Hue with Tez generates excessive jobs

to create new topics or reply. | New User Registration

This topic contains 1 reply, has 2 voices, and was last updated by  Dave 8 months, 1 week ago.

  • Creator
  • #55955

    Noam Cohen

    Hey guys –
    I’m running Hue over my HDP 2.1. It appears that when Hive is configured to use Tez as its execution engine (which is the default), Hue would generate excessive Tez applications for each query. This becomes a real problem when several queries are executed one after the other, because it fills up the Resource Manager queue and makes all jobs to freeze until timeout.
    This does not happen when execution engine is set to “mr” (MapReduce) or when the query is executed in “hive” or “hiveserver2″. It looks like a Hue-only problem.

    For example – see the log below. I ran a single query which generated 3 different Tez applications (jobs):
    “application_1401210649023_0227″ and “application_1401210649023_0228″ were created straight away. “application_1401210649023_0229″ was added once the query completed successfully.
    (I published the log as a Google drive document:)

    My assumption is that the “application_1401210649023_0227″ is a job used to describe the table, “application_1401210649023_0228″ is the query itself and “application_1401210649023_0229″ is used to format the output.

Viewing 1 replies (of 1 total)

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

  • Author
  • #64145


    Hi Noam,

    This is a known issue which is resolved in HDP 2.2 by using hiveserver2.
    We are working on investigating the root cause.



Viewing 1 replies (of 1 total)
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.