The Hortonworks Community Connection is now live. A completely rebuilt Q&A forum, Knowledge Base, Code Hub and more, backed by the experts in the industry.

You will be redirected here in 10 seconds. If your are not redirected, click here to visit the new site.

The legacy Hortonworks Forum is now closed. You can view a read-only version of the former site by clicking here. The site will be taken offline on January 31,2016

HDFS Forum

YARN – 'availability' mechanisms

  • #24560
    Tobias Herb

    I´m a student and just experimenting with YARN (I hope this is the right place to post, if not correct me please!). I write a simple YARN demo (at the moment on a single node setup) to understand the framework components and their interactions. At the moment I play with some ‘Availability’ mechanisms of YARN. In one simple scenario I kill the local ‘ApplicationMaster (AM)’ process and would expect that the RM restarts the AM from alone and re-execute the embedded job (the tasks that haven´t been run), but unfortunately nothing happens, no restarting/reallocation of the AM happen. Must this logic be implemented on the client side? Or did something else forgot? I expect this behaviour on the basis of the design document “Architecture of Next Generation Apache Hadoop MapReduce Framework”.

    Thanks in advance

  • Author
  • #24562
    Yi Zhang

    Hi Tobi,

    This question may get a better answer from the HDP 2.0 Alpha forum.

    The AM can reschedule tasks if a task fails. AM negotiates resources with RM, but RM doesn’t restart AM. They are of different realm.


The topic ‘YARN – 'availability' mechanisms’ is closed to new replies.

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.