in the result.
My input/output format is in the following way:
Sample input: List<Input> each item is run in a separate thread, where each input has a uniqueIdentifier.
Sample output: List<Output> such that:
Output :{
uniqueIdentifier: // Same as input to map for which input this output was generated
result: success/fail // This Field I want to add. Currently it's not there
data: {
// From output, e.g., addFun1 and subFun1
}
}
The semantics of what you want to achieve matter very much. On one hand, you say that you want an alternative for orTimeout for Java 8; on the other hand you kind of imply that you want to drop execution of a certain CompletableFuture if it goes beyond a certain threshold.
These are very different things, because orTimeout says in the documentation:
Exceptionally completes this CompletableFuture with a TimeoutException if not otherwise completed before the given timeout.
So something like:
CompletableFuture<Integer> addAsy =
supplyAsync(() -> addFun1(10,5), executorService)
.orTimeout(5, TimeUnit.MILLISECONDS);
will result in a CompletableFuture that is completed exceptionally (assuming that addFun1 takes more than 5 ms). At the same time, this:
CompletableFuture<Void> allFutures = CompletableFuture
.allOf(futuresList.toArray(new CompletableFuture[0]));
as the documentation states in the allOf:
... If any of the given CompletableFutures complete exceptionally, then the returned CompletableFuture also does so...
means that allFutures is a CompletableFuture that is completed exceptionally too (because addAsy is).
Now, because you have this:
CompletableFuture<List<Integer>> allCompletableFuture = allFutures.thenApply(future -> {
return futuresList.stream().map(CompletableFuture::join)
.collect(Collectors.toList());
});
And again, the documentation of thenApply says:
Returns a new CompletionStage that, when this stage completes normally, is executed with this stage's result as the argument to the supplied function.
Your allFutures did not complete normally, as such this is not even called.
So you need to understand what exactly you want to achieve. For a backport of orTimeout you could start by looking here.
You still need some kind of a backport for orTimeout. I will use the method as if it already exists.
static void parallelProcessing() throws Exception {
ExecutorService executorService = Executors.newFixedThreadPool(10);
List<CompletableFuture<Integer>> futuresList = new ArrayList<>();
futuresList.add(CompletableFuture.supplyAsync(() -> addFun1(10,5), executorService).orTimeout(2, TimeUnit.SECONDS));
futuresList.add(CompletableFuture.supplyAsync(() -> subFun1(10,5), executorService));
futuresList.add(CompletableFuture.supplyAsync(() -> mulFun1(10,5), executorService));
CompletableFuture<Void> all = CompletableFuture.allOf(futuresList.toArray(new CompletableFuture[0]));
Map<Boolean, List<CompletableFuture<Integer>>> map =
all.thenApply(x -> both(futuresList)).exceptionally(x -> both(futuresList)).get();
List<CompletableFuture<Integer>> failed = map.get(Boolean.TRUE);
List<CompletableFuture<Integer>> ok = map.get(Boolean.FALSE);
System.out.println("failed = " + failed.size());
System.out.println("ok = " + ok.size());
}
private static Map<Boolean, List<CompletableFuture<Integer>>> both(
List<CompletableFuture<Integer>> futuresList) {
return futuresList.stream().collect(Collectors.partitioningBy(
CompletableFuture::isCompletedExceptionally
));
}
The following is a single-file mre (paste the entire code into RunParallelTasks.java and run). It is a prototype of the structure I suggested in my comment aimed to achieve the required functionality by using simple means:
import java.util.Optional;
public class RunParallelTasks {
public static void main(String[] args) {
new Thread(()->{
long duration = 3000;
Callback<Long> cb = new LongTask(duration);
Output<Long> output = new TaskExecuter<Long>().work(cb);
System.out.println( output);
}).start();
new Thread(()->{
long duration = 300;
Callback<Long> cb = new LongTask(duration);
Output<Long> output = new TaskExecuter<Long>().work(cb);
System.out.println( output);
}).start();
new Thread(()->{
long duration = 4000;
Callback<Long> cb = new LongTask(duration);
Output<Long> output = new TaskExecuter<Long>().work(cb);
System.out.println( output);
}).start();
new Thread(()->{
long duration = 1000;
Callback<Long> cb = new LongTask(duration);
Output<Long> output = new TaskExecuter<Long>().work(cb);
System.out.println( output);
}).start();
}
}
class TaskExecuter<T>{
private static final long TIMEOUT = 2000;//millis
private T value = null;
public Output<T> work(Callback<T> call){
Thread t = new Thread(()->{
value = call.work();
});
t.start();
try {
t.join(TIMEOUT);
} catch (InterruptedException ex) {
ex.printStackTrace();
}
return new Output<>(t.getId(), value == null ? Optional.empty() : Optional.of(value)) ;
}
}
interface Callback<T> {
T work();
}
class LongTask implements Callback<Long>{
private final long durationInMillis;
public LongTask(long durationInMillis) {
this.durationInMillis = durationInMillis;
}
#Override
public Long work() {
try {
Thread.sleep(durationInMillis);
} catch (InterruptedException ex) {
ex.printStackTrace();
}
return durationInMillis;
}
}
class Output<T> {
private final long id;
private boolean success = false;
private T data;
public Output(long id, Optional<T> op) {
this.id = id;
if(!op.isEmpty()) {
data = op.get();
success = true;
}
}
//todo add getters
#Override
public String toString() {
return "task "+ id+ (success ? " Completed, returned "+data : " Failed" );
}
}
I believe that in order to achieve canceling a task if execution takes too long, you need two tasks:
The task itself doing the computation
Another task, which takes care to cancel the first, if it takes too long
This is inspired by my answer here, at least for now i still have not come up with a better way to do it.
