Some when we join two large table or small tables, HIVE Tez engine gets stuck @ 98 or 99%, mapper completes it task faster and reducer completes 90% of work and finally stuck @ 98% and fails the job. Is there way to trouble shoot reason why reducer fails @ 98%. we have cbo, vectorization properties. We are using HDInsight 3.1 version.
I’m looking for help like how to trouble shoot this kind of performance issues. where can I find execution life cycle and move some of the work to mapper rather than reducer performing activity
I’m looking for help like how to trouble shoot this kind of performance issues. where can I find execution life cycle and move some of the work to mapper rather than reducer performing activity.
538 INFO [Socket Reader #4 for port 60810] org.apache.hadoop.ipc.Server: Socket Reader #4 for port 60810: readAndProcess from client 100.112.170.101 threw exception [java 1="An" 2="existing" 3="connection" 4="was" 5="forcibly" 6="closed" 7="by" 8="the" 9="remote" 10="host" language=".io.IOException:"][/java]
2015-11-13 03:06:10,474 INFO [AMRM Callback Handler Thread] org.apache.tez.dag.app.rm.TaskScheduler: Released container completed:container_1446913430734_1459_01_000053 last allocated to task: attempt_1446913430734_1459_1_00_000050_0
2015-11-13 03:06:10,474 INFO [AsyncDispatcher event handler] org.apache.tez.dag.app.rm.container.AMContainerImpl: Container container_1446913430734_1459_01_000053 exited with diagnostics set to Container released by application 2015-11-13 03:06:10,474 INFO [AsyncDispatcher event handler] org.apache.tez.dag.app.rm.container.AMContainerImpl: AMContainer container_1446913430734_1459_01_000053 transitioned from STOPPING to COMPLETED via event C_COMPLETED 2015-11-13 07:45:43,778 INFO [Socket Reader #3 for port 60809] org.apache.hadoop.ipc.Server: Socket Reader #3 for port 60809: readAndProcess from client 100.112.192.28 threw exception [java.io.IOException: An existing connection was forcibly closed by the remote host