jealous commented on issue #24462: [SPARK-26268][CORE] Do not resubmit tasks when executors are lost URL: https://github.com/apache/spark/pull/24462#issuecomment-487772709 We have also developed our own shuffle manager. Here is the code: https://github.com/MemVerge/splash/ It doesn't rely on the executor. When the shuffle data is persisted to a storage cluster, you can still retrieve it when the executor dies. Spark has defined an interface for shuffle manager and it's already a configuration option in the application. So I agree that we should also make it a configuration option on whether we need to resubmit tasks when the executors are lost.
---------------------------------------------------------------- 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]
