线程池之七大参数
我们知道，线程池的核心类是 ThreadPoolExecutor，构造一个ThreadPoolExecutor需要7个参数！
int corePoolSize
线程池中的常驻核心线程数
int maximumPoolSize
线程池中能够容纳同时执行的最大线程数，此值必须大于等于1
long keepAliveTime
多余的空闲线程的存活时间。当前池中的线程数量超过corePoolSize时，多余线程会被销毁到只剩下corePoolSize个线程为至
TimeUnit unit
keepAliveTime的单位
BlockingQueue<Runnable> workQueue
任务队列，被提交但未被执行的任务。由阻塞队列实现，关于阻塞队列我这篇文章有讲解 https://www.jianshu.com/p/862c93ab3203
ThreadFactory threadFactory
表示生产线程池中工作线程的工厂。用于创建线程，一般默认即可
RejectedExecutionHandler handler
拒绝策略，表示当队列满了，并且工作线程大于等于线程池的最大线程数（maximumPoolSize）时如何来拒绝请求执行的runnable饿策略。
我的这篇文章有详细的讲解拒绝策略
https://www.jianshu.com/p/bf6fabb7b459
实际上我们应该如何使用线程池？
不允许直接使用jdk自带的三种线程池
Executors.newCachedThreadPool();
Executors.newFixedThreadPool(5);
Executors.newSingleThreadExecutor();
工作中请使用自定义的写法：直接使用ThreadPoolExecutor类带上7个参数构造线程池！因为可能导致oom异常，所谓oom异常就是 java.lang.OutOfMemoryError:Java heap space
实际上请使用ThreadPoolExecutor类带上7个参数来构造线程池
-
注意指定队列的长度，我这里是new LinkedBlockingQueue<>(3)。如果不指定3，则队列长度为
Integer.MAX_VALUE,可能因为任务的积累导致oom。
image.png maximumPoolSize在cpu密级型项目中请配置成
cup核心数+1==>Runtime.getRuntime().availableProcessors()+1
maximumPoolSize在io密级型项目中请配置成cpu核心数/阻塞系数
参数调整试验
使用如下代码来构造线程池。2个核心线程，5个最大数量，3个等待队列席位。
ExecutorService executorService =
new ThreadPoolExecutor(
2,
5,
2, TimeUnit.SECONDS,
new LinkedBlockingQueue<>(3),
Executors.defaultThreadFactory(),
new ThreadPoolExecutor.AbortPolicy());
实验一，任务数量为6
import java.util.concurrent.*;
public class Test {
public static void main(String[] args) {
ExecutorService executorService = new ThreadPoolExecutor(2, 5, 2, TimeUnit.SECONDS, new LinkedBlockingQueue<>(3), Executors.defaultThreadFactory(), new ThreadPoolExecutor.AbortPolicy());
try {
for (int i = 1; i <= 6; i++) {
executorService.execute(new Runnable() {
@Override
public void run() {
System.out.println(Thread.currentThread().getName() + " 办理业务");
}
});
}
} finally {
executorService.shutdown();
}
}
}
这里有6个任务。所以线程池的核心线程、等待队列都被用完。还剩下1个任务（6-2-3=1），需要创建线程来执行。所以执行结果中打印了pool-1-thread-3，该线程即是线程池临时创建出来的线程!
实验二，任务数量为9
package com.springboot.study.demo1;
import java.util.concurrent.*;
class Test1 {
public static void main(String[] args) {
//2个核心线程 等待数量席位3个 最大线程数5个，那么最大容纳任务是8个
ExecutorService executorService = new ThreadPoolExecutor(2, 5, 2, TimeUnit.SECONDS, new LinkedBlockingQueue<>(3), Executors.defaultThreadFactory(), new ThreadPoolExecutor.AbortPolicy());
try {
//提交9次任务，超出的1个任务会被应用拒绝策略
for (int i = 1; i <= 9; i++) {
executorService.execute(new Runnable() {
@Override
public void run() {
System.out.println(Thread.currentThread().getName() + " 办理业务");
}
});
}
} finally {
executorService.shutdown();
}
}
}
执行结果：抱错。此时任务数量9超过线程池可接受的最大任务数8（maximumPoolSize 3 +workQueue的容量5）
Exception in thread "main" java.util.concurrent.RejectedExecutionException: Task Test$1@42110406 rejected from java.util.concurrent.ThreadPoolExecutor@531d72ca[Running, pool size = 5, active threads = 5, queued tasks = 3, completed tasks = 0]
创建自定义线程工厂为线程池中线程命名
创建线程池时使用自定义的线程工厂，这样可以为该线程池内的线程改名
package io.renren;
import java.util.concurrent.*;
import java.util.concurrent.atomic.AtomicInteger;
public class UserThreadFactory implements ThreadFactory {
private final String namePrefix;
private final AtomicInteger nextId = new AtomicInteger(1);
// 定义线程组名称，在 jstack 问题排查时，非常有帮助
public UserThreadFactory(String whatFeaturOfGroup) {
namePrefix = "来自 UserThreadFactory 线程工厂的 " + whatFeaturOfGroup + "-线程-";
}
@Override
public Thread newThread(Runnable task) {
String name = namePrefix + nextId.getAndIncrement();
Thread thread = new Thread(null, task, name, 0);
System.out.println(thread.getName());
return thread;
}
}
class TestFa{
public static void main(String[] args) {
UserThreadFactory userThreadFactory = new UserThreadFactory(" 用户 ");
ThreadPoolExecutor threadPoolExecutor =
new ThreadPoolExecutor(
2,
5,
2, TimeUnit.SECONDS,
new LinkedBlockingQueue<>(3),
userThreadFactory ,
new ThreadPoolExecutor.AbortPolicy());
for (int i = 0; i <8 ; i++) {
threadPoolExecutor.submit(new Runnable() {
@Override
public void run() {
System.out.println(Thread.currentThread().getName());
}
});
}
threadPoolExecutor.shutdown();
}
}
ThreadFactory 匿名
ThreadFactory threadFactory = new ThreadFactory() {
AtomicInteger nextId = new AtomicInteger(1);
@Override
public Thread newThread(Runnable task) {
Thread thread = new Thread(null, task, "测试redis分布式锁线程" + nextId.getAndIncrement(), 0);
return thread;
}
};
ThreadPoolExecutor threadPoolExecutor =
new ThreadPoolExecutor(2,
5,
3,
TimeUnit.MINUTES,
new ArrayBlockingQueue<>(5),
threadFactory,
new ThreadPoolExecutor.AbortPolicy()
);
for (int i = 0; i <10 ; i++) {
threadPoolExecutor.submit(new Runnable() {
@Override
public void run() {
System.out.println(Thread.currentThread().getName());
}
});
}
