一个jvm内存泄露问题

由于利用线程池提交任务的操作放到了while循环中，导致利用该线程池循环提交任务，导致任务队列爆满。应该改为线程池提交的子线程中循环处理取处理任务的逻辑。

一段时间后线程池中的WorkQueue挤压了4千多条待处理的任务

一种针对大批量数据处理，如从kafka中获取消息然后开多线处理消息的场景，考虑到如果用主线程去从kafka中拿消息可能发生异常，导致线程停止，可以从主线程中创造一个固定长度为1的线程池提交任务去从kafka中取，然后在主线程中检查futuer。这样可以保证如果取从kafka中取消息的线程挂了或者超时了，线程池可以重新创建线程继续处理

class KafkaMessagePublisher{
ExecutorService executorService = Executors.newFixedThreadPool(1);
KafkaBatchConsumer tcConsumer = new KafkaBatchConsumer();
//判断开关变量
while (!doneConsuming.get()) {
            Future future = executorService.submit(() -> {
                tcConsumer.batchProcess();
                return true;
            });

            try {
                future.get(40, TimeUnit.SECONDS);
            } catch (Exception e) {
                boolean cancelFlag = future.cancel(true);
                logger.warn(Thread.currentThread().getName() + "->>KafkaMessagePublisher cancelFlag:" + cancelFlag + ";future isCancelled:" + future.isCancelled()
                        + ";" + threadPoolExecutor.toString(), e);
            }
        }
}

class KafkaBatchConsumer{

public void batchProcess() {
        final String batchUUID = UUID.randomUUID().toString();
        try {
            ConsumerRecords<String, String> records = consumer.poll(1000);
            logger.info("records num:" + records.count());

            if (records.count() > 0) {
                for (ConsumerRecord<String, String> record : records) {
                    MessageTask messageTask = MessageTaskFactory.getInstance(downloadType, record.value(),null);
                    Future<Boolean> spiderFuture = threadPoolExecutor.submit(messageTask);

                    //检测各个子线程是否超时
                    Object[] args = new Object[2];
                    args[0] = spiderFuture;
                    args[1] = record.value();
                    disruptor.publishEvent(new FutureListEventTranslator(), args);

                    tpAndOffsetMetadata.put(new TopicPartition(record.topic(), record.partition()),
                            new OffsetAndMetadata(record.offset() + 1, batchUUID));
                }
                consumer.commitSync(tpAndOffsetMetadata);
                tpAndOffsetMetadata.clear();
            }

        } catch (Exception e) {
            logger.error("KafkaSource EXCEPTION: {}", e);
        }
    }
}

JVM常见问题：https://gitbook.cn/books/5948740540c65d603756a41f/index.html