HDFS Forum

sort example not running, /dev/mapper/vg_node1-lv_root directory is getting full

  • #48403
    Mitali Sodhi
    Participant

    Hi,

    I have a 4 node cluster setup, each with 32G RAM and 1T Hard Disk. I was trying to run the sort example, so for that I first ran the randomwriter example.
    hadoop jar hadoop-mapreduce-examples-2.2.0.2.0.6.0-101.jar randomwriter random-data

    The randomwriter example is running successfull and data is getting written on the hdfs. But, in sort example the job is getting errors.
    The job is running 100% for map and stopping after 40% of reduce. I am getting DiskErrorException : Could not find any valid local directory for output. I ran this command when the job was in progress:
    $ df -h
    This, showed the that /dev/mapper/vg_node1-lv_root directory which has 50G storage capacity is getting fulled when the reduce is reaching around 40%. Here is the output of this command:

    So, what should I do to rectify this problem?

    Thanks,
    Mitali

to create new topics or reply. | New User Registration

  • Author
    Replies
  • #49266
    Koelli Mungee
    Moderator

    Hi Mitali,

    How big is your cluster, is it multi-node -if so how big? Looking at the error you are reaching the disk capacity. You could potentially add more space to the mount or delete some unwanted files to free up disk space.

    thanks
    koelli

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.