Github user jerryshao commented on the issue: https://github.com/apache/spark/pull/19396 Sorry for the late response. I understand you purpose now. I think such behavior discrepancy is not a big problem. I guess the reason why NM still run with exception is that NM doesn't serve only for Spark, but also MR/TEZ, so the failure of Spark external service should not affect MR's. Based on your comment above, I don't have a strong preference on either, I think both are OK. Maybe you can ping others to get their feedbacks.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org