在春季计划任务

Question

We have a requirement where we have to collect documents for X minutes in the database where X could be 30-60. Once these documents are collected in the Database, we have to push them to a separate service over a REST call.

Since these documents won't be very large we aggregate them together and then push them together.

Now, these documents have to be pushed after a fixed interval, So I thought of creating a separate thread which pulls records from the database, pushes them to the service and once the service responds with 200 OK, I remove these records from the Database and the thread goes to Sleep for X minutes.

class PushDocumentsToService extends Thread{
    DocumentRepository documentRepository;
    DocumentProcessingService documentProcessingService
    public void run(){
        List<Document> list = documentRepository.getAllDocuments();
        Integer statusCode = documentProcessingService.sendDocuments(list);
        if(statusCode == 200)
            documentRepository.remove(list);
        try{
            Thread.sleep(30*60);
        }catch(..){..}
    }    

}

Now a few of my colleagues have advised me against using Thread.sleep in my code. I am not able to understand the ill-effects of

Thread.sleep()

Also what methods do libraries and frameworks as Spring Scheduler and Google Guava use for Scheduling task?

Answer 1

All you need to do here is use ScheduledExecutorService to schedule a thread and provide proper frequency rate like this.

ScheduledExecutorService scheduledThreadPool = Executors.newScheduledThreadPool(5);    

Thread t1= new PushDocumentsToService();    
scheduledThreadPool.scheduleAtFixedRate(t1, 0, 30,TimeUnit.SECONDS);

Answer 2

If you have the Spring Scheduling library in your classpath, add the @Scheduled annotation to the method you wish to execute over a period of time. No need to extend Thread or implement Runnable ; this is all taken care of on your behalf through Spring.

You know, for as long as the application is running .

@Scheduled(cron = "30 * * * * *")
public void loadDataIntoRepository() {
    // your code
}

It is ill advised to use Thread.sleep for this for several reasons:

• The thread executing your code is now blocked for the duration of the sleep period. 30 minutes is a very long period of time to have your thread blocked.
• Your thread could be interrupted, offering you no real way to recover from your error state.
• You don't want to have to reinvent the wheel with scheduling when that functionality already exists and is already well-tested, for maintainability's sake.