[ https://issues.apache.org/jira/browse/SPARK-11271?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Davies Liu reopened SPARK-11271: -------------------------------- https://github.com/apache/spark/pull/9243 is reverted > MapStatus too large for driver > ------------------------------ > > Key: SPARK-11271 > URL: https://issues.apache.org/jira/browse/SPARK-11271 > Project: Spark > Issue Type: Improvement > Components: Shuffle > Reporter: Kent Yao > Assignee: Liang-Chi Hsieh > Fix For: 1.6.0 > > > When I run a spark job contains quite a lot of tasks(in my case is > 200k[maptasks]*200k[reducetasks]), the driver occured OOM mainly caused by > the object MapStatus, > RoaringBitmap that used to mark which block is empty seems to use too many > memories. > I try to use org.apache.spark.util.collection.BitSet instead of > RoaringBitMap, and it can save about 20% memories. > For the 200K tasks job, > RoaringBitMap uses 3 Long[1024] and 1 Short[3392] > =3*64*1024+16*3392=250880(bit) > BitSet uses 1 Long[3125] = 3125*64=200000(bit) > Memory saved = (250880-200000) / 250880 ≈20% -- 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