Let's say this is Output:
public class Output {
private final String uniqueIdentifier;
private final boolean result;
private final Object data;
//all arguments constructor and getters
#Override
public String toString() {
return "Output{" +
"uniqueIdentifier='" + uniqueIdentifier + '\'' +
", result=" + result +
", data=" + data +
'}';
}
}
For simplicity i'll use only add integers task from your example, wrapping it in a Supplier.
public class AddIntegerTask implements Supplier<Integer> {
private static final long NANOSECONDS_IN_SECOND = 1_000_000_000;
private final String uniqueIdentifier;
private final boolean tooLong;
private final int a;
private final int b;
public AddIntegerTask(boolean tooLong, int a, int b) {
this.uniqueIdentifier = UUID.randomUUID().toString();
this.tooLong = tooLong;
this.a = a;
this.b = b;
}
#Override
public Integer get() {
long nanoseconds = this.tooLong ? 3 * NANOSECONDS_IN_SECOND : NANOSECONDS_IN_SECOND;
long start = System.nanoTime();
long toEnd = start + nanoseconds;
//simulate long execution
while (System.nanoTime() <= toEnd) {
//check for interruption at crucial points
boolean interrupted = Thread.currentThread().isInterrupted();
if (interrupted) {
//custom exception extending RuntimeException
throw new TooLongExecutionException();
}
}
return this.a + this.b;
}
public String getUniqueIdentifier() {
return this.uniqueIdentifier;
}
}
Most important here is, that you need to check the current thread for interruption at key moments in your own implementation.
The cancel task is quite straightforward:
public class CancelTask implements Runnable {
private final Future<?> future;
public CancelTask(Future<?> future) {
this.future = future;
}
#Override
public void run() {
this.future.cancel(true);
}
}
Wrap the canceling of a Future in a Runnable, so it can be scheduled for execution with approproate delay.
And the Runnable, which will wrap the result in an Output, and will be submitted for execution:
public class MyRunnable implements Runnable {
private final Map<String, Output> outputMap;
private final AddIntegerTask calcFunction;
private final CountDownLatch latch;
public MyRunnable(Map<String, Output> outputMap, AddIntegerTask calcFunction, CountDownLatch latch) {
this.outputMap = outputMap;
this.calcFunction = calcFunction;
this.latch = latch;
}
#Override
public void run() {
String uniqueIdentifier = this.calcFunction.getUniqueIdentifier();
Output output;
try {
Integer result = this.calcFunction.get();
output = new Output(uniqueIdentifier, true, result);
} catch (TooLongExecutionException exc) {
output = new Output(uniqueIdentifier, false, null);
}
this.outputMap.replace(uniqueIdentifier, output);
this.latch.countDown();
}
}
Things to note here: CountDownLatch, it looks to me that you know the number of tasks beforehand, so it's a good choice to force main thread to wait until all tasks have finished. TooLongExecutionException is a custom exception extending RuntimeException. If the job completes, set as successful with result, if it was interrupted set to not successful without result.
And a main to combine and test all that:
public class CancelingMain {
public static void main(String[] args) throws InterruptedException {
ScheduledExecutorService executorService = Executors.newScheduledThreadPool(10);
int taskCount = 6;
CountDownLatch latch = new CountDownLatch(taskCount);
long start = System.nanoTime();
Map<String, Output> outputMap = new LinkedHashMap<>();
for (int i = 1; i <= taskCount; i++) {
boolean tooLong = i % 2 == 0;
AddIntegerTask task = new AddIntegerTask(tooLong, 10, 7);
outputMap.put(task.getUniqueIdentifier(), null);
MyRunnable runnable = new MyRunnable(outputMap, task, latch);
Future<?> future = executorService.submit(runnable);
//schedule cancel task to run once, 2 seconds after scheduling
executorService.schedule(new CancelTask(future), 2, TimeUnit.SECONDS);
}
latch.await();
System.out.println("execution took - " + (System.nanoTime() - start) / 1_000_000_000D);
executorService.shutdown();
outputMap.values().forEach(System.out::println);
}
}
I am using LinkedHashMap in order to keep the tasks in their order of submission.
