【高并发系列】18、线程池那些事儿2 - 内部实现
线程池内部实现
1、常用线程池定义
public static ExecutorService newFixedThreadPool(int nThreads) {
return new ThreadPoolExecutor(nThreads, nThreads,
0L, TimeUnit.MILLISECONDS,
new LinkedBlockingQueue());
}
public static ExecutorService newSingleThreadExecutor() {
return new FinalizableDelegatedExecutorService
(new ThreadPoolExecutor(1, 1,
0L, TimeUnit.MILLISECONDS,
new LinkedBlockingQueue()));
}
public static ExecutorService newCachedThreadPool() {
return new ThreadPoolExecutor(0, Integer.MAX_VALUE,
60L, TimeUnit.SECONDS,
new SynchronousQueue());
} 2、ThreadPoolExecutor构造方法详解
由以上代码可以看到，常用的几个方法虽然有着完全不同的功能特点，但内部实现都是ThreadPoolExecutor类的封装，下面是ThreadPoolExecutor类的构造方法：
public ThreadPoolExecutor(int corePoolSize,
int maximumPoolSize,
long keepAliveTime,
TimeUnit unit,
BlockingQueue workQueue,
ThreadFactory threadFactory,
RejectedExecutionHandler handler) 其中：
corePoolSize
- the number of threads to keep in the pool, even if they are idle, unless allowCoreThreadTimeOut is set
- 线程池中线程数量
maximumPoolSize
- the maximum number of threads to allow in the pool
- 线程池中最大线程数量
keepAliveTime
- when the number of threads is greater than the core, this is the maximum time that excess idle threads will wait for new tasks before terminating.
- 当线程池数量超过corePoolSize时，多余的空闲线程的存活时间，即超过corePoolSize的空闲线程在多长时间内被销毁；
unit
- the time unit for the keepAliveTime argument
- keepAliveTime的单位
workQueue
- the queue to use for holding tasks before they are executed. This queue will hold only the Runnable tasks submitted by the execute method.
- 任务队列，被提交但尚未被执行的任务存放到该队列；
threadFactory
- the factory to use when the executor creates a new thread
- 线程工厂，用于创建线程，一般使用默认值；
handler
- the handler to use when execution is blocked because the thread bounds and queue capacities are reached
- 拒绝策略；当任务太多来不及处理时如何拒绝任务；
2.1 workQueue
指被提交但未执行的任务队列；是java.util.concurrent.BlockingQueue接口对象，仅用于存放Runnable对象；分类如下：
- 直接提交的队列 - SynchronousQueue
SynchronousQueue没有容量，每一个插入操作都要等待一个相应的删除操作，反之，每一个删除操作都要等待对应的插入操作；
提交的任务总是被提交给线程执行，如果没有空闲的线程则尝试新建线程，如果已经达到线程最大值则执行拒绝策略；
使用SynchronousQueue队列，通常设置很大的maximumPoolSize值，否则很容易执行拒绝策略；
如newCachedThreadPool()方法中：
public static ExecutorService newCachedThreadPool() {
return new ThreadPoolExecutor(0, Integer.MAX_VALUE,
60L, TimeUnit.SECONDS,
new SynchronousQueue());
} - 有界的任务队列 - ArrayBlockingQueue
构造方法中需传入队列最大容量值
public ArrayBlockingQueue(int capacity)当有新任务需要执行时：
- ）线程池实际线程数小于corePoolSize，则优先创建新线程；
- ）线程池实际线程数大于corePoolSize，若等待队列未满可以加入，则将新任务加入等待队列；
- ）若等待队列已满无法加入，并且总线程数不大于maximumPoolSize时，创建新线程执行任务；
- ）若等待队列已满无法加入，并且总线程数大于maximumPoolSize时，执行拒绝策略；
- 无界的任务队列 - LinkedBlockingQueue
与有界队列相比，除非系统资源耗尽，否则无界任务队列不存在任务入队失败的情况；
当有新任务需要执行时：
- ）线程池实际线程数小于corePoolSize，则优先创建新线程；
- ）线程池实际线程数大于corePoolSize，则将新任务加入等待队列，并且线程数不会增加了；
如newFixedThreadPool()方法中：
public static ExecutorService newFixedThreadPool(int nThreads) {
return new ThreadPoolExecutor(nThreads, nThreads,
0L, TimeUnit.MILLISECONDS,
new LinkedBlockingQueue());
} 如newSingleThreadExecutor()方法中：
public static ExecutorService newSingleThreadExecutor() {
return new FinalizableDelegatedExecutorService
(new ThreadPoolExecutor(1, 1,
0L, TimeUnit.MILLISECONDS,
new LinkedBlockingQueue()));
} - 优先任务队列 - PriorityBlockingQueue
可以根据任务自身的优先级顺序控制执行先后顺序；
attention：使用自定义线程池时，要根据具体情况选择合适的并发队列作为任务的缓冲；
3、ThreadPoolExecutor.execute()方法
/**
* Executes the given task sometime in the future. The task
* may execute in a new thread or in an existing pooled thread.
