squito commented on a change in pull request #23927: [SPARK-23433][CORE] avoid more than one active task set managers for a stage URL: https://github.com/apache/spark/pull/23927#discussion_r262584316
########## File path: core/src/main/scala/org/apache/spark/scheduler/TaskSetManager.scala ########## @@ -123,7 +123,7 @@ private[spark] class TaskSetManager( // state until all tasks have finished running; we keep TaskSetManagers that are in the zombie // state in order to continue to track and account for the running tasks. // TODO: We should kill any running task attempts when the task set manager becomes a zombie. - private[scheduler] var isZombie = false + @volatile private[scheduler] var isZombie = false Review comment: I don't think this is necessary. You still only touch `isZombie` with a lock on the TaskSchedulerImpl (*everything* in the TSM requires a lock on the TaskSchedulerImpl). Even before, there could be multiple task-result-getter threads interacting w/ this. ---------------------------------------------------------------- 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: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
