MRv2 job output compression issue

to create new topics or reply. | New User Registration

This topic contains 1 reply, has 2 voices, and was last updated by  Koelli Mungee 1 year, 7 months ago.

  • Creator
    Topic
  • #43357

    Hi all

    I’m using HDP 2.0 with MapReduce V2. I wanted to compress output of my M/R jobs so I set the properties mapreduce.map.output.compress and mapreduce.output.fileoutputformat.compress to true in mapred-site.xml using ambari. However, the job did not compress the output and both those properties were false in the configuration viewed by application master. The explanation in app master was:

    job.xml <- because mapreduce.output.fileoutputformat.compress is deprecated

    job.xml <- because mapreduce.map.output.compress is deprecated

    Hadoop documentation says that those properties are in fact new MR v2 properties so it’s strange that application master has marked them as deprecated and even stranger that it changed them from their values in mapred-site.xml (they were not overridden by my job definitions).

    It may be worth mentioning that I defined and submitted the job using HUE gui. It seems that HUE uses oozie to actually submit the job for the execution.

    Does anyone know what could be the problem?

Viewing 1 replies (of 1 total)

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

  • Author
    Replies
  • #43669

    Koelli Mungee
    Moderator

    Hi Miljan

    I am testing this on my side as well. In the mean time can you also try setting

    mapreduce.output.fileoutputformat.compress.codec
    mapreduce.map.output.compress.codec

    in addition to the properties you set to true?
    -Koelli

    Collapse
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.