为了体会守护线程的作用,我做了一个下载文件的demo,可以没事用来测测网速什么的,其特性如下
1、一共有三个线程,分别是主线程,下载线程,守护线程
2、主线程启动下载线程和守护线程
3、下载线程连续下载100个文件,如果出现异常自动捕获并进入下一个文件的下载
4、如果下载线程下载某个文件超过了30秒,就认为是超时,而这个超时的检测由守护线程执行
5、如果守护线程发现某个文件下载超时,就停掉下载线程,并想办法另起一个新的下载线程继续下载
首先我们不使用Java设置的守护线程,而是用一个普通线程充当“守护线程”,也就是设置一个计数器,下载成功一个文件就+1,加到100后这个伪“守护线程”就会自我终止,代码如下
package zz.vc.qduedu;
import java.io.BufferedInputStream;
import java.io.BufferedOutputStream;
import java.io.DataOutputStream;
import java.io.File;
import java.io.FileOutputStream;
import java.io.IOException;
import java.net.HttpURLConnection;
import java.net.URL;
import javax.net.ssl.HttpsURLConnection;
public class DownloadTest {
public static long totalDownloadTime;//全部文件共消耗的下载时间
public static long downloadStartTime;//某个文件下载开始的时间
public static DownloadThread downloadThread;
public static int completedFileCount = 0;//下载成功的文件数
public static final String url = "https://d25hz2nnwuc2tq.cloudfront.net/images/image/cache/data/2016/10/19/1476866804-800x450-1080x608.webp";
public static final int total_times = 100;
public static void main(String[] args) throws IOException {
// TODO Auto-generated method stub
downloadThread = new DownloadThread();
downloadThread.start();
AlxDemonThread demonThread = new AlxDemonThread();
demonThread.start();
}
public static class DownloadThread extends Thread{
public boolean shouldStop = false;//超时后终结此进程
public void run() {
for(int i=completedFileCount;i<total_times;i++){
if(shouldStop)return;
try {
long cost = downloadNow(url, i);
if(shouldStop)return;
System.out.println("第"+i+"次耗时"+cost);
} catch (Exception e) {
// TODO: handle exception
if(shouldStop)return;
System.out.println("下载失败");
e.printStackTrace();
}
if(shouldStop)return;
completedFileCount++;
totalDownloadTime += System.currentTimeMillis() - downloadStartTime;
downloadStartTime = 0;
}
if(!shouldStop)System.out.println("总耗时=="+totalDownloadTime);
}
}
public static class AlxDemonThread extends Thread{
@Override
public void run() {
// TODO Auto-generated method stub
super.run();
while (2>1) {
try {
Thread.sleep(1000);
// System.out.println("守护线程还活着");
} catch (InterruptedException e) {
}
if(completedFileCount == total_times)return;
if(System.currentTimeMillis() - downloadStartTime > 30*1000){
// System.out.println("第"+alreadyTime+"超时了");
System.out.println("the "+completedFileCount+" time out of time");
downloadThread.shouldStop = true;
downloadThread.interrupt();
downloadThread = new DownloadThread();
downloadThread.start();
completedFileCount++;
totalDownloadTime += 30*1000;
}
}
}
}
public static long downloadNow(String strUrl,int times) throws IOException{
downloadStartTime = System.currentTimeMillis();
URL url = new URL(strUrl);
HttpsURLConnection httpURLConnection = (HttpsURLConnection) url.openConnection();
httpURLConnection.setDoInput(true);
httpURLConnection.setUseCaches(false);
httpURLConnection.setRequestMethod("GET");
httpURLConnection.setRequestProperty("Connection", "Keep-Alive");
httpURLConnection.setRequestProperty("Charset", "UTF-8");
