I was trying to intentionally create visibility issues with threads and I got unexpected results:
public class DownloadStatus {
private int totalBytes;
private boolean isDone;
public void increment() {
totalBytes++;
}
public int getTotalBytes() {
return totalBytes;
}
public boolean isDone() {
return isDone;
}
public void done() {
isDone = true;
}
}
public class DownloadFileTask implements Runnable {
DownloadStatus status;
public DownloadFileTask(DownloadStatus status) {
this.status = status;
}
#Override
public void run() {
System.out.println("start download");
for (int i = 0; i < 10_000; i++) { //"download" a 10,000 bytes file each time you run
status.increment(); //each byte downloaded - update the status
}
System.out.println("download ended with: " + status.getTotalBytes()); //**NOTE THIS LINE**
status.done();
}
}
//creating threads, one to download, another to wait for the download to be done.
public static void main(String[] args) {
DownloadStatus status = new DownloadStatus();
Thread t1 = new Thread(new DownloadFileTask(status));
Thread t2 = new Thread(() -> {
while (!status.isDone()) {}
System.out.println("DONE!!");
});
t1.start();
t2.start();
}
So, running this would create a visibility problem - the second thread wouldn't see the updated value since it had cached it before it got written back by the first thread - this causes an endless (while) loop, the second thread is constantly checking the cached isDone(). (at least that's how I think it works).
The thing I don't get is why this visibility problem stops happening when I comment out the line from the second code block that calls status.getTotalBytes().
From my understanding both threads start by caching the status object as-is, so the second thread should constantly check his cached value (and essentially not see the new value updated by the first thread).
Why is this line calling a method in the status object causing this visibility issue? (and more interestingly - why not calling it fixes it?)
What you call a "visibility problem" is actually a data race.
A single thread sees the effects of its operations in the order they are written. That is if you update a variable and then read it, you'll always see the updated value within that thread.
The effects of a thread's execution may be different when viewed from another thread. This is mainly related to the language and the underlying hardware architecture. The compiler may reorder instructions, delay memory writes while keeping values in registers, or the values may be kept in a cache before written to the main memory. Without an explicit memory barrier, the value in the main memory would not be updated. That's what you call the "visibility problem".
It is likely that there is a memory barrier in System.println. So when you execute that line, all updates up to that point will be committed to the main memory, and the other threads can see it. Note that without explicit synchronization, there is still no guarantee that the other threads will see it, because those threads may re-use the value they got for that variable before. There is nothing in the program that tells the compiler/runtime that the values may be changed by other threads.
This is the race condition between two threads. There is nothing to do with status.getTotalBytes() statement in your code. It is the scheduler that decides which thread will run. It is by chance that you are not getting stuck in the infinit loop after commenting the println statement. The main problem in your code that increment and set status should be atomic operation and replace the definition of run method as below. Secondly increment is also not a atomic operation. You can unpredictable results if there is no proper synchronization.
#Override
public void run() {
System.out.println("start download");
incrementAndSetStatus();
}
public synchronized void incrementAndSetStatus(){
for (int i = 0; i < 100000; i++) { //"download" a 10,000 bytes file each time you run
status.increment(); //each byte downloaded - update the status
}
System.out.println("download ended with: " + status.getTotalBytes()); //**NOTE THIS LINE**
status.done();
}
I was going through the tutorial https://www.youtube.com/watch?v=SC2jXxOPe5E to understand how volatile variable works and came across a strange behavior.
