Sorry I have to open a new thread to describe this problem.
This morning I asked this question, there're some replies but my problem is still not solved.
This time I will attach some runnable code(simplified but with the same problem) for you to reproduce the problem:
public class ThreadPoolTest {
public static void main(String[] args) throws Exception {
final ExecutorService taskExecutor = Executors.newFixedThreadPool(5);
Future<Void> futures[] = new Future[5];
for (int i = 0; i < futures.length; ++i)
futures[i] = startTask(taskExecutor);
for (int i = 0; i < futures.length; ++i)
System.out.println("futures[i].cancel(true): " + futures[i].cancel(true));
System.out.println("Cancel DONE.");
taskExecutor.shutdown();
}
private static Future<Void> startTask(final ExecutorService taskExecutor) {
Future<Void> f = taskExecutor.submit(new Callable<Void>() {
public Void call() throws Exception {
try {
downloadFile(new URI("http://stackoverflow.com"));
while(true) {
System.out.println(Thread.currentThread().getName() + ": " + Thread.currentThread().isInterrupted());
if(Thread.currentThread().isInterrupted())
break;
}
} catch (Exception ex) {
ex.printStackTrace();
}
return null;
}
});
return f;
}
private static void downloadFile (final URI uri) throws Exception {
// if(true) return;
Socket socket = new Socket (uri.getHost(), uri.getPort() == -1 ? 80 : uri.getPort());
return;
}
}
The code above will most likely be trapped in an infinite loop(you may want to run the code multiple times to witness what I saw), as you can see in the main method I have called futures[i].cancel(true) for all tasks, I don't know why this is happening, this has been torturing me for more than a day.
Your help will be greatly appreciated.
I've played with your code, and noticed that the thread's interrupt status is sometimes true before the socket creation, and false after.
I have tried interrupting a thread and calling the Socket constructor, and the thread always stays interrupted after. I also tried removing the shutdown of the threadpool, and the problem continued to happen.
Then I have tried using 5 different URIs, rather than always the same one. And the problem never happened.
So I wrote this simple program, showing that the thread pool is not the culprit, but the socket is:
public static void main(String[] args) throws Exception {
final URI uri = new URI("http://stackoverflow.com");
for (int i = 0; i < 5; i++) {
Runnable r = new Runnable() {
#Override
public void run() {
Thread.currentThread().interrupt();
System.out.println(Thread.currentThread().isInterrupted());
try {
Socket socket = new Socket (uri.getHost(), uri.getPort() == -1 ? 80 : uri.getPort());
}
catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
System.out.println(Thread.currentThread().isInterrupted());
}
};
new Thread(r).start();
}
}
And indeed, when 5 threads create a socket to the same host and port, 4 of them have their interrupt status cleared.
Then I tried to synchronize the socket creation (on a single lock, but I guess you might use one lock per host/port) :
synchronized(lock) {
try {
Socket socket = new Socket (uri.getHost(), uri.getPort() == -1 ? 80 : uri.getPort());
}
catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
and TADA... the problem disappeared. I would open a bug at Oracle to signal the problem.
I ran your code, and it didn't stop, as you said.
Didn't have much time to investigate why it behaves so, but I found out that declaring the executor service's threads as daemons made the problem go away :
private static ExecutorService TaskExecutor = Executors.newFixedThreadPool(5, new ThreadFactory() {
public Thread newThread(Runnable r) {
Thread t = new Thread(r);
t.setDaemon(true);
return t;
}
});
I'll come back if I find a better explanation.
I think the problem that task are not started when you try to cancel them. I added Thread.sleep(100) like this:
for (int i = 0; i < futures.length; ++i)
futures[i] = startTask(taskExecutor);
Thread.sleep(100);
for (int i = 0; i < futures.length; ++i)
System.out.println("futures[i].cancel(true): " + futures[i].cancel(true));
and everything was cancelled ok.
Related
Below is my code:
public class Controller {
public Button button_submitWork;
#FXML
public void handleSubmitWork(ActionEvent event) {
final ExecutorService executorService = Executors.newFixedThreadPool(1, r -> {
Thread t = Executors.defaultThreadFactory().newThread(r);
t.setDaemon(true);
return t;
});//set thread daemon, let all threads terminate when the program is closed.
