[YARN-1265] Fair Scheduler chokes on unhealthy node reconnect

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.1.1-beta
• Fix Version/s: 2.3.0
• Component/s: resourcemanager, scheduler
• Hadoop Flags: Reviewed
• Assignee: Sandy Ryza
• Reporter: Sandy Ryza
• Created: 02/Oct/13 21:48
• Updated: 24/Feb/14 20:57
• Resolved: 10/Oct/13 22:54

Description

Only nodes in the RUNNING state are tracked by schedulers. When a node reconnects, RMNodeImpl.ReconnectNodeTransition tries to remove it, even if it's in the RUNNING state. The FairScheduler doesn't guard against this.

I think the best way to fix this is to check to see whether a node is RUNNING before telling the scheduler to remove it.