For the following code snippet
public class VolatileDemo {
static boolean running = false;
public static void main(String a[]) throws InterruptedException {
Thread t = new Thread(new Runnable() {
#Override
public void run() {
while (!running) {
}
System.out.print("Started");
while (running) {
}
System.out.print("Stopped");
}
});
t.start();
Thread.sleep(1000);
running = true;
System.out.print("Starting ");
Thread.sleep(1000);
running = false;
System.out.print("Stopping");
}
}
The output is : Starting Stopping (which is understandable by video)
But for the following code snippet
public class VolatileDemo {
static boolean running = false;
public static void main(String a[]) throws InterruptedException {
Thread t = new Thread(new Runnable() {
#Override
public void run() {
while (!running) {
System.out.print("Flag " + running);
}
System.out.print(" Started");
while (running) {
System.out.print(" Flag " + running);
}
System.out.print(" Stopped");
}
});
t.start();
Thread.sleep(1000);
running = true;
System.out.print(" Starting");
Thread.sleep(1000);
running = false;
System.out.print(" Stopping");
}
}
The output is Flag: false Starting Started Flag: true Stopping Stopped(ignore the output)
My concern here is why the thread was able to read the updated value of 'running' in case 2?
Edit: The difference between the two snippets is the addition below statement in later case
System.out.print("Flag " + running);
I think it's important to understand what the purpose of volatile is.
On a multi-processor system with multiple levels of cache an update to a variable can take some time to reach main memory and hence other threads depending on latency and the hardware design. The code basically is supposed to demo this happening, the change to the running variable is changed and the output should show some delay between when it was changed and when the thread actually starts. Adding the volatile keyword should reduce this delay as it forces the write through to main memory to happen immediately instead of when the cache decides to do it.
Please note that volatile doesn't make code thread safe, it just tells the JVM that the variable needs to be written directly to main memory bypassing any delayed writing scheme the caching hardware might otherwise do. It also means that the variable is read from main memory so that stale data is not used. It is for reducing the latency between a variable being updated in one thread and seen to be updated in another. This isn't something you'll need often, and should use sparingly as bypassing the cache will have performance ramifications for your code.
When you've added extra instructions to the code of the thread you've effectively significantly reduced the rate at which it can poll the running variable. I'd say it's likely that the time between running being changed and being updated in main memory is very small, much quicker than the time it takes to output on the console (which takes longer than you'd think). So it is very likely that you won't see what you expect except on the rare occasion that the boolean evaluation in the while loop happens at the exact correct moment.
Unlike deadlock, this particular property of multi-processing is quite hard to demonstrate on a single machine. It's much more likely to happen on a NUMA system architecture (or a cluster) where the cache to memory update latency can be much larger. On a single system the time between a variable being updated in cache and written to main memory is very small.
I have a method that I would like to call. However, I'm looking for a clean, simple way to kill it or force it to return if it is taking too long to execute.
I'm using Java.
to illustrate:
logger.info("sequentially executing all batches...");
for (TestExecutor executor : builder.getExecutors()) {
logger.info("executing batch...");
executor.execute();
}
I figure the TestExecutor class should implement Callable and continue in that direction.
But all i want to be able to do is stop executor.execute() if it's taking too long.
Suggestions...?
EDIT
Many of the suggestions received assume that the method being executed that takes a long time contains some kind of loop and that a variable could periodically be checked.
However, this is not the case. So something that won't necessarily be clean and that will just stop the execution whereever it is is acceptable.
You should take a look at these classes :
FutureTask, Callable, Executors
Here is an example :
public class TimeoutExample {
public static Object myMethod() {
// does your thing and taking a long time to execute
return someResult;
}
public static void main(final String[] args) {
Callable<Object> callable = new Callable<Object>() {
public Object call() throws Exception {
return myMethod();
}
};
ExecutorService executorService = Executors.newCachedThreadPool();
Future<Object> task = executorService.submit(callable);
try {
// ok, wait for 30 seconds max
Object result = task.get(30, TimeUnit.SECONDS);
System.out.println("Finished with result: " + result);
} catch (ExecutionException e) {
throw new RuntimeException(e);
} catch (TimeoutException e) {
System.out.println("timeout...");
} catch (InterruptedException e) {
System.out.println("interrupted");
}
}
}
Java's interruption mechanism is intended for this kind of scenario. If the method that you wish to abort is executing a loop, just have it check the thread's interrupted status on every iteration. If it's interrupted, throw an InterruptedException.
Then, when you want to abort, you just have to invoke interrupt on the appropriate thread.