Here's a fleshed out version of what I suggested in a comment to the question. The idea is to wrap a call to get(long timeout, TimeUnit unit) into another future. I encapsulate the required logic into a BetterFuture class, which delegates to a CompletableFuture under the hood:
import static java.util.concurrent.CompletableFuture.completedFuture;
import static java.util.concurrent.CompletableFuture.runAsync;
import static java.util.stream.Stream.concat;
import java.time.Duration;
import java.util.Optional;
import java.util.concurrent.Callable;
import java.util.concurrent.CompletableFuture;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.Executor;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.TimeoutException;
import java.util.function.Function;
import java.util.stream.Stream;
public class BetterFuture<T> {
private final CompletableFuture<T> delegate;
private BetterFuture(CompletableFuture<T> delegate) {
this.delegate = delegate;
}
public static <T> BetterFuture<T> completed(T value) {
return new BetterFuture<>(completedFuture(value));
}
public static <T> BetterFuture<T> future(Executor executor, Callable<T> callable) {
CompletableFuture<T> delegate = new CompletableFuture<T>();
runAsync(() -> {
try {
delegate.complete(callable.call());
} catch (Throwable e) {
delegate.completeExceptionally(e);
}
}, executor);
return new BetterFuture<>(delegate);
}
public static <T> BetterFuture<Optional<T>> future(Executor executor, Callable<T> callable, Duration timeout) {
return future(executor, () -> future(executor, callable).get(timeout));
}
public <R> BetterFuture<R> map(Function<T, R> fn) {
return new BetterFuture<>(delegate.thenApply(fn));
}
public <R> BetterFuture<R> andThen(Function<T, BetterFuture<R>> fn) {
return new BetterFuture<>(
delegate.thenCompose(value -> fn.apply(value).delegate));
}
public static <T> BetterFuture<Stream<T>> collect(Stream<BetterFuture<T>> futures) {
return futures
.map(future -> future.map(Stream::of))
.reduce(
BetterFuture.completed(Stream.empty()),
(future1, future2) ->
future1
.andThen(stream1 ->
future2
.map(stream2 ->
concat(stream1, stream2)))
);
}
public T get() throws ExecutionException, InterruptedException {
return delegate.get();
}
public Optional<T> get(Duration timeout) throws ExecutionException, InterruptedException {
try {
return Optional.of(delegate.get(timeout.toMillis(), TimeUnit.MILLISECONDS));
} catch (TimeoutException e) {
return Optional.empty();
}
}
}
Most of the methods just delegate to the underlying CompletableFuture without adding much additional functionality.
To start an async task with a timeout, use the method
<T> BetterFuture<Optional<T>> future(Executor executor, Callable<T> callable, Duration timeout)
If a timeout occurs, it completes with empty and with an optional of T otherwise.
In addition the method
public static <T> BetterFuture<Stream<T>> collect(Stream<BetterFuture<T>> futures)
provides a convenient way for collecting a stream of futures into a future of a stream of the same type:
Stream<BetterFuture<Optional<String>>> futures = ...
BetterFuture<Stream<Optional<String>>> futureStream = BetterFuture.collect(futures);
Here's a full fledged examples where the first future times out and the second one completes successfully:
#Test
public void timeoutTest() throws ExecutionException, InterruptedException {
ExecutorService executor = Executors.newCachedThreadPool();
BetterFuture<Optional<String>> fa = BetterFuture.future(executor, () -> {
Thread.sleep(3000);
return "a";
}, Duration.ofSeconds(2));
BetterFuture<Optional<String>> fb = BetterFuture.future(executor, () -> {
Thread.sleep(1000);
return "b";
}, Duration.ofSeconds(2));
Stream<BetterFuture<Optional<String>>> futures = Stream.of(fa, fb);
BetterFuture<Stream<Optional<String>>> c = BetterFuture.collect(futures);
System.out.println(c.get().toList());
}
When run, it prints
[Optional.empty, Optional[b]]
As a final note, the implementation does nothing about the running threads when a timeout occurs. That is, it only times out the future but it does not interrupt the running thread. The thread will keep running in the background until in completes naturally.
It totally depends on the tasks / calculations you are running in parallel if you can really cancel them. Keep in mind that the Java Runtime is not an operating system, and you cannot forcible kill a thread like you could do with a process.
