[YARN-999] In case of long running tasks, reduce node resource should balloon out resource quickly by calling preemption API and suspending running task.

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 3.3.0
• Component/s: graceful, nodemanager, scheduler
• Assignee: Íñigo Goiri
• Reporter: Junping Du
• Created: 30/Jul/13 15:56
• Updated: 07/Aug/20 01:30
• Resolved: 09/Apr/19 21:19

Description

In current design and implementation, when we decrease resource on node to less than resource consumption of current running tasks, tasks can still be running until the end. But just no new task get assigned on this node (because AvailableResource < 0) until some tasks are finished and AvailableResource > 0 again. This is good for most cases but in case of long running task, it could be too slow for resource setting to actually work so preemption could be used here.

Issue Links

• contains: YARN-2489 ResouceOption's overcommitTimeout should be respected during resource update on NM
• relates to: YARN-10388 RMNode updatedCapability flag not set while RecommissionNodeTransition
• relates to: YARN-9501 TestCapacitySchedulerOvercommit#testReducePreemptAndCancel fails intermittent
• relates to: YARN-9942 Reset overcommit timeout