0

I have installed an hadoop cluster with one master and 3 nodes. hadoop v 3.2.1, hive 3.1.2 and tez 0.10.0

I verified conf files and I can't find why tez is trying to connect to yarn throw localhost:18032 yarn is configured to be on hadoop-master:8032.

I tested directly tez with an example and it connects correctly to yarn

hadoop jar tez-examples-0.10.0.jar  orderedwordcount /user/hive/foo.txt /user/hive/tmp/out

So, when executing a SQL like insert into ... I have that log

2022-10-28 12:36:15,036 INFO ipc.Client (Client.java:handleConnectionFailure(1010)) - Retrying connect to server: localhost/127.0.0.1:18032. Already tried 9 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS)

Jeannot77680
  • 49
  • 6
  • 1
    I suggest searching `yarn.resourcemanager.address` property in your cluster and check the values are correct or not. – Sercan Oct 28 '22 at 16:15
  • 1
    Also make sure the value is the same across every node's config files , and that you have set environment variables such as `HADOOP_CONF_DIR` correctly. – OneCricketeer Oct 28 '22 at 22:08

1 Answers1

1

I encountered such a situation, no error (because I turned off yarn's memory check). If you don't use the tez engine, you can start hive normally. Later, I think it will be because the memory is too small to start it? I restarted many components of the cluster: including hdfs, yarn, flume, kafa, zookeeper. Then restart hdfs and yarn and zookeeper. Finally, hive can be started.

sssr
  • 364
  • 1
  • 6