vanzin commented on issue #22015: 
[SPARK-20286][SPARK-24786][Core][DynamicAllocation] Release executors on 
unpersisting RDD
URL: https://github.com/apache/spark/pull/22015#issuecomment-468437353
 
 
   So, I just happen to be playing with this part of the code, and have a 
different solution for this. It's part of a much bigger change, so if you want 
to implement the specific solution for the caching problem here, that would be 
fine.
   
   The main change in this part is that I'm getting rid to the calls to 
`blockManagerMaster.hasCachedBlocks`, which are sort of expensive. It's 
possible to track this information without that call; the listener just needs 
to process `onBlockUpdated` and `onUnpersistRDD`.
   
   With that info it's also possible to handle `onUnpersistRDD` without the 
extra changes to the event, since the listener can itself keep track of which 
executors hold blocks for each RDD.
   
   I think that's a better approach.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to