So if you want to interrupt long-running calculations, you will have to write time in a way so they regularly check if they should stop execution. For waiting on some other stuff (sleep, sync on other threads etc.) it is a totally different strategy: you can interrupt such threads and the code receives a InterruptedException that can be used to really stop the code with much less cooperation from the code.
I prepared a small example here to show you the difference:
package examples.stackoverflow.q71322315;
import java.util.concurrent.*;
public class Q71322315 {
public static final long COUNTER = 10000000000L;
public static final boolean SLEEP = false;
private static final ExecutorService taskExec = Executors.newCachedThreadPool();
public static void timedRun(Runnable r, long timeout, TimeUnit unit) throws InterruptedException {
Future<?> task = taskExec.submit(r);
try {
task.get(timeout, unit);
System.out.println("completed");
} catch (TimeoutException e) {
// task will be cancelled below
System.out.println("timeout");
} catch (ExecutionException e) {
System.out.println("exex");
// exception thrown in task; rethrow
throw new RuntimeException(e.getCause());
} finally {
// Harmless if task already completed
task.cancel(true); // interrupt if running
}
}
public static void main(String[] args) throws InterruptedException {
timedRun(new Task(SLEEP), 2000, TimeUnit.MILLISECONDS);
taskExec.shutdown();
System.out.println("finish");
}
private static class Task implements Runnable {
private final boolean sleep;
private Task(boolean sleep) {
this.sleep = sleep;
}
#Override
public void run() {
try {
if (sleep) {
Thread.sleep(5000L);
} else {
longRunningMethod(COUNTER);
}
System.out.println("Success");
} catch (Exception e) {
e.printStackTrace();
if (e instanceof InterruptedException) {
Thread.currentThread().interrupt();
}
}
}
private void longRunningMethod(long counter) {
for (long y = 0; y < counter; y++) {
Math.sqrt(y);
}
}
}
}
The example is based on some example code of the already mentioned Java Concurrency in Practice - "7.10 Cancelling a task using Future."
The code as above performs a long running calculation that doesn't care about any interruptions. (You might have to increase the value of COUNTER, just add zeros at the end until the whole method takes longer than 2 seconds.)
You will see that you first get the "timeout" message that indicates that the task wasn't finished in the wanted timeout. But the code continues running and also prints out "finish" and "Success".
When you flip the SLEEP constant to true it uses an interruptible call to Thread.sleep() instead and the output will not contain the "Success" message.
After you managed to build a cancellable / interruptible computation you can then set up multiple threads that each execute the timedRun execution in parallel, so the tasks are started in parallel and also interrupted after the timeout.
This does not yet include the collection of the results, but instead of the sysouts for completed and timeout you can collect results or count the timed out tasks.
(And if you want to use that code in production, please clean it up very thoroughly, it has really some huge smells that should never land in any production ready code ;-)
We had similar requirement where we need capture timeout of each thread and ignore the results. Java 8 doesn't have this in built.
One of the ways we achieved it,
List<CompletableFuture<?>> futures = new ArrayList<>();
List<?> results = new ArrayList<>(); // It can be anything you collect
futures.add(asyncService.fetchMethod()
.acceptEither(
timeoutAfter(timeout, TimeUnit.SECONDS),
results:add)
.handle(
(result, ex) -> {
//Handle the timeout exception
results.add(...);
return result
});
CompletableFuture.allOf(futures.toArray(new CompletableFuture[0])).join();
private <T> CompletableFuture<T> timeoutAfter(long timeout, TimeUnit unit) {
CompletableFuture<T> result = new CompletableFuture<>();
// We need a separate executor here
scheduledExecutor.schedule(
() -> result.completeExceptionally(new TimeoutException()), timeout, unit);
);
return result;
}
Suppose you want 10 threads running and want a returned value, you can use the Callable<Boolean> interface, submit it to ExecutorService, and then get the result using Future#get returned as a Boolean.
Here is an example usage.
final int NUM_THREADS = 10;
List<Boolean> results = new ArrayList<Boolean>();
List<Callable<Boolean>> callables = new ArrayList<Callable<Boolean>>();
for(int i=0; i<NUM_THREADS; ++i)
{
callables.add(new Callable<Boolean>()
{
public Boolean call()
{
// Add your task here
return isTaskCompleted;
}
});
}
ExecutorService executorService = ExecutorService.newFixedThreadPool(NUM_THREADS); // Run 10 threads
for(Callable<Boolean> callable:callables)
{
Future<Boolean> future = executor.submit(callable);
try
{
results.add(future.get(2, TimeUnit.SECONDS)); // Timeout 2 seconds and add the result
}
catch(Exception ex)
{
results.add(false); // Set result to false if task throw TimeOutExeption
}
}
If you want more information about these classes you can read this book: O'Reilly - Learning Java, Chapter 9: Threads.