*
* If the task cannot be submitted for execution, either because this
* executor has been shutdown or because its capacity has been reached,
* the task is handled by the current {@code RejectedExecutionHandler}.
*
* @param command the task to execute
* @throws RejectedExecutionException at discretion of
* {@code RejectedExecutionHandler}, if the task
* cannot be accepted for execution
* @throws NullPointerException if {@code command} is null
*/
public void execute(Runnable command) {
if (command == null)
throw new NullPointerException();
/*
* Proceed in 3 steps:
*
* 1. If fewer than corePoolSize threads are running, try to
* start a new thread with the given command as its first
* task. The call to addWorker atomically checks runState and
* workerCount, and so prevents false alarms that would add
* threads when it shouldn't, by returning false.
*
* 2. If a task can be successfully queued, then we still need
* to double-check whether we should have added a thread
* (because existing ones died since last checking) or that
* the pool shut down since entry into this method. So we
* recheck state and if necessary roll back the enqueuing if
* stopped, or start a new thread if there are none.
*
* 3. If we cannot queue task, then we try to add a new
* thread. If it fails, we know we are shut down or saturated
* and so reject the task.
*/
int c = ctl.get();
if (workerCountOf(c) < corePoolSize) {
if (addWorker(command, true))
return;
c = ctl.get();
}
if (isRunning(c) && workQueue.offer(command)) {
int recheck = ctl.get();
if (! isRunning(recheck) && remove(command))
reject(command);
else if (workerCountOf(recheck) == 0)
addWorker(null, false);
}
else if (!addWorker(command, false))
reject(command);
}先判断当前线程数是否小于corePoolSize核心线程数(workerCountOf(c) < corePoolSize)；
如果小于，则通过addWorker(command, true)直接调度执行；
如果不小于，加入等待队列(workQueue.offer(command))；
如果加入等待队列失败（如有界队列到达上限，或者使用了SynchronousQueue类等），则将任务直接提交给线程池执行else if (!addWorker(command, false))；
如果提交失败，执行拒绝策略reject(command)；
4、拒绝策略
是系统超负荷运行时的补救措施；
JDK内置拒绝策略，都实现了java.util.concurrent.RejectedExecutionHandler接口：
- AbortPolicy：直接抛出异常，阻止系统正常工作；
-
CallerRunsPolicy：只要线程池未关闭，该策略会直接在调用者线程中，运行当前被丢弃的任务；
-
DiscardOldestPolicy：丢弃未执行任务中最老的任务，就是即将被执行的一个任务，并尝试再次提交当前任务；
-
DiscardPolicy：默默丢弃无法处理的任务，不做任何处理；
如果内置策略无法满足实际应用的需求，可以自己扩展RejectedExecutionHandler接口：
public interface RejectedExecutionHandler {
/**
* Method that may be invoked by a {@link ThreadPoolExecutor} when
* {@link ThreadPoolExecutor#execute execute} cannot accept a
* task. This may occur when no more threads or queue slots are
* available because their bounds would be exceeded, or upon
* shutdown of the Executor.
*
* In the absence of other alternatives, the method may throw
* an unchecked {@link RejectedExecutionException}, which will be
* propagated to the caller of {@code execute}.
