Remove the ZNode for HDFS, YARN, Hive, HBase, and Oozie
If the cluster is configured for Namenode HA, the Failover controllers cannot
determine the Active Namenode after the migration. This is because the naming
conventions used to identify Namenodes for the HA service are different for Ambari and
Cloudera Manager. You must format the ZooKeeper ZNode for the Failover controller to
continue further.If the cluster is configured for Yarn HA, the Resource Managers
cannot elect an Active Resource Manager after the migration. You must clear the
previous Znode from ZooKeeper so that YARN replaces it.
You must format the
ZooKeeper Znode for the failover controller to continue further.
-
Start ZooKeeper client CLI session from a master node. (Assuming that the
skipACL is set to Yes to avoid
authentication issues).
zookeeper-client -server
<zk_hostname>
-
Remove the HDFS HA Failover controller Znode
deleteall /hadoop-ha
-
Remove the YARN ZNode
deleteall /yarn-leader-election
deleteall /rmstore
-
Start all role instances of YARN.
-
Remove the Hive ZNode
deleteall /hive
deleteall /hiveserver2
-
Remove the HBase ZNode
deleteall /hbase-secure
-
Remove the Oozie ZNode
deleteall /oozie
