[YARN-141] NodeManager shuts down if it can't find the ResourceManager

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Assignee: Ahmed Radwan
• Reporter: Ahmed Radwan
• Created: 02/Oct/12 21:23
• Updated: 14/Mar/13 06:53
• Resolved: 14/Mar/13 06:52

Description

When starting yarn services, and if the NodeManager is started but the ResourceManager is not. The NodeManager tries 10 times (the default setting), and then shuts down.

I understand that this default setting can be changed and possibly wait for a longer period, but I think it is better to keep the NodeManager trying without shutting down. This can accomodate cases where the ResourceManager is late to start for any problems and it will preserve the same behavior to what the DataNode does when it cannot find the NameNode at startup as it keeps trying without shutting down.

Issue Links

• is duplicated by: YARN-196 Nodemanager should be more robust in handling connection failure to ResourceManager when a cluster is started