https://community.cloudera.com/t5/Support-Questions/Yarn-containers-get-KILLED-automatically-MapReduce-Hive-jobs/m-p/125805#M88549 <P><A rel="user" href="https://community.cloudera.com/users/10163/clukasik.html" nodeid="10163">@clukasik</A> This is for sure that no one has issued the yarn kill command. Moreover, the machine 172.23.35.6 is a machine where only master services are running. And access is restricted to this machine. The services on this machine are -</P><P>namenode, hive_server, hive_metastore, journalnode, kerberos_client,mapreduce2_client,</P><P>falcon_client, hbase_client, hbase_master, hcat, hdfs_client, hive_client, mahout, metrics_monitor, oozie_client, phoenix_query_server, pig, ranger_admin, ranger_usersync, resourcemanager, spark_client, spark_thriftserver, sqoop, tez_client, webhcat_server, yarn_client, zkfc, zookeeper_client, zookeeper_server</P><P>Since these are hive jobs, there could be a possibility that hive_server itself is issuing sort of a kill command. Not sure though - will check hive logs as well.</P> Wed, 30 Nov 2016 14:50:01 GMT dosigary 2016-11-30T14:50:01Z