记一次线程堵塞

程序的主流程是 线程池A fetch数据，然后线程池 B delete数据

	public void init() {
		final String fetchPoolName = "fetch-pool";
		fetchVsearchDataExecutor = new ThreadPoolExecutor(4, threadNum, 10, TimeUnit.SECONDS,
				new ArrayBlockingQueue<Runnable>(20), new NameThreadFactory(fetchPoolName),
				new ThreadPoolExecutor.DiscardPolicy());

		final String deletePoolName = "delete-pool";
		deleteVsearchDataExecutor = new ThreadPoolExecutor(4, threadNum, 10, TimeUnit.SECONDS,
				new ArrayBlockingQueue<Runnable>(20), new NameThreadFactory(deletePoolName),
				new BlockRejectedExecutionHandler<Runnable>(deletePoolName));
	}

假如：fetch线程很耗时，fetch的线程池 的RejectedExecutionHandler handler 使用ThreadPoolExecutor.CallerRunsPolicy()或者类似假如阻塞队列当前线程处理，在同时加入线程数 >  maximumPoolSize+BlockingQueue.size() 会导致主线程堵塞，程序无法执行下去，删除线程无法添加到线程池中，删除线程无法执行。这样就需要考虑fetch线程池的阻塞队列的长度要大于fetch线程数，考虑是否要丢弃过多的fetch线程，以保证删除线程能够顺利得到执行。