dlmarion commented on PR #3652:
URL: https://github.com/apache/accumulo/pull/3652#issuecomment-1649680070

   > The test CompactionIT.testErrorDuringUserCompaction() was timing out. The 
test would create a tablet with 1000 files. These would compact in batches of 
10 files at a time with a 4 or 5 seconds between each compaction. So this could 
take 400 to 500 seconds. The few second delay between compactions comes from 
waiting for the manager to scan the metadata table again. When compactions ran 
in the tablet server it would immediately start another compaction when one 
finished.
   
   This is likely due to the value of `MANAGER_TABLET_GROUP_WATCHER_INTERVAL` 
which is set to 5s.
   
   Overall, this is a side effect of moving compactions to the Manager. It 
sounds like the TGW is only queueing up 1 compaction per tablet per iteration. 
Is that the case? While lowering the files or the watcher interval will get the 
test to work, I wonder if that is the right fix here.
   
   My concern is the behavior on large systems where it may take several 
minutes for the TGW to evaluate a Tablet again.
   


-- 
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