mcvsubbu commented on issue #5559:
URL: 
https://github.com/apache/incubator-pinot/issues/5559#issuecomment-649634794


   @kishoreg Maybe you forget, :) but I know how hard it was to detect (let 
alone debug) the phenomena of two consuming segments for the same partition in 
the same replica. 
   
   Yes, this is a corner case, but when an issue is reported, we need to debug 
it. If it was hard when we had all the machines and logs at hand and could look 
at zookpeer logs at will, imagine doing it remotely when some open source user 
reports a problem when we can't even GUESS that there are parallel consuming 
segments.
   
   We added plenty of comments and put the logic in one class so that we don't 
stumble on such things again.
   
   
https://github.com/apache/incubator-pinot/blob/master/pinot-controller/src/main/java/org/apache/pinot/controller/helix/core/realtime/PinotLLCRealtimeSegmentManager.java#L665
   
   
https://github.com/apache/incubator-pinot/blob/master/pinot-controller/src/main/java/org/apache/pinot/controller/helix/core/realtime/PinotLLCRealtimeSegmentManager.java#L715
   
   
https://github.com/apache/incubator-pinot/blob/master/pinot-controller/src/main/java/org/apache/pinot/controller/helix/core/realtime/PinotLLCRealtimeSegmentManager.java#L754
   
   We make a strong assumption here that there are only two threads in the 
entire cluster that create segments. By starting a new validation manager, we 
violate that assumption.
   


----------------------------------------------------------------
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.

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



---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@pinot.apache.org
For additional commands, e-mail: commits-h...@pinot.apache.org

Reply via email to