Append HBASE_CLASSPATH to HADOOP_CLASSPATH like below:- export HADOOP_CLASSPATH=$HADOOP_CLASSPATH:$HBASE_CLASSPATH //but before that confirm HBASE_CLASSPATH is properly set. And also you may need to set hive.aux.jars.path in hive-site.xml so that libraries are available in MR job. <property> <name>hive.aux.jars.path</name> <value><hive_home>/lib/hive-hbase-handler-<version>.jar, <hbase_home>/lib/hbase-client-<version>.jar, <hbase_home>/lib/hbase-server-<version>.jar,<hbase_home>/lib/hbase-protocol-<version>.jar,<hbase_home>/lib/zookeeper-<version>.jar</value> <description>A comma separated list (with no spaces) of the jar files required for Hive-HBase integration</description> </property> replace the above paths with actual paths as per your setup. ... View more

Upgrading any component out of HDP is not recommended , however, you can upgrade to phoenix 4.6 with just changing phoenix-server.jar on all nodes with phoenix-4.6 server jar and upgrading client jar to phoenix-4.6 . Updates in the schema will be automatically taken care by new phoenix-client jar. And just look for upgrade option in bin/psql.py. u,--upgrade Upgrades tables specified as arguments by rewriting them with the correct row key for descending columns. If no arguments are specified, then tables that need to be upgraded will be displayed without being upgraded. Use the -b option to bypass the rewrite if you know that your data does not need to be upgrade. This would only be the case if you have not relied on auto padding for BINARY and CHAR data, but instead have always provided data up to the full max length of the column. See PHOENIX-2067 and PHOENIX-2120 for more information. Note that phoenix.query.timeoutMs and hbase.regionserver.lease.period parameters must be set very high to prevent timeouts when upgrading. if you have data affected by above jira tickets. ... View more

Hi @Pedro Gandola This problem occurs when meta regions are not assigned yet and preScannerOpen coprocessor waits for reading meta table for local indexes, which results in openregionthreads to wait forever because of deadlock. you can solve this by increasing number of threads required to open the regions so that meta regions can be assigned even threads for local index table is still waiting to remove the deadlock. <property> <name>hbase.regionserver.executor.openregion.threads</name> <value>100</value> </property> ... View more