Github user caneGuy commented on the issue: https://github.com/apache/spark/pull/18723 @kiszk Thanks for your time. Yes,you are right. 1.OOM happens at io.netty.buffer.CompositeByteBuf.allocBuffer when one executor(client) fetch blocks from an other(server) , oom happens at server side before response to client. 2.Components number depends on chunks' length.Low-level is a list,since we do not need to remove or update component in the list, i can not think of any downside in our usage.However, if we do not consolidate we can avoid copy,which will improve the performance.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org