Here is what helped me.
Problem in hand: Given X seconds as timeout return the value from the Task which completes first, in case none of the Task is able to finish then return default value.
import java.util.concurrent.ExecutionException;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.TimeoutException;
public class Test {
public static void main(String[] args) {
CompletableFuture<String> f1 = CompletableFuture.supplyAsync(() -> func("Task 1", 1000));
CompletableFuture<String> f2 = CompletableFuture.supplyAsync(() -> func("Task 2", 2000));
String str = null;
try {
str = (String) CompletableFuture.anyOf(f1, f2).get(3, TimeUnit.SECONDS);
} catch (InterruptedException | ExecutionException e) {
e.printStackTrace();
} catch (TimeoutException e) {
str = "Default";
e.printStackTrace();
}
System.out.println(str);
}
public static String func(String task, int sleepTime) {
try {
Thread.sleep(sleepTime);
} catch (InterruptedException e) {
e.printStackTrace();
}
return task;
}
}
I know that CompletableFuture design does not control its execution with interruptions, but I suppose some of you might have this problem. CompletableFutures are very good way to compose async execution, but given the case when you want the underlying execution to be interrupted or stopped when future is canceled, how do we do that? Or we must just accept that any canceled or manually completed CompletableFuture will not impact the thread working out there to complete it?
That is, in my opinion, obviously a useless work that takes time of executor worker. I wonder what approach or design might help in this case?
UPDATE
Here is a simple test for this
public class SimpleTest {
#Test
public void testCompletableFuture() throws Exception {
CompletableFuture<Void> cf = CompletableFuture.runAsync(()->longOperation());
bearSleep(1);
//cf.cancel(true);
cf.complete(null);
System.out.println("it should die now already");
bearSleep(7);
}
public static void longOperation(){
System.out.println("started");
bearSleep(5);
System.out.println("completed");
}
private static void bearSleep(long seconds){
try {
TimeUnit.SECONDS.sleep(seconds);
} catch (InterruptedException e) {
System.out.println("OMG!!! Interrupt!!!");
}
}
}
A CompletableFuture is not related to the asynchronous action that may eventually complete it.
Since (unlike FutureTask) this class has no direct control over the
computation that causes it to be completed, cancellation is treated as
just another form of exceptional completion. Method cancel has the
same effect as completeExceptionally(new CancellationException()).
There may not even be a separate thread working on completing it (there may even be many threads working on it). Even if there is, there's no link from a CompletableFuture to any thread that has a reference to it.
As such, there's nothing you can do through CompletableFuture to interrupt any thread that may be running some task that will complete it. You'll have to write your own logic which tracks any Thread instances which acquire a reference to the CompletableFuture with the intention to complete it.
Here's an example of the type of execution I think you could get away with.
public static void main(String[] args) throws Exception {
ExecutorService service = Executors.newFixedThreadPool(1);
CompletableFuture<String> completable = new CompletableFuture<>();
Future<?> future = service.submit(new Runnable() {
#Override
public void run() {
for (int i = 0; i < 10; i++) {
if (Thread.interrupted()) {
return; // remains uncompleted
}
try {
Thread.sleep(1000);
} catch (InterruptedException e) {
return; // remains uncompleted
}
}
completable.complete("done");
}
});
Thread.sleep(2000);
// not atomic across the two
boolean cancelled = future.cancel(true);
if (cancelled)
completable.cancel(true); // may not have been cancelled if execution has already completed
if (completable.isCancelled()) {
System.out.println("cancelled");
} else if (completable.isCompletedExceptionally()) {
System.out.println("exception");
} else {
System.out.println("success");
}
service.shutdown();
}
This assumes that the task being executed is setup to handle interruptions correctly.
What about this?
public static <T> CompletableFuture<T> supplyAsync(final Supplier<T> supplier) {
final ExecutorService executorService = Executors.newFixedThreadPool(1);
final CompletableFuture<T> cf = new CompletableFuture<T>() {
#Override
public boolean complete(T value) {
if (isDone()) {
return false;
}
executorService.shutdownNow();
return super.complete(value);
}
#Override
public boolean completeExceptionally(Throwable ex) {
if (isDone()) {
return false;
}
executorService.shutdownNow();
return super.completeExceptionally(ex);
}
};
// submit task
executorService.submit(() -> {
try {
cf.complete(supplier.get());
} catch (Throwable ex) {
cf.completeExceptionally(ex);
}
});
return cf;
}
Simple Test:
CompletableFuture<String> cf = supplyAsync(() -> {
try {
Thread.sleep(1000L);
} catch (Exception e) {
System.out.println("got interrupted");
return "got interrupted";
}
System.out.println("normal complete");
return "normal complete";
});
cf.complete("manual complete");
System.out.println(cf.get());
I don't like the idea of having to create an Executor service every time, but maybe you can find a way to reuse the ForkJoinPool.
