[YARN-2555] Effective max-allocation-* should consider biggest node

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 2.5.1
• Target Version/s: 2.6.0
• Assignee: Unassigned
• Reporter: Karthik Kambatla
• Created: 15/Sep/14 19:47
• Updated: 03/Nov/14 18:33
• Resolved: 16/Sep/14 03:26

Description

The effective max-allocation-mb should be min(admin-configured-max-allocation-mb, max-mb-on-one-node), so we can reject container requests for resources larger than any node. Today, these requests wait forever.

We should do this for all resources and update the effective value on node updates.

Issue Links

• is duplicated by: YARN-56 Handle container requests that request more resources than currently available in the cluster
• relates to: YARN-2422 yarn.scheduler.maximum-allocation-mb should not be hard-coded in yarn-default.xml