[MAPREDUCE-3121] DFIP aka 'NodeManager should handle Disk-Failures In Place'

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 0.23.0
• Fix Version/s: 0.23.1
• Component/s: mrv2, nodemanager
• Target Version/s: 0.23.1
• Assignee: Ravi Gummadi
• Reporter: Vinod Kumar Vavilapalli
• Created: 29/Sep/11 14:34
• Updated: 05/Mar/12 02:49
• Resolved: 29/Nov/11 23:28

Description

This is akin to MAPREDUCE-2413 but for YARN's NodeManager. We want to minimize the impact of transient/permanent disk failures on containers. With larger number of disks per node, the ability to continue to run containers on other disks is crucial.

Issue Links

• is related to: MAPREDUCE-3473 A single task tracker failure shouldn't result in Job failure

Sub-Tasks

• DISKS_FAILED -101 error code should be defined in same location as ABORTED_CONTAINER_EXIT_STATUS: Closed
• Deadlock in LocalDirsHandlerService and ShuffleHandler: Closed