Callable<String> callable = new Callable<String>() {
#Override
public String call() throws Exception {
System.out.println("Executor Service thread");
StringBuilder stringBuilder_output = new StringBuilder();
for (int k = 0; k < 5; k++) {
stringBuilder_output.append(k);
}
//Thread.sleep(1000);
return stringBuilder_output.toString() + "\n";
}
};
Future<String> future = executorService.submit(callable);//Weird line.
//This line must be placed inside the "watchThread" to get the result, but why???
Thread watchThread = new Thread(new Runnable() {
#Override
public void run() {
//<----------Moving to here solve the problem!
System.out.println("Watch thread");
while (!Thread.currentThread().isInterrupted() && !future.isDone()) {
try {
String result = future.get();
System.out.println(result);
} catch (InterruptedException e) {
e.printStackTrace();
} catch (ExecutionException e) {
e.printStackTrace();
} finally {
executorService.shutdownNow();
}
}
}
});
watchThread.setDaemon(true);
watchThread.start();
System.out.println("FX thread");
}
}
The question is that the System.out.println(result); inside "watchThread" is never been called. The console output looks like this:
Executor Service thread
FX thread
Watch thread
But when I move the Future<String> future = executorService.submit(callable); to the inside of run method of "watchThread", the output change to:
FX thread
Watch thread
Executor Service thread
01234
which is I expected.
I also discovered that if the call() method has a longer task, say a Thread.sleep(1000), the output change to the result I expected.
So why is that?
The thread you submit to executorService finishes before this line:
while (!Thread.currentThread().isInterrupted() && !future.isDone()) { is called so future.isDone returns true and the while loop is not executed.
If you add Thread.sleep(1000) then it still runs and future.isDone returns false and the while loop executes. The same thing happens when you move Future<String> future = executorService.submit(callable); inside watchThread.
I have the following work queue implementation, which I use to limit the number of threads in use. It works by me initially adding a number of Runnable objects to the queue, and when I am ready to begin, I run "begin()". At this point I do not add any more to the queue.
public class WorkQueue {
private final int nThreads;
private final PoolWorker[] threads;
private final LinkedList queue;
Integer runCounter;
boolean hasBegun;
public WorkQueue(int nThreads) {
runCounter = 0;
this.nThreads = nThreads;
queue = new LinkedList();
threads = new PoolWorker[nThreads];
hasBegun = false;
for (int i = 0; i < nThreads; i++) {
threads[i] = new PoolWorker();
threads[i].start();
}
}
public boolean isQueueEmpty() {
synchronized (queue) {
if (queue.isEmpty() && runCounter == 0) {
return true;
} else {
return false;
}
}
}
public void begin() {
hasBegun = true;
synchronized (queue) {
queue.notify();
}
}
public void add(Runnable r) {
if (!hasBegun) {
synchronized (queue) {
queue.addLast(r);
runCounter++;
}
} else {
System.out.println("has begun executing. Cannot add more jobs ");
}
}
private class PoolWorker extends Thread {
public void run() {
Runnable r;
while (true) {
synchronized (queue) {
while (queue.isEmpty()) {
try {
queue.wait();
} catch (InterruptedException ignored) {
}
}
r = (Runnable) queue.removeFirst();
}
// If we don't catch RuntimeException,
// the pool could leak threads
try {
r.run();
synchronized (runCounter) {
runCounter--;
}
} catch (RuntimeException e) {
// You might want to log something here
}
}
}
}
}
This is a runnable I use to keep track of when all the jobs on the work queue have finished:
public class QueueWatcher implements Runnable {
private Thread t;
private String threadName;
private WorkQueue wq;
public QueueWatcher(WorkQueue wq) {
this.threadName = "QueueWatcher";
this.wq = wq;
}
#Override
public void run() {
while (true) {
if (wq.isQueueEmpty()) {
java.util.Date date = new java.util.Date();
System.out.println("Finishing and quiting at:" + date.toString());
System.exit(0);
break;
} else {
try {
Thread.sleep(1000);
} catch (InterruptedException ex) {
Logger.getLogger(PlaneGenerator.class.getName()).log(Level.SEVERE, null, ex);
}
}
}
}
public void start() {
wq.begin();
System.out.println("Starting " + threadName);
if (t == null) {
t = new Thread(this, threadName);
t.setDaemon(false);
t.start();
}
}
}
This is how I use them:
Workqueue wq = new WorkQueue(9); //Get same results regardless of 1,2,3,8,9
QueueWatcher qw = new QueueWatcher(wq);
SomeRunnable1 sm1 = new SomeRunnable1();
SomeRunnable2 sm2 = new SomeRunnable2();
SomeRunnable3 sm3 = new SomeRunnable3();
SomeRunnable4 sm4 = new SomeRunnable4();
SomeRunnable5 sm5 = new SomeRunnable5();
wq.add(sm1);
wq.add(sm2);
wq.add(sm3);
wq.add(sm4);
wq.add(sm5);
qw.start();
But regardless of how many threads I use, the result is always the same - it always takes about 1m 10seconds to complete. This is about the same as when I just did a single threaded version (when everything ran in main()).
