[ https://issues.apache.org/jira/browse/YARN-4677?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16418525#comment-16418525 ]
Wilfred Spiegelenburg commented on YARN-4677: --------------------------------------------- [~djp] It has been a long time since anyone looked at this but I ran into this issue in the FS. I asume you don't mind that I take over. The code has been refactored and changed since this was reported but the issue still exists. I created a patch to fix and test the issue. [^YARN-4677.01.patch] > RMNodeResourceUpdateEvent update from scheduler can lead to race condition > -------------------------------------------------------------------------- > > Key: YARN-4677 > URL: https://issues.apache.org/jira/browse/YARN-4677 > Project: Hadoop YARN > Issue Type: Sub-task > Components: graceful, resourcemanager, scheduler > Affects Versions: 2.7.1 > Reporter: Brook Zhou > Assignee: Junping Du > Priority: Major > Attachments: YARN-4677.01.patch > > > When a node is in decommissioning state, there is time window between > completedContainer() and RMNodeResourceUpdateEvent get handled in > scheduler.nodeUpdate (YARN-3223). > So if a scheduling effort happens within this window, the new container could > still get allocated on this node. Even worse case is if scheduling effort > happen after RMNodeResourceUpdateEvent sent out but before it is propagated > to SchedulerNode - then the total resource is lower than used resource and > available resource is a negative value. -- This message was sent by Atlassian JIRA (v7.6.3#76005) --------------------------------------------------------------------- To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org