[YARN-8992] Fair scheduler can delete a dynamic queue while an application attempt is being added to the queue

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.1.1
• Fix Version/s: 3.3.0, 3.2.3
• Component/s: fairscheduler
• Labels: pull-request-available release-blocker
• Target Version/s: 3.2.3
• Hadoop Flags: Reviewed
• Assignee: Wilfred Spiegelenburg
• Reporter: Haibo Chen
• Created: 08/Nov/18 23:59
• Updated: 04/Aug/21 01:18
• Resolved: 04/Aug/21 01:18

Description

As discovered in YARN-8990, QueueManager can see a leaf queue being empty while FSLeafQueue.addApp() is called in the middle of      

return queue.getNumRunnableApps() == 0 &&
          leafQueue.getNumNonRunnableApps() == 0 &&
          leafQueue.getNumAssignedApps() == 0;

Issue Links

• is related to: YARN-8990 Fix fair scheduler race condition in app submit and queue cleanup
• links to: GitHub Pull Request #3257