If I set wq to (1,2,3--9) threads it is always between 1m8s-1m10s. What is the problem ? The jobs (someRunnable) have nothing to do with each other and cannot block each other.
EDIT: Each of the runnables just read some image files from the filesystems and create new files in a separate directory. The new directory eventually contains about 400 output files.
EDIT: It seems that only one thread is always doing work. I made the following changes:
I let the Woolworker store an Id
PoolWorker(int id){
this.threadId = id;
}
Before running I print the id of the worker.
System.out.println(this.threadId + " got new task");
r.run();
In WorkQueue constructor when creating the poolworkers I do:
for (int i = 0; i < nThreads; i++) {
threads[i] = new PoolWorker(i);
threads[i].start();
}
But it seems that that only thread 0 does any work, as the output is always:
0 got new task
Use queue.notifyAll() to start processing.
Currently you're using queue.notify(), which will only wake a single thread. (The big clue that pointed me to this was when you mentioned only a single thread was running.)
Also, synchronizing on Integer runCounter isn't doing what you think it's doing - runCounter++ is actually assigning a new value to the Integer each time, so you're synchronizing on a lot of different Integer objects.
On a side note, using raw threads and wait/notify paradigms is complicated and error-prone even for the best programmers - it's why Java introduced the java.util.concurrent package, which provide threadsafe BlockingQueue implementations and Executors for easily managing multithreaded apps.
I am using single thread executor for long-running threads like this:
executor = Executors.newSingleThreadExecutor(THREAD_FACTORY);
executor.submit(new LongRunnable());
which checks a flag to be stopped:
private class LongRunnable implements Runnable {
#Override
public void run() {
while (isRunning.get()) {
try {
doSomething();
} catch (InterruptedException e) {
...
}
}
}
}
and whole execution is interrupted that way:
#Override
public void close() throws Exception {
isRunning.set(false);
executor.shutdownNow();
}
Still I can see some threads not gc-ed in profiler (while by logs, runnable they were executing has quit outermost while loop).
Question: does provided working with threads strategy memory-leak-free and thread-leak-free?
I am not able to see any issue with executor or shutDownNow. Probably you are looking at different threads in your profiler.
Try this program which is similar to the one in your question and you can see the thread is no longer there after successful shutdown.
public class ExecutorShutdownTest {
private static ExecutorService executor;
private static AtomicLong executorThreadId = new AtomicLong(0);
public static void main(String[] args) {
// get thread MX bean
ThreadMXBean threadMXBean = ManagementFactory.getThreadMXBean();
// create an executor and start the task
executor = Executors.newSingleThreadExecutor(new TestThreadFactory());
LongRunnable runnable = new LongRunnable();
executor.submit(runnable);
// main thread: keep running for sometime
int count = 5;
while (count-- > 0) {
try {
Thread.sleep(1000);
System.out.println(String.valueOf(threadMXBean.getThreadInfo(executorThreadId.longValue())).replace("\r", "").replace(
"\n", ""));
} catch (InterruptedException e) {
e.printStackTrace();
}
}
// main thread: stop the task
try {
runnable.close();
System.out.println(String.valueOf(threadMXBean.getThreadInfo(executorThreadId.longValue())).replace("\r", "").replace("\n", ""));
} catch (Exception e) {
e.printStackTrace();
}
// main thread: run some more time to verify the executor thread no longer exists
count = 5;
while (count-- > 0) {
try {
Thread.sleep(1000);
System.out.println(String.valueOf(threadMXBean.getThreadInfo(executorThreadId.longValue())).replace("\r", "").replace("\n", ""));
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
private static class LongRunnable implements Runnable {
private volatile boolean isRunning = true;
#Override
public void run() {
while (isRunning) {
System.out.println("Running");
try {
Thread.sleep(1000);
} catch (InterruptedException e) {
//ignore
}
}
System.out.println("Stopped");
}
public void close() throws Exception {
System.out.println("Stopping");
isRunning = false;
executor.shutdownNow();
}
}
private static class TestThreadFactory implements ThreadFactory {
private static final AtomicInteger poolNumber = new AtomicInteger(1);
