[YARN-455] NM warns about stopping an unknown container under normal circumstances

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 2.0.3-alpha, 0.23.6
• Component/s: nodemanager
• Assignee: Omkar Vinit Joshi
• Reporter: Jason Darrell Lowe
• Created: 07/Mar/13 16:00
• Updated: 18/Jul/13 19:13
• Resolved: 18/Jul/13 19:13

Description

During normal operations the NM can log warnings to its audit log about unknown containers. For example:

2013-03-06 21:04:55,327 WARN nodemanager.NMAuditLogger: USER=UnknownUser	IP=xx	OPERATION=Stop Container Request	TARGET=ContainerManagerImpl	RESULT=FAILURE	DESCRIPTION=Trying to stop unknown container!	APPID=application_1359150825713_3947178	CONTAINERID=container_1359150825713_3947178_01_001266

Looking closer at the audit log and the NM log shows that the container completed successfully and was forgotten by the NM before the stop request arrived. The NM should avoid warning in these situations since this is a "normal" race condition.

Issue Links

• duplicates: YARN-903 DistributedShell throwing Errors in logs after successfull completion
• is related to: YARN-903 DistributedShell throwing Errors in logs after successfull completion