Spring線程池ThreadPoolTaskExecutor配置及詳情

1. ThreadPoolTaskExecutor配置

<!-- spring thread pool executor -->           
    <bean id="taskExecutor" class="org.springframework.scheduling.concurrent.ThreadPoolTaskExecutor">
        <!-- 線程池維護線程的最少數量 -->
        <property name="corePoolSize" value="5" />
        <!-- 允許的空閑時間 -->
        <property name="keepAliveSeconds" value="200" />
        <!-- 線程池維護線程的最大數量 -->
        <property name="maxPoolSize" value="10" />
        <!-- 緩存隊列 -->
        <property name="queueCapacity" value="20" />
        <!-- 對拒絕task的處理策略 -->
        <property name="rejectedExecutionHandler">
            <bean class="java.util.concurrent.ThreadPoolExecutor$CallerRunsPolicy" />
        </property>
    </bean>

屬性字段說明

corePoolSize：線程池維護線程的最少數量

keepAliveSeconds：允許的空閑時間

maxPoolSize：線程池維護線程的最大數量

queueCapacity：緩存隊列

rejectedExecutionHandler：對拒絕task的處理策略

2. execute(Runable)方法執行過程

如果此時線程池中的數量小於corePoolSize，即使線程池中的線程都處於空閑狀態，也要創建新的線程來處理被添加的任務。

如果此時線程池中的數量等於 corePoolSize，但是緩沖隊列 workQueue未滿，那么任務被放入緩沖隊列。

如果此時線程池中的數量大於corePoolSize，緩沖隊列workQueue滿，並且線程池中的數量小於maxPoolSize，建新的線程來處理被添加的任務。

如果此時線程池中的數量大於corePoolSize，緩沖隊列workQueue滿，並且線程池中的數量等於maxPoolSize，那么通過handler所指定的策略來處理此任務。也就是：處理任務的優先級為：核心線程corePoolSize、任務隊列workQueue、最大線程 maximumPoolSize，如果三者都滿了，使用handler處理被拒絕的任務。

當線程池中的線程數量大於corePoolSize時，如果某線程空閑時間超過keepAliveTime，線程將被終止。這樣，線程池可以動態的調整池中的線程數。

3. 示例代碼

Junit Test

@RunWith(SpringJUnit4ClassRunner.class)
@ContextConfiguration(classes = { MultiThreadConfig.class })
public class MultiThreadTest {

    @Autowired
    private ThreadPoolTaskExecutor taskExecutor;

    @Autowired
    private MultiThreadProcessService multiThreadProcessService;
    
    @Test
    public void test() {
        
        int n = 20;
        for (int i = 0; i < n; i++) {
            taskExecutor.execute(new MultiThreadDemo(multiThreadProcessService));
            System.out.println("int i is " + i + ", now threadpool active threads totalnum is " + taskExecutor.getActiveCount());
        }
        
        try {
            System.in.read();
        } catch (IOException e) {
            throw new RuntimeException(e);
        }
    }
}

MultiThreadDemo

/**
 * 多線程並發處理demo
 * @author daniel.zhao
 *
 */
public class MultiThreadDemo implements Runnable {

    private MultiThreadProcessService multiThreadProcessService;
    
    public MultiThreadDemo() {
    }
    
    public MultiThreadDemo(MultiThreadProcessService multiThreadProcessService) {
        this.multiThreadProcessService = multiThreadProcessService;
    }
    
    @Override
    public void run() {
        multiThreadProcessService.processSomething();
    }

}

MultiThreadProcessService

@Service
public class MultiThreadProcessService {

    public static final Logger logger = Logger.getLogger(MultiThreadProcessService.class);
    
    /**
     * 默認處理流程耗時1000ms
     */
    public void processSomething() {
        logger.debug("MultiThreadProcessService-processSomething" + Thread.currentThread() + "......start");
        try {
            Thread.sleep(1000);
        } catch (InterruptedException e) {
            throw new RuntimeException(e);
        }
        logger.debug("MultiThreadProcessService-processSomething" + Thread.currentThread() + "......end");
    }
}

MultiThreadConfig

@Configuration
@ComponentScan(basePackages = { "com.xxx.multithread" })
@ImportResource(value = { "classpath:config/application-task.xml" })
@EnableScheduling
public class MultiThreadConfig {
}