If you use
cf.get();
instead of
cf.join();
The thread waiting on the completion can be interrupted. This bit me in the a**, so I'm just putting it out there. You'd then need to propagate this interruption further / use cf.cancel(...) to really finish the execution.
I had similar issue wherein I needed to simulate a InterruptedException.
I mocked the method call that is supposed to return the CompletetableFuture, and I put a spy on return value such that CompletableFuture#get will throw the exception.
It worked as I expected, and I was able to test that code handled the exception correctly.
CompletableFuture spiedFuture = spy(CompletableFuture.completedFuture(null));
when(spiedFuture .get()).thenThrow(new InterruptedException());
when(servuce.getById(anyString())).thenReturn(spiedFuture );
Here is a ultra-short version to create a Future task that can be cancelled:
public static <T> Future<T> supplyAsync(Function<Future<T>, T> operation) {
CompletableFuture<T> future = new CompletableFuture<>();
return future.completeAsync(() -> operation.apply(future));
}
The CompletableFuture is passed to the operation Function to be able to check the cancel status of the Future:
Future<Result> future = supplyAsync(task -> {
while (!task.isCancelled()) {
// computation
}
return result;
});
// later you may cancel
future.cancel(false);
// or retrieve the result
Result result = future.get(5, TimeUnit.SECONDS);
This however does not interrupt the Thread running the operation. If you also want to be able to interrupt the Thread, then you have to store a reference to it and override Future.cancel(..) to interrupt it.
public static <T> Future<T> supplyAsync(Function<Future<T>, T> action) {
return supplyAsync(action, r -> new Thread(r).start());
}
public static <T> Future<T> supplyAsync(Function<Future<T>, T> action, Executor executor) {
AtomicReference<Thread> interruptThread = new AtomicReference<>();
CompletableFuture<T> future = new CompletableFuture<>() {
#Override
public boolean cancel(boolean mayInterruptIfRunning) {
if (!interruptThread.compareAndSet(null, Thread.currentThread())
&& mayInterruptIfRunning) {
interruptThread.get().interrupt();
}
return super.cancel(mayInterruptIfRunning);
}
};
executor.execute(() -> {
if (interruptThread.compareAndSet(null, Thread.currentThread())) try {
future.complete(action.apply(future));
} catch (Throwable e) {
future.completeExceptionally(e);
}
});
return future;
}
The following test checks that the Thread executing our Function got interrupted:
#Test
void supplyAsyncWithCancelOnInterrupt() throws Exception {
Object lock = new Object();
CountDownLatch done = new CountDownLatch(1);
CountDownLatch started = new CountDownLatch(1);
Future<Object> future = supplyAsync(m -> {
started.countDown();
synchronized (lock) {
try {
lock.wait(); // let's get interrupted
} catch (InterruptedException e) {
done.countDown();
}
}
return null;
});
assertFalse(future.isCancelled());
assertFalse(future.isDone());
assertTrue(started.await(5, TimeUnit.SECONDS));
assertTrue(future.cancel(true));
assertTrue(future.isCancelled());
assertTrue(future.isDone());
assertThrows(CancellationException.class, () -> future.get());
assertTrue(done.await(5, TimeUnit.SECONDS));
}
What about?
/** #return {#link CompletableFuture} which when cancelled will interrupt the supplier
*/
public static <T> CompletableFuture<T> supplyAsyncInterruptibly(Supplier<T> supplier, Executor executor) {
return produceInterruptibleCompletableFuture((s) -> CompletableFuture.supplyAsync(s, executor), supplier);
}
// in case we want to do the same for similar methods later
private static <T> CompletableFuture<T> produceInterruptibleCompletableFuture(
Function<Supplier<T>,CompletableFuture<T>> completableFutureAsyncSupplier, Supplier<T> action) {
FutureTask<T> task = new FutureTask<>(action::get);
return addCancellationAction(completableFutureAsyncSupplier.apply(asSupplier(task)), () ->
task.cancel(true));
}
/** Ensures the specified action is executed if the given {#link CompletableFuture} is cancelled.
*/
public static <T> CompletableFuture<T> addCancellationAction(CompletableFuture<T> completableFuture,
#NonNull Runnable onCancellationAction) {
completableFuture.whenComplete((result, throwable) -> {
if (completableFuture.isCancelled()) {
onCancellationAction.run();
}
});
return completableFuture; // return original CompletableFuture
}
/** #return {#link Supplier} wrapper for the given {#link RunnableFuture} which calls {#link RunnableFuture#run()}
* followed by {#link RunnableFuture#get()}.