httpURLConnection.setRequestProperty("Content-Type","multipart/form-data;boundary=" + "******");
httpURLConnection.connect();
// System.out.println("响应码是::"+httpURLConnection.getResponseCode());
File pictureFile = new File("d:/"+times+".jpg");
BufferedInputStream bis = new BufferedInputStream(httpURLConnection.getInputStream());
FileOutputStream outputStream = new FileOutputStream(pictureFile);
byte[] buffer = new byte[1024];
int len = 0;
BufferedOutputStream out = new BufferedOutputStream(outputStream);
while ((len = bis.read(buffer)) != -1) {
if(System.currentTimeMillis() - downloadStartTime > 30*1000)throw new RuntimeException("超时");
out.write(buffer, 0, len);
}
out.flush();
out.close();
bis.close();
return System.currentTimeMillis()-downloadStartTime;
}
}
运行上面的程序,可以在控制台看见每次文件下载的耗时,如果出现了异常,并不会中断子线程,而是会开始下载下一个文件
但是如果伪“守护线程”发现了某个文件下载超时,就会调用downloadTread.interrupt()方法停止当前的下载线程并且重新new一个下载线程继续之前的进度下载。
在这个地方,我发现
Thread.interrupt()方法并不会立即停止当前线程,当前线程会继续运行十几秒的时间
我的解决方法是修改Tread里的一个私有变量shouldStop,在跨过耗时操作之后检查这个私有变量,如果被修改过那么就return掉当前线程。
现在,我们斗胆把这个伪“守护线程”编程真的守护线程,用setDaemon(true)在start()之前实现。
public static void main(String[] args) throws IOException {
// TODO Auto-generated method stub
downloadThread = new DownloadThread();
downloadThread.start();
AlxDemonThread demonThread = new AlxDemonThread();
demonThread.setDaemon(true);
demonThread.start();
}
然后把我们用于停止伪“守护线程”的计数器去掉
//if(completedFileCount == total_times)return;
然后重新执行这个程序
1、如果下载很顺畅没有任何一个文件超时,那么守护线程会随着下载线程的终止而终止
2、如果有一个文件下载超时了,守护线程会自动new线程,但是这个线程和守护线程跑了没一会就停了,守护线程退出。
得出如下几个结论
1、守护线程由主线程创建,守护主线程,同样也守护主线程创建的所有子线程。
2、主线程如果终止(包括crash),由主线程创建的子线程并不会终止,守护线程也不会终止。
3、守护线程创建的所有线程都是守护线程,当父守护线程守护的线程终止,父守护线程及其创建的子守护线程都会一起停止
第三条有点拗口,简单解释一下就是:当守护线程把下载超时的线程停止掉之后,这个下载线程并不会立即停掉,此时守护线程又创建了一个下载线程,而这个由守护线程创建的下载线程并不好用,它会莫名其妙的随着上个下载线程的终止而终止。原因就是这个新下载线程其实时候守护线程,守护主线程(其实就是守护父守护线程,也就间接守护旧的下载线程),所以没等下完就自动终止了。
那么看来通过简单粗暴的设置守护线程并不好用,那么我们根据“守护线程会守护创建它的线程”的原理,让下载线程创建一个线程守护它自己,当这个守护线程发现下载线程超时以后,通知主线程创建一个新的下载线程进行下载,同时也有一个新的守护线程守护它
import java.io.BufferedInputStream;
import java.io.BufferedOutputStream;
import java.io.DataOutputStream;
import java.io.File;
import java.io.FileOutputStream;
import java.io.IOException;
import java.net.HttpURLConnection;
import java.net.URL;
import java.util.Random;
import javax.net.ssl.HttpsURLConnection;
public class CopyOfDownloadTest {
public static long total;
public static long startTime;
public static int alreadyTime = 0;
public static final String url = "https://d25hz2nnwuc2tq.cloudfront.net/images/image/cache/data/2016/10/19/1476866804-800x450-1080x608.webp";
public static final int total_times = 100;
private static Random threadLock = new Random();//线程锁
public static void main(String[] args) throws IOException, InterruptedException {
// TODO Auto-generated method stub
new DownloadThread().start();
while (1<2) {
synchronized (threadLock) {//让主线程持有线程锁
threadLock.wait();//锁住主线程
}
System.out.println("主线程解锁,准备重新new下载线程");
new DownloadThread().start();
}
}
public static class DownloadThread extends Thread{