Alternatively, you can use the approach Sun suggest as an alternative to the deprecated stop method. This doesn't involve throwing any exceptions, the method would just return normally.
I'm assuming the use of multiple threads in the following statements.
I've done some reading in this area and most authors say that it's a bad idea to kill another thread.
If the function that you want to kill can be designed to periodically check a variable or synchronization primitive, and then terminate cleanly if that variable or synchronization primitive is set, that would be pretty clean. Then some sort of monitor thread can sleep for a number of milliseconds and then set the variable or synchronization primitive.
Really, you can't... The only way to do it is to either use thread.stop, agree on a 'cooperative' method (e.g. occassionally check for Thread.isInterrupted or call a method which throws an InterruptedException, e.g. Thread.sleep()), or somehow invoke the method in another JVM entirely.
For certain kinds of tests, calling stop() is okay, but it will probably damage the state of your test suite, so you'll have to relaunch the JVM after each call to stop() if you want to avoid interaction effects.
For a good description of how to implement the cooperative approach, check out Sun's FAQ on the deprecated Thread methods.
For an example of this approach in real life, Eclipse RCP's Job API's 'IProgressMonitor' object allows some management service to signal sub-processes (via the 'cancel' method) that they should stop. Of course, that relies on the methods to actually check the isCancelled method regularly, which they often fail to do.
A hybrid approach might be to ask the thread nicely with interrupt, then insist a couple of seconds later with stop. Again, you shouldn't use stop in production code, but it might be fine in this case, esp. if you exit the JVM soon after.
To test this approach, I wrote a simple harness, which takes a runnable and tries to execute it. Feel free to comment/edit.
public void testStop(Runnable r) {
Thread t = new Thread(r);
t.start();
try {
t.join(2000);
} catch (InterruptedException e) {
throw new RuntimeException(e);
}
if (!t.isAlive()) {
System.err.println("Finished on time.");
return;
}
try {
t.interrupt();
t.join(2000);
if (!t.isAlive()) {
System.err.println("cooperative stop");
return;
}
} catch (InterruptedException e) {
throw new RuntimeException(e);
}
System.err.println("non-cooperative stop");
StackTraceElement[] trace = Thread.getAllStackTraces().get(t);
if (null != trace) {
Throwable temp = new Throwable();
temp.setStackTrace(trace);
temp.printStackTrace();
}
t.stop();
System.err.println("stopped non-cooperative thread");
}
To test it, I wrote two competing infinite loops, one cooperative, and one that never checks its thread's interrupted bit.
public void cooperative() {
try {
for (;;) {
Thread.sleep(500);
}
} catch (InterruptedException e) {
System.err.println("cooperative() interrupted");
} finally {
System.err.println("cooperative() finally");
}
}
public void noncooperative() {
try {
for (;;) {
Thread.yield();
}
} finally {
System.err.println("noncooperative() finally");
}
}
Finally, I wrote the tests (JUnit 4) to exercise them:
#Test
public void testStopCooperative() {
testStop(new Runnable() {
#Override
public void run() {
cooperative();
}
});
}
#Test
public void testStopNoncooperative() {
testStop(new Runnable() {
#Override
public void run() {
noncooperative();
}
});
}
I had never used Thread.stop() before, so I was unaware of its operation. It works by throwing a ThreadDeath object from whereever the target thread is currently running. This extends Error. So, while it doesn't always work cleanly, it will usually leave simple programs with a fairly reasonable program state. For example, any finally blocks are called. If you wanted to be a real jerk, you could catch ThreadDeath (or Error), and keep running, anyway!
If nothing else, this really makes me wish more code followed the IProgressMonitor approach - adding another parameter to methods that might take a while, and encouraging the implementor of the method to occasionally poll the Monitor object to see if the user wants the system to give up. I'll try to follow this pattern in the future, especially methods that might be interactive. Of course, you don't necessarily know in advance which methods will be used this way, but that is what Profilers are for, I guess.
As for the 'start another JVM entirely' method, that will take more work. I don't know if anyone has written a delegating class loader, or if one is included in the JVM, but that would be required for this approach.
