[ https://issues.apache.org/jira/browse/SPARK-15496?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Eric Liang resolved SPARK-15496. -------------------------------- Resolution: Fixed Fix Version/s: 2.0.0 Ah, actually the reproduction was incorrect. This issue I saw before seems to have been fixed in master. > Spill metrics not updated when off-heap memory is enabled > --------------------------------------------------------- > > Key: SPARK-15496 > URL: https://issues.apache.org/jira/browse/SPARK-15496 > Project: Spark > Issue Type: Bug > Affects Versions: 2.0.0 > Reporter: Eric Liang > Fix For: 2.0.0 > > > This is reproducible by looking in the SQL UI for the Sort operator spill > metrics after running the following query: > {code} > scala> val table = spark.range(0, 1000000000) > table: org.apache.spark.sql.Dataset[Long] = [id: bigint] > scala> table.join(table, "id").count() > res0: Long = 1000000000 > > {code} > If off-heap memory is enabled, e.g. bin/spark-shell > --master=local-cluster[1,16,250000] --conf spark.executor.memory=2g > --driver-memory 1g --conf spark.memory.offHeap.enabled=true --conf > spark.memory.offHeap.size=15g, then the spill metric will always be zero. > cc [~davies] -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org