private final ThreadGroup group;
private final AtomicInteger threadNumber = new AtomicInteger(1);
private final String namePrefix;
TestThreadFactory() {
SecurityManager s = System.getSecurityManager();
group = (s != null) ? s.getThreadGroup() : Thread.currentThread().getThreadGroup();
namePrefix = "pool-" + poolNumber.getAndIncrement() + "-thread-";
}
public Thread newThread(Runnable r) {
Thread t = new Thread(group, r, namePrefix + threadNumber.getAndIncrement(), 0) {
#Override protected void finalize() throws Throwable {
super.finalize();
// probably bad idea but lets see if it gets here
System.out.println("Executor thread removed from JVM");
}
};
if (t.isDaemon())
t.setDaemon(false);
if (t.getPriority() != Thread.NORM_PRIORITY)
t.setPriority(Thread.NORM_PRIORITY);
executorThreadId.set(t.getId());
System.out.println("Executor thread created");
return t;
}
}
}
Here's a sample program using the single-thread Executor that manages to strand a thread so that the JVM can't shut down, but it only manages to do it by not calling shutdownNow:
import java.util.concurrent.*;
public class Exec {
public static void main(String[] args) throws Exception {
ExecutorService executor = Executors.newSingleThreadExecutor();
executor.submit(new MyTask());
Thread.sleep(20000L);
// executor.shutdownNow();
int retryCount = 4;
while (!executor.isTerminated() && retryCount > 0) {
System.out.println("waiting for tasks to terminate");
Thread.sleep(500L);
retryCount -= 1;
}
}
}
class MyTask implements Runnable {
public void run() {
int count = 0;
try {
while (!Thread.currentThread().isInterrupted() && count < 10) {
Thread.sleep(1000L);
count += 1;
}
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
}
System.out.println("all done");
}
}
The thread used by the executor has a separate life cycle from the task, this example shows how the task finishes but the thread goes on. Uncommenting the shutdownNow results in the executor's thread terminating. Otherwise the main thread sleeps for a while and exits, leaving the executor's thread hanging out, preventing the JVM from exiting.
My guess is that your close method isn't getting called and your executor never gets shut down. To get more useful answers please add a MVCE so that we can reproduce the problem.
Consider that with interruption there's no need to keep a reference to the Runnable to set the flag. As I read the question the task not finishing is not an issue here, but it would still be better to make the Runnable respond to interruption and lose the flag, just because having less things to keep track of is always an improvement.
I am trying to signal between two threads using the below FutureResult class which extends FutureTask class. When run the script, it prints the following result.
SENDING: 0
SENT: 0
POLL: FutureResult#513431
SIGNALLED: FutureResult#513431
Then the program hang up forever. I expect FutureResult instance should return the value from it's blocking get method. Then print the result in the console. But FutureResult.get is blocking forever.
import java.util.concurrent.*;
/**
* Created by someone on 20/08/2015.
*/
final public class FutureResult<T> extends FutureTask<T> {
private static final Object SS = "SS";
public FutureResult() {
super(() -> null);
}
public void signal(final T value) {
set(value);
}
public void signalError(final Throwable throwable) {
setException(throwable);
}
public static void main(String... args) throws Exception {
final ArrayBlockingQueue<FutureResult> queue = new ArrayBlockingQueue<>(1000000);
new Thread(() -> {
while (true) {
try {
final FutureResult poll = queue.take();
System.out.println("POLL: " + poll);
if (poll != null) {
poll.signal(SS);
System.out.println("SIGNALLED: " + poll);
}
} catch (Exception e) {
e.printStackTrace();
}
}
}).start();
new Thread(() -> {
for (int i = 0; i < 1; i++) {
final FutureResult<Object> result = new FutureResult<>();
System.out.println("SENDING: " + i);
queue.offer(new FutureResult());
try {
System.out.println("SENT: " + i);
result.get();
System.out.println("GOT : " + i);
} catch (InterruptedException e) {
e.printStackTrace();
} catch (ExecutionException e) {
e.printStackTrace();
}
}
}).start();
}
}
This is the problem:
queue.offer(new FutureResult());
You're setting the value on one FutureResult, but that's not the one you're waiting for. Just change that line to:
queue.offer(result);
and it works fine.