public boolean shouldStop = false;//超时后终结次进程
public void run() {
AlxDemonThread demonThread = new AlxDemonThread(this);
demonThread.setDaemon(true);
demonThread.start();
for(int i=alreadyTime;i<total_times;i++){
if(shouldStop)return;
try {
long cost = downloadNow(url, i);
if(shouldStop)return;
System.out.println("第"+i+"次耗时"+cost);
} catch (Exception e) {
// TODO: handle exception
if(shouldStop)return;
System.out.println("下载失败");
e.printStackTrace();
}
if(shouldStop)return;
alreadyTime++;
total += System.currentTimeMillis() - startTime;
startTime = 0;
}
if(!shouldStop)System.out.println("总耗时=="+total);
}
}
public static class AlxDemonThread extends Thread{
private DownloadThread mDownloadThread = null;
public AlxDemonThread(DownloadThread t) {
// TODO Auto-generated constructor stub
this.mDownloadThread = t;
}
@Override
public void run() {
// TODO Auto-generated method stub
super.run();
while (2>1) {
try {
Thread.sleep(1000);
// System.out.println("守护线程还活着");
} catch (InterruptedException e) {
}
if(alreadyTime == total_times)return;
if(System.currentTimeMillis() - startTime > 30*1000){
System.out.println("第"+alreadyTime+"超时了");
mDownloadThread.shouldStop = true;
mDownloadThread.interrupt();
alreadyTime++;
total += 30*1000;
synchronized (threadLock) {
threadLock.notify();
}
return;//停掉守护线程,防止再new一个子线程
}
//因为是守护线程,所以在下载线程结束后会自动停止
}
}
}
public static long downloadNow(String strUrl,int times) throws IOException{
startTime = System.currentTimeMillis();
URL url = new URL(strUrl);
HttpsURLConnection httpURLConnection = (HttpsURLConnection) url.openConnection();
httpURLConnection.setDoInput(true);
httpURLConnection.setUseCaches(false);
httpURLConnection.setRequestMethod("GET");
httpURLConnection.setRequestProperty("Connection", "Keep-Alive");
httpURLConnection.setRequestProperty("Charset", "UTF-8");
httpURLConnection.setRequestProperty("Content-Type","multipart/form-data;boundary=" + "******");
httpURLConnection.connect();
// System.out.println("响应码是::"+httpURLConnection.getResponseCode());
File pictureFile = new File("d:/speed/test"+times+".jpg");
BufferedInputStream bis = new BufferedInputStream(httpURLConnection.getInputStream());
FileOutputStream outputStream = new FileOutputStream(pictureFile);
byte[] buffer = new byte[1024];
int len = 0;
BufferedOutputStream out = new BufferedOutputStream(outputStream);
while ((len = bis.read(buffer)) != -1) {
if(System.currentTimeMillis() - startTime > 30*1000)throw new RuntimeException("超时");
out.write(buffer, 0, len);
}
out.flush();
out.close();
bis.close();
return System.currentTimeMillis()-startTime;
}
}
上面的程序中,守护线程由下载子线程创建,守护下载子线程。当守护线程发现下载线程超时之后,就会终止下载线程,然后通知主线程再重新创建一个下载线程继续之前的下载,然后这个守护线程会随着要终止的下载线程一起结束。一对新的下载线程和守护线程继续当前的下载任务。
但是上面的程序在下载完成之后并不会退出,因为主线程还在wait()中。所有的下载子线程和守护线程均退出。
从这里可以得出一个结论:
1、object.wait()和object.notify必须在synchronized(object){}代码块中执行,否则会报java.lang.IllegalMonitorStateException异常
2、一个线程wait()之后并不会结束。并且守护它的守护线程也不会结束。
3、主线程也可以wait()
4、主线程出现crash崩溃了之后,主线程开启的子线程会继续执行。由主线程产生的守护线程会守护主线程产生的子线程。