Hive / HCatalog Forum

Hive+HBase table creation freezing and failing

  • #43737
    Thomas Vial
    Member

    Hi!
    We are trying to create HBase-backed tables in Hive. We are using HDP 2 with HBase 0.95.2.2.0.5.0-67 and Hive 0.11.0.2.0.5.0-67 or 0.12.0.2.0.6.0-76 (neither will do).

    Table creation is failing, and we are wondering why. The classpath appears to be correctly set, so does the connection to zookeper, after setting HADOOP_CLASSPATH, hbase.zookeeper.quorum and hbase.master.

    When we issue the CREATE TABLE statement, HIve client freezes for a minute or so then fails with “java.io.IOException: Failed to find location, tableName=hbase:meta, …” on the client. Apparently, it’s happening when Hive is reading table metadata from HBase.

    Any idea what might be causing that?

    Thanks!

to create new topics or reply. | New User Registration

  • Author
    Replies
  • #43745
    Nick Dimiduk
    Moderator

    Hi Thomas,

    Can you try including the hbase config directory in your hive classpath? I usually specify:

    $ export HADOOP_CLASSPATH=/etc/hbase/conf:/usr/lib/hbase/hbase-security.jar:…
    $ hive …

    Also note that there is a known bug in the combination of HBase + Hive we shipped in HDP-2.0; you won’t be able to load data into HBase from Hive. See HIVE-5515. I believe we’ll be shipping that patch with the next patch release. In the mean time, things should work for you on a HDP-1.3 release (HBase 0.94 + Hive 0.11).

    Cheers,
    Nick

You must be to reply to this topic. | Create Account

Support from the Experts

A HDP Support Subscription connects you experts with deep experience running Apache Hadoop in production, at-scale on the most demanding workloads.

Enterprise Support »

Become HDP Certified

Real world training designed by the core architects of Hadoop. Scenario-based training courses are available in-classroom or online from anywhere in the world

Training »

Hortonworks Data Platform
The Hortonworks Data Platform is a 100% open source distribution of Apache Hadoop that is truly enterprise grade having been built, tested and hardened with enterprise rigor.
Get started with Sandbox
Hortonworks Sandbox is a self-contained virtual machine with Apache Hadoop pre-configured alongside a set of hands-on, step-by-step Hadoop tutorials.
Modern Data Architecture
Tackle the challenges of big data. Hadoop integrates with existing EDW, RDBMS and MPP systems to deliver lower cost, higher capacity infrastructure.