Return-Path: X-Original-To: apmail-spark-reviews-archive@minotaur.apache.org Delivered-To: apmail-spark-reviews-archive@minotaur.apache.org Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by minotaur.apache.org (Postfix) with SMTP id 9251518B12 for ; Wed, 16 Sep 2015 02:07:41 +0000 (UTC) Received: (qmail 70543 invoked by uid 500); 16 Sep 2015 02:07:41 -0000 Delivered-To: apmail-spark-reviews-archive@spark.apache.org Received: (qmail 70517 invoked by uid 500); 16 Sep 2015 02:07:41 -0000 Mailing-List: contact reviews-help@spark.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Delivered-To: mailing list reviews@spark.apache.org Received: (qmail 70505 invoked by uid 99); 16 Sep 2015 02:07:41 -0000 Received: from git1-us-west.apache.org (HELO git1-us-west.apache.org) (140.211.11.23) by apache.org (qpsmtpd/0.29) with ESMTP; Wed, 16 Sep 2015 02:07:41 +0000 Received: by git1-us-west.apache.org (ASF Mail Server at git1-us-west.apache.org, from userid 33) id 181ACE03C8; Wed, 16 Sep 2015 02:07:41 +0000 (UTC) From: vanzin To: reviews@spark.apache.org Reply-To: reviews@spark.apache.org References: In-Reply-To: Subject: [GitHub] spark pull request: [SPARK-10515] When killing executor, there is ... Content-Type: text/plain Message-Id: <20150916020741.181ACE03C8@git1-us-west.apache.org> Date: Wed, 16 Sep 2015 02:07:41 +0000 (UTC) Github user vanzin commented on the pull request: https://github.com/apache/spark/pull/8668#issuecomment-140601612 Ok, I think I see what the problem is. But your fix is not correct. The problem is here: doRequestTotalExecutors( numExistingExecutors + numPendingExecutors - executorsPendingToRemove.size) By subtracting `executorsPendingToRemove.size` when that list contains an executor that is pending replacement, that replacement will be lost. The fix is to keep track of how many replacement executors the code is waiting for, and account for that in the above equation, not to remove that code altogether. > So there is no need to change the number of executors when killing executors. That's not true, in YARN, at least. See SPARK-6325. So you can't make your current change unless you also change how the YARN backend does accounting for the running executors. --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastructure@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org For additional commands, e-mail: reviews-help@spark.apache.org