[ https://issues.apache.org/jira/browse/SPARK-1112?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14011978#comment-14011978 ]
Matei Zaharia commented on SPARK-1112: -------------------------------------- I'm curious, why did you want to make the frameSize this big -- are the tasks themselves also big or just the results? There might be other buffers in Akka that can't be made bigger than this. It's possible that this changed in a newer Akka version (because larger frame sizes used to work before). > When spark.akka.frameSize > 10, task results bigger than 10MiB block execution > ------------------------------------------------------------------------------ > > Key: SPARK-1112 > URL: https://issues.apache.org/jira/browse/SPARK-1112 > Project: Spark > Issue Type: Bug > Components: Spark Core > Affects Versions: 0.9.0 > Reporter: Guillaume Pitel > Priority: Critical > Fix For: 0.9.2 > > > When I set the spark.akka.frameSize to something over 10, the messages sent > from the executors to the driver completely block the execution if the > message is bigger than 10MiB and smaller than the frameSize (if it's above > the frameSize, it's ok) > Workaround is to set the spark.akka.frameSize to 10. In this case, since > 0.8.1, the blockManager deal with the data to be sent. It seems slower than > akka direct message though. > The configuration seems to be correctly read (see actorSystemConfig.txt), so > I don't see where the 10MiB could come from -- This message was sent by Atlassian JIRA (v6.2#6252)