HDP on Linux – Installation Forum

Hadoop HA

  • #38036
    lou82
    Participant

    Just looking to see if anyone has set up HA on Centos servers. Seems like a pretty large task in that it wants the name nodes to be in RHEL Cluster – any comments?

    Also – anyone using bonded NICS in their hadoop cluster?

to create new topics or reply. | New User Registration

  • Author
    Replies
  • #38189
    Dave
    Moderator

    Hi Lou,

    Currently HDP 1.3.2 supports HA in RHEL or VMWare cluster.

    HDP 2.0 has HA of it’s own, which you can try out using the Beta version:
    http://docs.hortonworks.com/HDPDocuments/HDP2/HDP-2.0.5.0/bk_system-admin-guide/content/ch_hadoop-ha.html

    I have seen bonded NICS used in environments before and they should not cause any issues.

    Thanks

    Dave

    #38191
    lou82
    Participant

    Oh Ok so with the 2.0 you can just use that HA set up process. Do you need to have a backup server for the namenode and a server for the secondarynamenode or can I have one server for both?

    Anything special with the bonded NICs? Is that a better performance? I did read sometimes it can have the opposite effect and cause bottle necks.

    I appreciate the feedback
    thanks

    #38205
    Dave
    Moderator

    Hi Lou,

    It really depends on your environment. You can have a backup server for everything and this is fine if you intend on bringing the unavailable primary back up.
    Usually 1 server for both is fine as long as you do not intend to use it for an extended period of time (as it will run low on resource)

    As for the bonded NICs, I can’t comment as I’m not a network guy, but for anything which relies heavily on network and connectivity, I would recommend to use a Physical NIC.

    Thanks

    Dave

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.