*
* @param r the runnable task requested to be executed
* @param executor the executor attempting to execute this task
* @throws RejectedExecutionException if there is no remedy
*/
void rejectedExecution(Runnable r, ThreadPoolExecutor executor);
}
示例：核心线程、最大线程数都是5，有界任务队列容量为10，所以最多可以容纳15个线程，多余的走拒绝策略；
public class RejectExecutionHandlerDemo {
public static class TaskDemo implements Runnable {
@Override
public void run() {
System.out.println(System.currentTimeMillis() + " : " + Thread.currentThread().getName());
try {
TimeUnit.MILLISECONDS.sleep(1000);
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
public static void main(String[] args) throws InterruptedException {
TaskDemo task = new TaskDemo();
ExecutorService es = new ThreadPoolExecutor(5, 5, 0L, TimeUnit.MILLISECONDS, new LinkedBlockingQueue<>(10),
new RejectedExecutionHandler() {
@Override
public void rejectedExecution(Runnable r, ThreadPoolExecutor executor) {
System.out.println(r.toString() + " is discard.");
}
});
for (int i = 0; i < 16; i++) {
es.submit(task);
TimeUnit.MILLISECONDS.sleep(10);
}
es.shutdown();
}
}console
1550415828931 : pool-1-thread-1
1550415828943 : pool-1-thread-2
1550415828954 : pool-1-thread-3
1550415828964 : pool-1-thread-4
1550415828975 : pool-1-thread-5
java.util.concurrent.FutureTask@7d4991ad is discard.
1550415829933 : pool-1-thread-1
1550415829943 : pool-1-thread-2
1550415829955 : pool-1-thread-3
1550415829969 : pool-1-thread-4
1550415829978 : pool-1-thread-5
1550415830936 : pool-1-thread-1
1550415830944 : pool-1-thread-2
1550415830959 : pool-1-thread-3
1550415830974 : pool-1-thread-4
1550415830979 : pool-1-thread-55、自定义线程工厂 ThreadFactory
java.util.concurrent.ThreadFactory接口用于创建线程；
public interface ThreadFactory {
/**
* Constructs a new {@code Thread}. Implementations may also initialize
* priority, name, daemon status, {@code ThreadGroup}, etc.
*
* @param r a runnable to be executed by new thread instance
* @return constructed thread, or {@code null} if the request to
* create a thread is rejected
*/
Thread newThread(Runnable r);
}示例：自定义线程工厂，设置所有线程为守护线程，当main线程退出后，强制注销该线程池；
public class ThreadFactoryDemo {
public static class TaskDemo implements Runnable {
@Override
public void run() {
System.out.println(System.currentTimeMillis() + " : " + Thread.currentThread().getName());
try {
TimeUnit.MILLISECONDS.sleep(1000);
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
public static void main(String[] args) {
TaskDemo task = new TaskDemo();
ExecutorService es = new ThreadPoolExecutor(5, 5, 0L, TimeUnit.MILLISECONDS, new SynchronousQueue(),
new ThreadFactory() {
@Override
public Thread newThread(Runnable r) {
Thread t = new Thread(r);
t.setDaemon(true);
System.out.println("create " + t);
return t;
}
});
for (int i = 0; i < 5; i++) {
es.submit(task);
}
}
} console
create Thread[Thread-0,5,main]
create Thread[Thread-1,5,main]
create Thread[Thread-2,5,main]
1550416438149 : Thread-1
1550416438149 : Thread-0
create Thread[Thread-3,5,main]
1550416438150 : Thread-2
create Thread[Thread-4,5,main]
1550416438150 : Thread-3
1550416438150 : Thread-46、线程池扩展
可以覆写ThreadPoolExecutor中的以下三个方法，来对线程池进行控制：
protected void beforeExecute(Thread t, Runnable r) { }
protected void afterExecute(Runnable r, Throwable t) { }
protected void terminated() { }
public class ExtThreadPool {
public static class TaskDemo implements Runnable {
private String name;
public TaskDemo(String name) {
this.name = name;
}
@Override
public void run() {
System.out.println(Thread.currentThread().getName() + " - " + this.name + " - is running.");
try {
TimeUnit.MILLISECONDS.sleep(100);
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
public static void main(String[] args) throws InterruptedException {
ExecutorService es = new ThreadPoolExecutor(5, 5, 0L, TimeUnit.MILLISECONDS,
new LinkedBlockingQueue()) {
@Override
protected void beforeExecute(Thread t, Runnable r) {
System.out.println(t.getName() + " - " + ((TaskDemo) r).name + " - beforeExecute");
}
@Override
protected void afterExecute(Runnable r, Throwable t) {
System.out.println(Thread.currentThread().getName() + " - " + ((TaskDemo) r).name + " - afterExecute");
}
@Override
protected void terminated() {
System.out.println("ThreadPool is terminated.");
}
};
for (int i = 0; i < 5; i++) {
TaskDemo task = new TaskDemo("TASK-" + i);
es.execute(task);
TimeUnit.MILLISECONDS.sleep(10);
}
es.shutdown();
}
} console
pool-1-thread-1 - TASK-0 - beforeExecute
pool-1-thread-1 - TASK-0 - is running.
