I have been seeing this warning a lot in the yarn logs for every job that is run different users.... any idea about this?
2017-02-07 16:47:01,477 WARN resourcemanager.RMAuditLogger (RMAuditLogger.java:logFailure(267)) - USER=user IP=x.x.x.x OPERATION=AM Released Container TARGET=Scheduler RESULT=FAILURE DESCRIPTION=Trying to release container not owned by app or with invalid id. PERMISSIONS=Unauthorized access or invalid container APPID=application_1485795502013_2891 CONTAINERID=container_e31_1485795502013_2891_01_000313
This could occur if you have an overloaded NM and the liveness monitor expiration has occurred. Are you seeing an Nodemanagers in a Lost state? What does resource consumption look like on your nodemanagers when this occurs?