Github user holdenk commented on a diff in the pull request: https://github.com/apache/spark/pull/11919#discussion_r61769074 --- Diff: mllib/src/main/scala/org/apache/spark/ml/recommendation/ALS.scala --- @@ -1355,4 +1359,28 @@ object ALS extends DefaultParamsReadable[ALS] with Logging { * satisfies this requirement, we simply use a type alias here. */ private[recommendation] type ALSPartitioner = org.apache.spark.HashPartitioner + + /** + * Private function to clean up all of the shuffles files from the dependencies and their parents. + */ + private[spark] def cleanShuffleDependencies[T](sc: SparkContext, deps: Seq[Dependency[_]], + blocking: Boolean = false): Unit = { + // If there is no reference tracking we skip clean up. + sc.cleaner.foreach{ cleaner => + /** + * Clean the shuffles & all of its parents. + */ + def cleanEagerly(dep: Dependency[_]): Unit = { + if (dep.isInstanceOf[ShuffleDependency[_, _, _]]) { + val shuffleId = dep.asInstanceOf[ShuffleDependency[_, _, _]].shuffleId + cleaner.doCleanupShuffle(shuffleId, blocking) --- End diff -- So my understanding is that previously the ttl cleaner could clean up check point files and Spark would still need to handle recompute. Of course I've gone ahead and gone with the safer choice - although I also modified the first test to illustrate that recompute works fine (we clean up the shuffle files of a non-cached RDD and then call count).
--- 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