[YARN-880] Configuring map/reduce memory equal to nodemanager's memory, hangs the job execution

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Not A Problem
• Affects Version/s: 2.0.1-alpha
• Component/s: resourcemanager
• Assignee: Unassigned
• Reporter: Nishan Shetty
• Created: 24/Jun/13 14:21
• Updated: 01/May/15 11:34
• Resolved: 01/May/15 11:34

Description

Scenario:
=========
Cluster is installed with 2 Nodemanagers

Configuraiton:

NM memory (yarn.nodemanager.resource.memory-mb): 8 gb
map and reduce memory : 8 gb
Appmaster memory: 2 gb

If map task is reserved on the same nodemanager where appmaster of the same job is running then job execution hangs.

Issue Links

• is blocked by: YARN-957 Capacity Scheduler tries to reserve the memory more than what node manager reports.