Nobody answered it directly, so here's the closest thing i can give you in a short amount of psuedo code:
wrap the method in a runnable/callable. The method itself is going to have to check for interrupted status if you want it to stop (for example, if this method is a loop, inside the loop check for Thread.currentThread().isInterrupted and if so, stop the loop (don't check on every iteration though, or you'll just slow stuff down.
in the wrapping method, use thread.join(timeout) to wait the time you want to let the method run. or, inside a loop there, call join repeatedly with a smaller timeout if you need to do other things while waiting. if the method doesn't finish, after joining, use the above recommendations for aborting fast/clean.
so code wise, old code:
void myMethod()
{
methodTakingAllTheTime();
}
new code:
void myMethod()
{
Thread t = new Thread(new Runnable()
{
public void run()
{
methodTakingAllTheTime(); // modify the internals of this method to check for interruption
}
});
t.join(5000); // 5 seconds
t.interrupt();
}
but again, for this to work well, you'll still have to modify methodTakingAllTheTime or that thread will just continue to run after you've called interrupt.
The correct answer is, I believe, to create a Runnable to execute the sub-program, and run this in a separate Thread. THe Runnable may be a FutureTask, which you can run with a timeout ("get" method). If it times out, you'll get a TimeoutException, in which I suggest you
call thread.interrupt() to attempt to end it in a semi-cooperative manner (many library calls seem to be sensitive to this, so it will probably work)
wait a little (Thread.sleep(300))
and then, if the thread is still active (thread.isActive()), call thread.stop(). This is a deprecated method, but apparently the only game in town short of running a separate process with all that this entails.
In my application, where I run untrusted, uncooperative code written by my beginner students, I do the above, ensuring that the killed thread never has (write) access to any objects that survive its death. This includes the object that houses the called method, which is discarded if a timeout occurs. (I tell my students to avoid timeouts, because their agent will be disqualified.) I am unsure about memory leaks...
I distinguish between long runtimes (method terminates) and hard timeouts - the hard timeouts are longer and meant to catch the case when code does not terminate at all, as opposed to being slow.
From my research, Java does not seem to have a non-deprecated provision for running non-cooperative code, which, in a way, is a gaping hole in the security model. Either I can run foreign code and control the permissions it has (SecurityManager), or I cannot run foreign code, because it might end up taking up a whole CPU with no non-deprecated means to stop it.
double x = 2.0;
while(true) {x = x*x}; // do not terminate
System.out.print(x); // prevent optimization
I can think of a not so great way to do this. If you can detect when it is taking too much time, you can have the method check for a boolean in every step. Have the program change the value of the boolean tooMuchTime to true if it is taking too much time (I can't help with this). Then use something like this:
Method(){
//task1
if (tooMuchTime == true) return;
//task2
if (tooMuchTime == true) return;
//task3
if (tooMuchTime == true) return;
//task4
if (tooMuchTime == true) return;
//task5
if (tooMuchTime == true) return;
//final task
}
I was expecting this code to be thread safe. I ran it a few times, but got different results. However, if I uncomment the sleep(1000) part, it prints 10000 every time (at least from the results from my test runs).
So what's wrong? Could it be something to do with thread.join()?
public class Test implements Runnable{
private int x;
public synchronized void run(){
x++;
}
public static void main(String args[]){
Test test = new Test();
Thread thread = null;
for (int i = 0; i < 10000; i++) {
thread = new Thread(test);
try {
thread.join();
} catch (InterruptedException e) {}
thread.start();
}
// try {
// Thread.sleep(1000);
// } catch (InterruptedException e) {
// e.printStackTrace();
// }
System.out.println(test.x);
}
}
edit: oops, my bad. I misunderstood how Thread#join functions. And synchronizing on run() method is a bad idea.
thread.join() should be called after thread.start().
join() means "block until the thread finishes". That only makes sense after the thread has started.
Presumably your Thread.sleep() call actually waits long enough for all the threads (that you effectively didn't join) to finish. Without it, the threads might not all have finished when you print out the value of x.
