tgravescs commented on pull request #30650: URL: https://github.com/apache/spark/pull/30650#issuecomment-770930298
> After times thinking, I think we should just keep the original behavior for the barrier taskset with the legacy delay scheduling. That means we should still abort the taskset and throw an exception when tasks are partially launched in that case. Right that means this case will never work with the legacy delay scheduling, which I think is what we were saying seems broken to me. I guess it was this way before so its no worse from that sense and with change in locality algorithm maybe its not a big deal. @mridulm thoughts? ---------------------------------------------------------------- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