*/
public static <T> Supplier<T> asSupplier(RunnableFuture<T> futureTask) throws CompletionException {
return () -> {
try {
futureTask.run();
try {
return futureTask.get();
} catch (ExecutionException e) { // unwrap ExecutionExceptions
final Throwable cause = e.getCause();
throw (cause != null) ? cause : e;
}
} catch (CompletionException e) {
throw e;
} catch (Throwable t) {
throw new CompletionException(t);
}
};
}
How do I notify my main class which instantiates a ThreadPoolExecutor when all threads within the ThreadPoolExecutor are completed?
ThreadPoolExecutor threadPool = null;
ThreadClass threadclass1;
ThreadClass threadclass2;
final ArrayBlockingQueue<Runnable> queue = new ArrayBlockingQueue<Runnable>(maxPoolSize);
puclic MyClass(){
threadPool = new ThreadPoolExecutor(poolSize, maxPoolSize, keepAliveTime, TimeUnit.SECONDS, queue);
threadClass1 = new ThreadClass;
threadClass2 = new ThreadClass;
threadPool.execute(threadClass1);
threadPool.execute(threadClass2);
//Now I would like to do something until the threadPool is done working
//The threads fill a ConcurrentLinkedQueueand I would like to poll
//the queue as it gets filled by the threads and output
//it to XML via JAX-RS
}
EDIT 1
Wile my threads fetch data from somewhere and fill this information into a ConcurrentLinkedQueue I basically would like to perform some action in MyClass to update the XML output with the results. When all threads are terminated I would like to return true to the JAX-RS webservice which instantiated MyClass so the webservice knows all data has been fetched and it can now display the final XML file
EDIT 2
I am passing a Queue to threads so they can add items to the queue. When one driver is done adding items to the articleQueue I want to perform an action within my main class, polling the entity from the Queue and handing it over to the response object to display it in some way.
When I pass the queue to the threads, are they working with the same object or with a "copy" of the object so that changes within the thread do not effect the main object? That is not the behavior I want. When I check the size of the articleQueue within the Driver it is 18, the size of the articleQueue in the DriverController is 0.
Is there a nicer way to react when a thread has added something to the queue other than my while loop? How do I have to modify my code to acces the same object within different classes?
DriverController
public class DriverController {
Queue<Article> articleQueue;
ThreadPoolExecutor threadPool = null;
final ArrayBlockingQueue<Runnable> queue = new ArrayBlockingQueue<Runnable>(
maxPoolSize);
public DriverController(Response response) {
articleQueue = new ConcurrentLinkedQueue<Article>();
threadPool = new ThreadPoolExecutor();
Driver driver = new Driver(this.articleQueue);
threadPool.execute(driver);
// More drivers would be executed here which add to the queue
while (threadPool.getActiveCount() > 0) {
// this.articleQueue.size() gives back 0 here ... why?
if(articleQueue.size()>0){
response.addArticle(articleQueue.poll());
}
}
}
}
Driver
public class Driver implements Runnable{
private Queue<Article> articleQueue;
public DriverAlliedElectronics(Queue articleQueue) {
this.articleQueue = articleQueue;
}
public boolean getData() {
// Here would be the code where the article is created ...
this.articleQueue.offer(article);
return true;
}
public void run() {
this.getData();
// this.articleQueue.size() gives back 18 here ...
}
}
You should try to use following snippet
//Now I would like to wait until the threadPool is done working
threadPool.shutdown();
while (!threadPool.isTerminated()) {
try {
threadPool.awaitTermination(10, TimeUnit.MILLISECONDS);
} catch (InterruptedException e) {
e.printStackTrace();
}
}
Maybe a ExecutorCompletionService might be the right thing for you:
http://download.oracle.com/javase/1.5.0/docs/api/java/util/concurrent/ExecutorCompletionService.html
Example from the link above:
void solve(Executor e, Collection<Callable<Result>> solvers)
throws InterruptedException, ExecutionException {
CompletionService<Result> ecs = new ExecutorCompletionService<Result>(e);
for (Callable<Result> s : solvers)
ecs.submit(s);
int n = solvers.size();
for (int i = 0; i < n; ++i) {
Result r = ecs.take().get();
if (r != null)
use(r);
}
}
Instead of using execute you should use submit. This will return a Future instance on which you can wait for the task(s) to complete. That way you don't need polling or shutting down the pool.
I don't think there's a way to do this explicitly. You could poll the getCompletedTaskCount() to wait for that to become zero.
Why not collect the Future objects returned upon submission and check for all of those being completed ? Simply call get() on each one in turn. Since that call blocks you'll simply wait for each in turn and gradually fall through the set until you've waited on each on.
Alternatively you could submit the threads, and call shutdown() on the executor. That way, the submitted tasks will be executed, and then the terminated() method is called. If you override this then you'll get a callback once all tasks are completed (you couldn't use that executor again, obviously).
