Important: These numbers are general guidance only, and can be affected by factors such as number of columns, partitions, complex joins, and client activity. Based on your anticipated deployment, refine through testing to arrive at the best values for your environment.

For information on configuring heap for HiveServer2, as well as Hive metastore and Hive clients, see Heap Size and Garbage Collection for Hive Components and the following video:

You must properly configure and enable Hive's Table Lock Manager. This requires installing ZooKeeper and setting up a ZooKeeper ensemble; see ZooKeeper Installation.

  Important: Failure to do this will prevent HiveServer2 from handling concurrent query requests and may result in data corruption.

Enable the lock manager by setting properties in /etc/hive/conf/hive-site.xml as follows (substitute your actual ZooKeeper node names for those in the example):

<property>
  <name>hive.support.concurrency</name>
  <description>Enable Hive's Table Lock Manager Service</description>
  <value>true</value>
</property>

<property>
  <name>hive.zookeeper.quorum</name>
  <description>Zookeeper quorum used by Hive's Table Lock Manager</description>
  <value>zk1.myco.com,zk2.myco.com,zk3.myco.com</value>
</property>

  Important: Enabling the Table Lock Manager without specifying a list of valid Zookeeper quorum nodes will result in unpredictable behavior. Make sure that both properties are properly configured.

(The above settings are also needed if you are still using HiveServer1. HiveServer1 is deprecated; migrate to HiveServer2 as soon as possible.)

If ZooKeeper is not using the default value for ClientPort, you need to set hive.zookeeper.client.port in /etc/hive/conf/hive-site.xml to the same value that ZooKeeper is using. Check /etc/zookeeper/conf/zoo.cfg to find the value for ClientPort. If ClientPort is set to any value other than 2181 (the default), sethive.zookeeper.client.port to the same value. For example, if ClientPort is set to 2222, set hive.zookeeper.client.port to 2222 as well:

<property>
  <name>hive.zookeeper.client.port</name>
  <value>2222</value>
  <description>
  The port at which the clients will connect.
  </description>
</property>

The connection URL format and the driver class are different for HiveServer2 and HiveServer1:

jdbc:hive2://<host>:<port>

org.apache.hive.jdbc.HiveDriver

jdbc:hive://<host>:<port>

org.apache.hadoop.hive.jdbc.HiveDriver

HiveServer2 can be configured to authenticate all connections; by default, it allows any client to connect. HiveServer2 supports either Kerberos or LDAP authentication; configure this in the hive.server2.authentication property in the hive-site.xml file. You can also configure Pluggable Authentication, which allows you to use a custom authentication provider for HiveServer2; and HiveServer2 Impersonation, which allows users to execute queries and access HDFS files as the connected user rather than the super user who started the HiveServer2 daemon. For more information, see Hive Security Configuration.

  Warning: Because of concurrency and security issues, HiveServer1 and the Hive CLI are deprecated in CDH 5 and will be removed in a future release. Cloudera recommends you migrate to Beeline and HiveServer2 as soon as possible. The Hive CLI is not needed if you are using Beeline with HiveServer2.

HiveServer2 and HiveServer1 can be run concurrently on the same system, sharing the same data sets. This allows you to run HiveServer1 to support, for example, Perl or Python scripts that use the native HiveServer1 Thrift bindings.

<property>
  <name>hive.server2.thrift.port</name>
  <value>10001</value>
  <description>TCP port number to listen on, default 10000</description>
</property>

You can also specify the port (and the host IP address in the case of HiveServer2) by setting these environment variables: