[GitHub] [rocketmq] XiaoyiPeng edited a comment on pull request #3509: [ISSUE #2516]: Fix the value of sendThreadPoolQueueHeadWaitTimeMills is 0 most of the time

[GitBox <gi...@apache.org>]

XiaoyiPeng edited a comment on pull request #3509:
URL: https://github.com/apache/rocketmq/pull/3509#issuecomment-995634452


   Hi, @areyouok  @RongtongJin  @duhenglucky 
   
   I reviewed the given test cases again, the broker will never hangs in your test, @areyouok 
   
   In your test case, the reason you made the judgment:  **broker hangs**, is due to your console println "begin scan, i still alive", but not println "finish scan, i still alive" finally, or vice versa. 
   
   **That's because the IDE console lied to us！**
   **Our naked eyes do not perceive the tiny change in the console !**
   
   You can use `AtomicInteger` to print the times of prints before you print "begin scan, i still alive" and "finish scan, i still alive". The code shown below: 
   
   ```java
   public class TestLinkedBlockingQueue {
       public static void main(String[] args) throws InterruptedException {
           AtomicInteger startScanTimes = new AtomicInteger();
           AtomicInteger finishScanTimes = new AtomicInteger();
           LinkedBlockingQueue<Object> queue = new LinkedBlockingQueue<>();
           for (int i = 0; i < 10; i++) {
               new Thread(() -> {
                   while (true) {
                       queue.offer(new Object());
                       queue.remove();
                   }
               }).start();
           }
   
           while (true) {
               System.out.println("begin scan, i still alive, times: " + startScanTimes.incrementAndGet());
               queue.stream().filter(o -> o != null)
                       .findFirst().isPresent();
               Thread.sleep(1000);
               System.out.println("finish scan, i still alive, times: " + finishScanTimes.incrementAndGet());
           }
       }
   }
   ```
   We can run a fatigue test with this test case, it will never hangs and print the given info forever!
   
   As for the **deadlock**, that was my initial misjudgment. In methods of class `LinkedBlockingQueue`, `putLock` and `takeLock` are never acquired simultaneously in a different order.
   
   Meanwhile, the author of class LinkedBlockingQueue is **Doug Lea**, I don't think their test team would be so lax.
   
   best regards!
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: dev-unsubscribe@rocketmq.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org