You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by GitBox <gi...@apache.org> on 2021/12/14 06:13:26 UTC

[GitHub] [spark] venkata91 commented on pull request #33896: [SPARK-33701][SHUFFLE] Adaptive shuffle merge finalization for push-based shuffle

venkata91 commented on pull request #33896:
URL: https://github.com/apache/spark/pull/33896#issuecomment-993193098


   > Instead of FakeShuffleClient, we can mock shuffleClient.pushBlocks such that for specific host:port, we delay responding.
   > For example:
   > a) For requests to host1, we wait on a latch.
   > b) For other requests we return immediately (as in interceptPushedBlocksForSuccess).
   > 
   > This can be used to check all blocks are in flight, some have completed and reqsInFlight > 0 ?
   
   Tried waiting on the latch for one of the `shuffleClient.pushBlocks` call but this is still not working. Need to have a latch on the `notifyDriverAboutPushCompletion` so that the tests won't complete before all the blocks are pushed. So instead just checking `assert(bytesInFlight <= 0)` in `notifyDriverAboutPushCompletion` itself. Let me know if you have other thoughts.


-- 
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: reviews-unsubscribe@spark.apache.org

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



---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org