Just to be sure that YARM/MR works fine on your setup, could you show the output that is printed when you run an example MapReduce jobs?
$ hadoop jar <path-to-hadoop-mapreduce-jars>/hadoop-mapreduce-examples.jar pi 10 100 # it runs PiEstimator example to estimate value of Pi and $ hadoop distcp /hbase /tmp/hbase-copy # it copies content from <src-dir> to <target-dir> Last but not least, it's much more comfortable and easier to manage Hadoop/HBase with distro like HDP or CDH. There are few reasons (1) they make sure that the versions components (Hadoop, HBase, ZooKeeper, Hive etc.) that you use are compatible, (2) they provide relatively smart default configurations and recommendations, (3) they minimize the risk of misconfiguration by deploying configuration changes to proper hosts, (4) they install everything in proper order, directories and hosts, (5) they give you monitoring, alerts, warnings etc. -- View this message in context: http://apache-hbase.679495.n3.nabble.com/Hbase-exportSnapshot-fails-in-HA-missing-hbase-prefix-tree-jar-tp4087506p4087722.html Sent from the HBase User mailing list archive at Nabble.com.