多线程java优化

Question

在我的程序中，我尝试掌握如何使用ExecutorService来优化我的程序。由于某种原因，它在两个Urls上有点卡住了。http://sjsu.edu/和https://paypal.com。当它位于这两个站点上时，它不会继续执行其他URLS。

即使这两个域的响应不够快，其他3个可用线程是否应该继续运行呢？

如何以最好的方式解决这个问题？

public class SequentialPinger {
    public static void main(String args[]) throws Exception {

        String[] hostList = {"http://crunchify.com", "http://yahoo.com",
            "http://www.ebay.com", "http://google.com",
            "http://www.example.co", "https://paypal.com",
            "http://bing.com/", "http://techcrunch.com/",
            "http://mashable.com/", "http://thenextweb.com/",
            "http://wordpress.com/", "http://cphbusiness.dk/",
            "http://example.com/", "http://sjsu.edu/",
            "http://ebay.co.uk/", "http://google.co.uk/",
            "http://www.wikipedia.org/",
            "http://dr.dk", "http://pol.dk", "https://www.google.dk",
            "http://phoronix.com", "http://www.webupd8.org/",
            "https://studypoint-plaul.rhcloud.com/", "http://stackoverflow.com",
            "http://docs.oracle.com", "https://fronter.com",
            "http://imgur.com/", "http://www.imagemagick.org"
        };

        List<CallableImpl> callList = new ArrayList();
        ExecutorService es = Executors.newFixedThreadPool(4);

        for (String url : hostList) {
            CallableImpl callable = new CallableImpl(url);
            callList.add(callable);
        }
        for (CallableImpl callableImpl : callList) {
            System.out.println("Trying to connect to: " + callableImpl.getUrl());
            Future<String> lol = es.submit(callableImpl);
            System.out.println("status: " + lol.get());
        }

        es.shutdown();

    }
}

我的可调用实现

public class CallableImpl implements Callable<String> {

    private final String url;

    public CallableImpl(String url) {
        this.url = url;
    }

    public String getUrl() {
        return url;
    }

    @Override
    public String call() {
        String result = "Error";

        try {
            URL siteURL = new URL(url);
            HttpURLConnection connection = (HttpURLConnection) siteURL
                    .openConnection();
            connection.setRequestMethod("GET");

            connection.connect();

            int code = connection.getResponseCode();
            if (code == 200) {
                result = "Green";
            }

            if (code == 301) {
                result = "Redirect";
            }

        } catch (IOException e) {
            result = "->Red<-";
        }
        return result;
    }
}

Answer 1

在您的代码中，您一个接一个地向Callable提交ExecutorService，然后立即调用Future.get()，这将阻塞，直到结果准备就绪(或者在运行时抛出异常)。

最好用ExecutorService包装CompletionSerivce，一旦结果准备就绪，它就会提供结果。并将for-循环分成两个循环:一个提交所有Callables，另一个检查结果。

ExecutorService es = Executors.newFixedThreadPool(4);
ExecutorCompletionService<String> completionService = new ExecutorCompletionService<>(es);

for (CallableImpl callableImpl : callList) {
        System.out.println("Trying to connect to: " + callableImpl.getUrl());
        completionService.submit(callableImpl);
}
for (int i = 0; i < callList.size(); ++i) {
    completionService.take().get();   //fetch next finished Future and check its result
}

Answer 2

问题

创建get()后直接在Future上调用它，阻塞主线程。因此，您根本没有任何并行调用，使得ExecutorService基本上是无用的。您的代码相当于自己简单地调用callableImpl.call()。

溶液

如果您想要继续执行并让每个get()并行运行，请不要调用CallableImpl。相反，您可以在es.awaitTermination()之后调用es.shutdown()。

Answer 3

我建议使用在Java8中添加的CompletableFuture，并向其添加一个回调。

CompletableFuture.supplyAsync(myCallable::call, es)
        .thenAccept(result -> {
            something(result);
        });

我建议让你的可调用成为一个供应商，使这更简单。