Judging from the reference documentation you have a few options:
ThreadPoolExecutor threadPool = null;
ThreadClass threadclass1;
ThreadClass threadclass2;
final ArrayBlockingQueue<Runnable> queue = new ArrayBlockingQueue<Runnable>(maxPoolSize);
puclic MyClass(){
threadPool = new ThreadPoolExecutor(poolSize, maxPoolSize, keepAliveTime, TimeUnit.SECONDS, queue);
threadClass1 = new ThreadClass;
threadClass2 = new ThreadClass;
threadPool.execute(threadClass1);
threadPool.execute(threadClass2);
//Now I would like to wait until the threadPool is done working
//Option 1: shutdown() and awaitTermination()
threadPool.shutDown();
try {
threadPool.awaitTermination(Long.MAX_VALUE, TimeUnit.SECONDS)
}
catch (InterruptedException e) {
e.printStackTrace();
}
//Option 2: getActiveCount()
while (threadPool.getActiveCount() > 0) {
try {
Thread.sleep(1000);
}
catch (InterruptedException ignored) {}
}
//Option 3: getCompletedTaskCount()
while (threadPool.getCompletedTaskCount() < totalNumTasks) {
try {
Thread.sleep(1000);
}
catch (InterruptedException ignored) {}
}
}
All things considered, I think shutdown() and awaitTermination() is the best option of the three.
I think you're overengineering things a bit. You don't really care about the threads or the thread pool, and rightly so. Java provides nice abstractions so that you don't have to. You just need to know when your tasks are complete, and methods exist for that. Just submit your jobs, and wait for the futures to say they're done. If you really want to know as soon as a single task completes, you can watch all the futures and take action as soon as any one is finished. If not and you only care that everything is finished, you can remove some complexity from the code I'm about to post. Try this on for size (note MultithreadedJaxrsResource is executable):
import javax.ws.rs.*;
import javax.ws.rs.core.MediaType;
import java.util.*;
import java.util.concurrent.*;
#Path("foo")
public class MultithreadedJaxrsResource {
private ExecutorService executorService;
public MultithreadedJaxrsResource(ExecutorService executorService) {
this.executorService = executorService;
}
#GET
#Produces(MediaType.APPLICATION_XML)
public AllMyArticles getStuff() {
List<Future<Article>> futures = new ArrayList<Future<Article>>();
// Submit all the tasks to run
for (int i = 0; i < 10; i++) {
futures.add(executorService.submit(new Driver(i + 1)));
}
AllMyArticles articles = new AllMyArticles();
// Wait for all tasks to finish
// If you only care that everything is done and not about seeing
// when each one finishes, this outer do/while can go away, and
// you only need a single for loop to wait on each future.
boolean allDone;
do {
allDone = true;
Iterator<Future<Article>> futureIterator = futures.iterator();
while (futureIterator.hasNext()) {
Future<Article> future = futureIterator.next();
if (future.isDone()) {
try {
articles.articles.add(future.get());
futureIterator.remove();
} catch (InterruptedException e) {
// thread was interrupted. don't do that.
throw new IllegalStateException("broken", e);
} catch (ExecutionException e) {
// execution of the Callable failed with an
// exception. check it out.
throw new IllegalStateException("broken", e);
}
} else {
allDone = false;
}
}
} while (!allDone);
return articles;
}
public static void main(String[] args) {
ExecutorService executorService = Executors.newFixedThreadPool(10);
AllMyArticles stuff =
new MultithreadedJaxrsResource(executorService).getStuff();
System.out.println(stuff.articles);
executorService.shutdown();
}
}
class Driver implements Callable<Article> {
private int i; // Just to differentiate the instances
public Driver(int i) {
this.i = i;
}
public Article call() {
// Simulate taking some time for each call
try {
Thread.sleep(1000 / i);
} catch (InterruptedException e) {
System.err.println("oops");
}
return new Article(i);
}
}
class AllMyArticles {
public final List<Article> articles = new ArrayList<Article>();
}
class Article {
public final int i;
public Article(int i) {
this.i = i;
}
#Override
public String toString() {
return "Article{" +
"i=" + i +
'}';
}
}
Done that way, you can plainly see that the tasks are returned in the order they complete, as the last task finishes first thanks to sleeping the shortest time. If you don't care about completion order and just want to wait for all to finish, the loop becomes much simpler:
for (Future<Article> future : futures) {
try {
articles.articles.add(future.get());
} catch (InterruptedException e) {
// thread was interrupted. don't do that.
throw new IllegalStateException("broken", e);
} catch (ExecutionException e) {
// execution of the Callable failed with an exception. check it out.
throw new IllegalStateException("broken", e);
}
}