attilapiros commented on a change in pull request #29014: URL: https://github.com/apache/spark/pull/29014#discussion_r459924690
########## File path: core/src/main/scala/org/apache/spark/scheduler/TaskSchedulerImpl.scala ########## @@ -939,12 +941,43 @@ private[spark] class TaskSchedulerImpl( override def executorDecommission( executorId: String, decommissionInfo: ExecutorDecommissionInfo): Unit = { + synchronized { + // The scheduler can get multiple decommission updates from multiple sources, + // and some of those can have isHostDecommissioned false. We merge them such that + // if we heard isHostDecommissioned ever true, then we keep that one since it is + // most likely coming from the cluster manager and thus authoritative + val oldDecomInfo = executorsPendingDecommission.get(executorId) + // Don't bother noting decommissioning for executors that don't have running tasks + // to prevent a race where the `executorDecommission` can be called after `removeExecutor` + if (executorIdToRunningTaskIds.contains(executorId) && Review comment: 1) I think avoiding the above with `executorIdToRunningTaskIds.contains(executorId)` assumes to much knowledge about the interactions behind. **What about just checking whether the `executorId` is contained in `executorIdToHost`?** I think that is more straightforward: removed executors will be sooner or latter will be cleaned from the `executorIdToHost`: https://github.com/apache/spark/blob/2ebfaccd7710e4152ccb73abfe91ecddf694356d/core/src/main/scala/org/apache/spark/scheduler/TaskSchedulerImpl.scala#L1072 If the `LossReasonPending` that just mean latter: https://github.com/apache/spark/blob/2ebfaccd7710e4152ccb73abfe91ecddf694356d/core/src/main/scala/org/apache/spark/scheduler/TaskSchedulerImpl.scala#L991-L993 2) Move this condition up: we do not need the `oldDecomInfo` when the executor is already removed. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org