NAME

Thrift::API::HiveClient2 - Perl to HiveServer2 Thrift API wrapper

VERSION

version 0.011

METHODS

new

Initialize the client object with the Hive server parameters

    my $client = Thrift::API::HiveClient2->new(
        host    => <host name or IP, defaults to localhost>,
        port    => <port, defaults to 10000>,
        timeout => <seconds timeout, defaults to 1 hour>,
    );

connect

Open the connection on the server declared in the object's constructor.

     $client->connect() or die "Failed to connect";

execute

Run an HiveQl statement on an open connection.

    my $rh = $client->execute( <HiveQL statement> );

fetch

Returns an array(ref) of arrayrefs, like DBI's fetchall_arrayref, and a boolean indicator telling wether or not a subsequent call to fetch() will return more rows.

    my ($rv, $has_more_rows) = $client->fetch( $rh, <maximum records to retrieve> );

IMPORTANT: The version of HiveServer2 that we use for testing is the one bundled with CDH 4.2.1. The hasMoreRows method is currently broken, and always returns false. So the right way of obtaining the resultset is to keep using fetch() until it returns an empty array. For this reason the behaviour of fetch has been altered in scalar context (which becomes the current advised way of retrieving the data):

    # $rv will be an arrayref is anything was fetched, and undef otherwise.
    #
    while (my $rv = $client->fetch( $rh, <maximum records to retrieve> )) {
        # ... do something with @$rv
    }

This is the approach adopted in https://github.com/cloudera/hue/blob/master/apps/beeswax/src/beeswax/server/hive_server2_lib.py

WARNING

Thrift in Perl currently doesn't support SASL, so authentication needs to be disabled for now on HiveServer2 by setting this property in your /etc/hive/conf/hive-site.xml. Although the property is documented, this *value* -which disables the SASL server transport- is not, AFAICT.

  <property>
    <name>hive.server2.authentication</name>
    <value>NOSASL</value>
  </property>

CAVEATS

The instance of hiveserver2 we have didn't return results encoded in UTF8, for the reason mentioned here: https://groups.google.com/a/cloudera.org/d/msg/cdh-user/AXeEuaFP0Ro/Txmn1OHleAsJ

So we had to change the init script for hive-server2 to make it behave, adding '-Dfile.encoding=UTF-8' to HADOOP_OPTS

AUTHOR

David Morel <david.morel@amakuru.net>

COPYRIGHT AND LICENSE

This software is Copyright (c) 2013 by David Morel & Booking.com. Portions are (c) R.Scaffidi, Thrift files are (c) Apache Software Foundation..

This is free software, licensed under:

  The Apache License, Version 2.0, January 2004

cpanm Thrift::API::HiveClient2

perl -MCPAN -e shell
install Thrift::API::HiveClient2