Looks like the confusion is in the use of FutureTask. FutureTask is designed as a Runnable; running it is necessary.
Honestly, based on the code, it looks like the custom code is implementing something similar to FutureTask. If the intent here is to learn to use FutureTask, then create a FutureTask instance with a "run" method, and then execute that run method. On completion of the run method, the FutureTask.get() will complete.
I call the below testMethod, after putting it into a Callable(with other few Callable tasks), from an ExecutorService. I suspect that, the map.put() suffers OutOfMemoryError, as I'm trying to put some 20 million entries.
But, I'm not able to see the error trace in console. Just the thread stops still. I tried to catch the Error ( I know.. we shouldnt, but for debug I caught). But, the error is not caught. Directly enters finally and stops executing.. and the thread stands still.
private HashMap<String, Integer> testMethod(
String file ) {
try {
in = new FileInputStream(new File(file));
br = new BufferedReader(new InputStreamReader(in), 102400);
for (String line; (line= br.readLine()) != null;) {
map.put(line.substring(1,17),
Integer.parseInt(line.substring(18,20)));
}
System.out.println("Loop End"); // Not executed
} catch(Error e){
e.printStackTrace(); //Not executed
}finally {
System.out.println(map.size()); //Executed
br.close();
in.close();
}
return map;
}
Wt could be the mistake, I'm doing?
EDIT: This is how I execute the Thread.
Callable<Void> callable1 = new Callable<Void>() {
#Override
public Void call() throws Exception {
testMethod(inputFile);
return null;
}
};
Callable<Void> callable2 = new Callable<Void>() {
#Override
public Void call() throws Exception {
testMethod1();
return null;
}
};
List<Callable<Void>> taskList = new ArrayList<Callable<Void>>();
taskList.add(callable1);
taskList.add(callable2);
// create a pool executor with 3 threads
ExecutorService executor = Executors.newFixedThreadPool(3);
List<Future<Void>> future = executor.invokeAll(taskList);
//executor.invokeAll(taskList);
latch.await();
future.get(0);future.get(1); //Added this as per SubOptimal'sComment
But, this future.get() didn't show OOME in console.
You should not throw away the future after submitting the Callable.
Future future = pool.submit(callable);
future.get(); // this would show you the OOME
example based on the informations of the requestor to demonstrate
public static void main(String[] args) throws InterruptedException, ExecutionException {
Callable<Void> callableOOME = new Callable<Void>() {
#Override
public Void call() throws Exception {
System.out.println("callableOOME");
HashMap<String, Integer> map = new HashMap<>();
// some code to force an OOME
try {
for (int i = 0; i < 10_000_000; i++) {
map.put(Integer.toString(i), i);
}
} catch (Error e) {
e.printStackTrace();
} finally {
System.out.println("callableOOME: map size " + map.size());
}
return null;
}
};
Callable<Void> callableNormal = new Callable<Void>() {
#Override
public Void call() throws Exception {
System.out.println("callableNormal");
// some code to have a short "processing time"
try {
TimeUnit.SECONDS.sleep(5);
} catch (InterruptedException ex) {
System.err.println(ex.getMessage());
}
return null;
}
};
List<Callable<Void>> taskList = new ArrayList<>();
taskList.add(callableOOME);
taskList.add(callableNormal);
ExecutorService executor = Executors.newFixedThreadPool(3);
List<Future<Void>> future = executor.invokeAll(taskList);
System.out.println("get future 0: ");
future.get(0).get();
System.out.println("get future 1: ");
future.get(1).get();
}
Try catching Throwable as it could be an Exception like IOException or NullPointerException, Throwable captures everything except System.exit();
Another possibility is that the thread doesn't die, instead it becomes increasingly slower and slower due to almost running out of memory but never giving up. You should be able to see this with a stack dump or using jvisualvm while it is running.
BTW Unless all you strings are exactly 16 characters long, you might like to call trim() on the to remove any padding in the String. This could make them shorter and use less memory.
I assume you are using a recent version of Java 7 or 8. If you are using Java 6 or older, it will use more memory as .substring() doesn't create a new underlying char[] to save CPU, but in this case wastes memory.