Github user suyanNone commented on the pull request: https://github.com/apache/spark/pull/4055#issuecomment-113356815 @andrewor14 The rest failed, because I change the `DagScheduler.executorLost` logical DAGSCheduler.scala: Origin version: Change mapOutPutTracker epoch for all stage as long as a Executor lost handleExecutorLost { ...... for ((shuffleId, stage) <- shuffleToMapStage) { stage.removeOutputsOnExecutor(execId) val locs = stage.outputLocs.map(list => if (list.isEmpty) null else list.head).toArray mapOutputTracker.registerMapOutputs(shuffleId, locs, changeEpoch = true) ...... } This patch version: For running Stage is not need to re-register mapOutTracker, so it will not change the epoch. ``` for ((shuffleId, stage) <- shuffleToMapStage) { stage.removeOutputsOnExecutor(execId) if (!runningStages.contains(stage)) { val locs = stage.outputLocs.map(list => if (list.isEmpty) null else list.head).toArray mapOutputTracker.registerMapOutputs(shuffleId, locs, changeEpoch = true) } } ```
--- 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