pool-1-thread-2 - TASK-1 - beforeExecute
pool-1-thread-2 - TASK-1 - is running.
pool-1-thread-3 - TASK-2 - beforeExecute
pool-1-thread-3 - TASK-2 - is running.
pool-1-thread-4 - TASK-3 - beforeExecute
pool-1-thread-4 - TASK-3 - is running.
pool-1-thread-5 - TASK-4 - beforeExecute
pool-1-thread-5 - TASK-4 - is running.
pool-1-thread-1 - TASK-0 - afterExecute
pool-1-thread-2 - TASK-1 - afterExecute
pool-1-thread-3 - TASK-2 - afterExecute
pool-1-thread-4 - TASK-3 - afterExecute
pool-1-thread-5 - TASK-4 - afterExecute
ThreadPool is terminated.7、优化线程池线程数量
Ncpu = CPU数量
Ucpu = 目标CPU使用率， 0<=Ucpu<=1
W/C = 等待时间与计算时间的比率为保持处理器达到期望的使用率，最优的线程池大小为
Nthreads = Ncpu * Ucpu * (1 + W/C)在Java中，通过如下代码获取可用CPU数量：
Runtime.getRuntime().availableProcessors()8、打印线程池中的错误堆栈信息
public class TraceErrorDemo {
public static class DivTask implements Runnable {
private int a, b;
public DivTask(int a, int b) {
this.a = a;
this.b = b;
}
@Override
public void run() {
double result = a / b;
System.out.println(result);
}
}
public static void main(String[] args) throws InterruptedException, ExecutionException {
ThreadPoolExecutor pools = new ThreadPoolExecutor(0, Integer.MAX_VALUE, 0L, TimeUnit.MILLISECONDS,
new SynchronousQueue<>());
for (int i = 0; i < 5; i++) {
pools.submit(new DivTask(100, i));
}
}
}console
100.0
25.0
33.0
50.0只有4个输出，把i=0的情况漏掉了，0不能作为除数，但没有任何日志，没有错误提示，好像程序正常执行了一样；
找回部分异常信息：
1、放弃submit()方法，改用execute()方法；
pools.execute(new DivTask(100, i));2、使用Future接收submit()方法返回值；
Future result = pools.submit(new DivTask(100, i));
result.get();得到部分异常堆栈信息如下：
Exception in thread "pool-1-thread-1" java.lang.ArithmeticException: / by zero
at com.threadpool.TraceErrorDemo$DivTask.run(TraceErrorDemo.java:18)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:748)
100.0
25.0
33.0
50.0但是还是不知道任务是在哪里提交的，提交位置已经被线程池淹没了，所以自己动手扩展ThreadPoolExecutor线程池，在调度任务之前先保存一下提交任务线程的堆栈信息：
public class TraceThreadPoolExecutor extends ThreadPoolExecutor {
public TraceThreadPoolExecutor(int corePoolSize, int maximumPoolSize, long keepAliveTime, TimeUnit unit,
BlockingQueue workQueue) {
super(corePoolSize, maximumPoolSize, keepAliveTime, unit, workQueue);
}
@Override
public void execute(Runnable command) {
super.execute(wrap(command, clientTrace(), Thread.currentThread().getName()));
}
@Override
public Future submit(Runnable command) {
return super.submit(wrap(command, clientTrace(), Thread.currentThread().getName()));
}
private Runnable wrap(final Runnable task, final Exception clientStack, String clientThreadName) {
return new Runnable() {
@Override
public void run() {
try {
task.run();
} catch (Exception e) {
clientStack.printStackTrace();
throw e;
}
}
};
}
private Exception clientTrace() {
return new Exception("Client stack trace");
}
public static class DivTask implements Runnable {
private int a, b;
public DivTask(int a, int b) {
this.a = a;
this.b = b;
}
@Override
public void run() {
double result = a / b;
System.out.println(result);
}
}
public static void main(String[] args) {
ThreadPoolExecutor pools = new TraceThreadPoolExecutor(0, Integer.MAX_VALUE, 0L, TimeUnit.MILLISECONDS,
