[GitHub] [druid] acherla commented on issue #13692: Druid Version 0.22.1 - Long Coordinator handoff times

["acherla (via GitHub)" <gi...@apache.org>]

acherla commented on issue #13692:
URL: https://github.com/apache/druid/issues/13692#issuecomment-1500210922

   > The chat requests are going from Overlord to the workers. Surprising that it's causing Overlord CPU to become high. When you see spike in Overlord CPU usage, can you check through thread dumps where are Overlord threads spending their time? Maybe it is something else.
   > 
   > We have also seen segment allocation cause significant overhead at that scale. This overhead can be reduced by enabling batch segment allocation. `druid.indexer.tasklock.batchSegmentAllocation` needs to be set to `true` on Overlord.
   
   Thanks we've enabled batchSegmentAllocation on the overlord since the 25.0 upgrade and increased the batchsegment interval to 10 seconds.  
   
   Attached a copy of the flamegraph we took from the overlord during this period when it stalls out
   
   [druid-flame-117758.zip](https://github.com/apache/druid/files/11178605/druid-flame-117758.zip)
   


-- 
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: commits-unsubscribe@druid.apache.org

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


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@druid.apache.org
For additional commands, e-mail: commits-help@druid.apache.org