Github user jerryshao commented on the issue: https://github.com/apache/spark/pull/21758 Hi @jiangxb1987 one question about barrier task. For example, `rddA` is marked as barrier, and `rddB` is a normal RDD, if `rddC = rddA.union(rddB)`, seems it contains both normal task and barrier, will you try to mark tasks generated from `rddB` also as barrier task, or you will only mark tasks from 'rddA' as barrier tasks? This will potentially affect the resource demands, as `rddA` requires gang semantics, whereas `rddB` doesn't require that.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org