keith-turner opened a new issue, #2700:
URL: https://github.com/apache/accumulo/issues/2700

   **Describe the bug**
   
   While testing the scan server feature #2665 I was running a query client I 
wrote that was doing 512 concurrent small random lookups.  These lookups were 
going to scan servers which were not behaving as expected.  I ran java flight 
recorder and saw contention on 3 locks.  Two locks were scan servers specific 
and one was related to the sha512 computation.  I also saw that outside of lock 
contention, the sha512 computation was the top method for profiling.  Even 
though I saw these issues in the scan server, I suspect the tserver would have 
the same problem because it uses the same code.  
   
   **To Reproduce**
   Steps to reproduce the behavior (or a link to an example repository that 
reproduces the problem):
   1. Go to '...'
   2. Click on '....'
   3. Scroll down to '....'
   4. See error
   
   **Expected behavior**
   Would be good to avoid the lock contention and research the performance of 
the sha512 computation relative to small scan.
   
   **Screenshots**
   
   The following is the JFR data that shows lock contention
   
   ![Screenshot 2022-05-12 
192749](https://user-images.githubusercontent.com/1268739/168185291-8fa7e50b-69b3-4619-a756-90b2b0762352.png)
   
   The following is the JFR data that shows hot methods.
   
   ![Screenshot 2022-05-12 
193928](https://user-images.githubusercontent.com/1268739/168185353-963b506f-03f1-471e-b280-e57a5e586203.png)
   
   The JFR data came from profiling the scan server for 2 minutes while my test 
code was running doing 512 concurrent small lookups.
   
   


-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to