There are two problems here:
a race condition where the main thread finishes before all the worker threads.
a memory visibility issue where the main thread is not guaranteed to see the updated value of x.
Thread#join is implemented using Object#wait. The condition variable used is the alive flag on the Thread:
groovy:000> new Thread().isAlive()
===> false
Thread.join is checking the alive flag before the thread has started, so isAlive returns false and join returns before the thread can start. The counter still gets incremented eventually, but since the join doesn't happen for that thread then the main thread may be printing out the results for x before all the threads can execute.
Adding the sleep gives all the threads enough time to finish up that x is what you expect by the time that the main thread prints it out.
In addition to the race condition, there is a memory visibility issue since the main thread is accessing x directly and is not using the same lock as the other threads. You should add an accessor to your Runnable using the synchronized keyword:
public class Test implements Runnable{
private int x;
public synchronized void run(){
x++;
}
public synchronized int getX() {
return x;
}
and change the main method to use the accessor:
System.out.println(test.getX());
Memory visibility issues may not be apparent since they depend on how aggressive the JVM is about caching and optimizing. If your code runs against a different JVM implementation in production, and you don't adequately guard against these issues, you may see errors there that you can't reproduce locally on a PC.
Using AtomicInteger would simplify this code and allow solving the memory visibility problem while removing synchronization.
You don't add synchronized to the run method. Each thread gets its own.
You have to synchronize the mutable, shared data. In your case, that's the integer x. You can synchronize get/set or use AtomicInteger.
I have made a method to send an e-mail, and I wanted to try if it was possible to call a method inside the method itself with a timer, to do like a scheduler
public void createExcel(){
int year = Calendar.getInstance().get(Calendar.YEAR);
int num_week = data.getCurrentWeek()-1;
ArrayList<DHDemande> ListeDemandes = d.getDemandesForPaie(num_week, year);
try {
data.createFile(ListeDemandes);
Thread.sleep(20000);
createExcel();
} catch(InterruptedException ex) {
Thread.currentThread().interrupt();
}
}
But now the method doesn't stop (it was obvious) but even if I refresh Apache and if I change the method. How can I stop it ? because I receive an email every 20 second now
The thing with Treads is that there is now save way to tell it to just stop without memory leaks. You can use thread.stop(), this will kill the thread but it may cause memory problems if your objects are too big.
Quote from java doc:
stop() Deprecated. This method is inherently unsafe. Stopping a
thread with Thread.stop causes it to unlock all of the monitors that
it has locked (as a natural consequence of the unchecked ThreadDeath
exception propagating up the stack). If any of the objects previously
protected by these monitors were in an inconsistent state, the damaged
objects become visible to other threads, potentially resulting in
arbitrary behavior. Many uses of stop should be replaced by code that
simply modifies some variable to indicate that the target thread
should stop running. The target thread should check this variable
regularly, and return from its run method in an orderly fashion if the
variable indicates that it is to stop running. If the target thread
waits for long periods (on a condition variable, for example), the
interrupt method should be used to interrupt the wait. For more
information, see Why are Thread.stop, Thread.suspend and Thread.resume
Deprecated?.
Explore the Thread api and see if you can find anything else that suites your needs.
As Borislav mentioned, stop() is incorrect here, you need to call and handle interrupt().
class Test
{
public static Thread helloWorldLater()
{
Thread t = new Thread(new Runnable(){
#Override public void run()
{
try {
Thread.sleep(200);
System.out.println("Hello World");
}
catch(InterruptedException ex)
{
// clean up here
return;
}
}
});
t.start();
return t;
}
public static void main(String[] args) throws InterruptedException
{
Thread t = helloWorldLater();
Thread.sleep(100);
t.interrupt();
// no "Hello World" to be seen
}
}
Alternatively, depending on your Java version, you can also use the more modern concurrency Java APIs.
Besides Borislavs and Konrads legitimate comments on thread.stop(), you are calling the createExcel() method recursively (after Thread.sleep(20000)), so obviously your method will run forever. If you remove that call it should work like you expect it.