new SynchronousQueue<>());
for (int i = 0; i < 5; i++) {
pools.execute(new DivTask(100, i));
}
}
} console
java.lang.Exception: Client stack trace
at com.threadpool.TraceThreadPoolExecutor.clientTrace(TraceThreadPoolExecutor.java:36)
at com.threadpool.TraceThreadPoolExecutor.execute(TraceThreadPoolExecutor.java:16)
at com.threadpool.TraceThreadPoolExecutor.main(TraceThreadPoolExecutor.java:54)
Exception in thread "pool-1-thread-1" java.lang.ArithmeticException: / by zero
at com.threadpool.TraceThreadPoolExecutor$DivTask.run(TraceThreadPoolExecutor.java:46)
at com.threadpool.TraceThreadPoolExecutor$1.run(TraceThreadPoolExecutor.java:27)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:748)
100.0
25.0
33.0
50.09、Fork/Join框架
“分而治之”的思想体现；
ForkJoinPool线程池对fork/join框架支持的接口
public ForkJoinTask submit(ForkJoinTask task) ForkJoinTask有两个重要的子类：RecursiveAction和RecursiveTask，分别表示没有返回值和可以携带返回值的任务；
public abstract class RecursiveAction extends ForkJoinTask
public abstract class RecursiveTask extends ForkJoinTask 示例：计算数列求和
public class CountTask extends RecursiveTask {
private static final long serialVersionUID = 8000196622093620501L;
// 任务分解规模
private static final int THRESHOLD = 10000;
private long start;
private long end;
public CountTask(long start, long end) {
this.start = start;
this.end = end;
}
@Override
protected Long compute() {
long sum = 0;
boolean canCompute = (end - start) < THRESHOLD;
if (canCompute) {
for (long i = start; i <= end; i++) {
sum += i;
}
} else {
// 任务再次分解
long step = (start + end) / 100;
List subTasks = new ArrayList<>();
long pos = start;
for (int i = 0; i < 100; i++) {
long lastOne = pos + step;
if (lastOne > end) {
lastOne = end;
}
CountTask subTask = new CountTask(pos, lastOne);
pos += step + 1;
subTasks.add(subTask);
// 提交子任务
subTask.fork();
}
// 等待所有子任务结束，并将结果再次求和
for (CountTask t : subTasks) {
sum += t.join();
}
}
return sum;
}
public static void main(String[] args) {
ForkJoinPool forkJoinPool = new ForkJoinPool();
CountTask task = new CountTask(0, 200000L);
ForkJoinTask result = forkJoinPool.submit(task);
try {
long res = result.get();
System.out.println(res);
} catch (InterruptedException e) {
e.printStackTrace();
} catch (ExecutionException e) {
e.printStackTrace();
}
}
} console
2000010000010、Guava中对线程池的扩展
10.1 DirectExecutor
MoreExecutors中的DirectExecutor，将任务在当前线程中直接执行；
使用统一的编码风格来处理同步和异步调用，简化设计：
public class DirectExecutorDemo {
public static void main(String[] args) {
Executor exec = MoreExecutors.directExecutor();
exec.execute(() -> System.out.println("I am runnint in " + Thread.currentThread().getName() + "."));
}
}console
I am runnint in main.10.2 将普通线程池转为Daemon线程池
使用MoreExecutors中的如下方法：
public static ExecutorService getExitingExecutorService(ThreadPoolExecutor executor)public class DaemonThreadPoolDemo {
public static void main(String[] args) {
ThreadPoolExecutor exec = (ThreadPoolExecutor) Executors.newFixedThreadPool(2);
// 如果不设置，程序不会退出
MoreExecutors.getExitingExecutorService(exec);
exec.execute(() -> System.out.println("I am running in " + Thread.currentThread().getName() + "."));
}
}