Grokbase › Groups › Hadoop › common-user › August 2011
FAQ
Hi,

Has anybody been able to run hadoop standalone mode on fedora 15 ?
I have installed it correctly. It runs till map but gets stuck in reduce.
It fails with the error "mapred.JobClient Status : FAILED Too many
fetch-failures". I read several articles on net for this problem, all of them
say about the /etc/hosts and some say firewall issue.
I enabled firewall for the port range and also checked my /etc/hosts file. its
content is "localhost" and that is the only line in it.

Is sun-java absolute necessary or open-jdk will work ?

can someone give me some suggestion to get along with this problem ?

Thanks & regard

Manish

Search Discussions

  • Madhu phatak at Aug 5, 2011 at 1:25 pm ⇧
    disable iptables and try again
    On Fri, Aug 5, 2011 at 2:20 PM, Manish wrote:

    Hi,

    Has anybody been able to run hadoop standalone mode on fedora 15 ?
    I have installed it correctly. It runs till map but gets stuck in reduce.
    It fails with the error "mapred.JobClient Status : FAILED Too many
    fetch-failures". I read several articles on net for this problem, all of
    them
    say about the /etc/hosts and some say firewall issue.
    I enabled firewall for the port range and also checked my /etc/hosts file.
    its
    content is "localhost" and that is the only line in it.

    Is sun-java absolute necessary or open-jdk will work ?

    can someone give me some suggestion to get along with this problem ?

    Thanks & regard

    Manish

    --
    Join me at http://hadoopworkshop.eventbrite.com/
  • Harsh J at Aug 5, 2011 at 1:44 pm ⇧
    Sun JDK is what its been thoroughly tested upon. You can run on
    OpenJDK perhaps, but YMMV.

    Hadoop has a strict requirement of having a proper network setup before use.

    What port range did you open? TaskTracker would use 50060 for
    intercommunication (over lo, if its bound to that). Check if your
    daemons are binding to the right interfaces and have proper name->IP
    resolutions, and then check if that port is allowed communication
    upon.
    On Fri, Aug 5, 2011 at 2:20 PM, Manish wrote:
    Hi,

    Has anybody been able to run hadoop standalone mode on fedora 15 ?
    I have installed it correctly. It runs till map but gets stuck in reduce.
    It fails with the error "mapred.JobClient Status : FAILED Too many
    fetch-failures". I read several articles on net for this problem, all of them
    say about the /etc/hosts and some say firewall issue.
    I enabled firewall for the port range and also checked my /etc/hosts file. its
    content is "localhost" and that is the only line in it.

    Is sun-java absolute necessary or open-jdk will work ?

    can someone give me some suggestion to get along with this problem ?

    Thanks & regard

    Manish


    --
    Harsh J
  • John cohen at Apr 26, 2012 at 5:55 am ⇧
    I had the same issue. My problem was the use of VPN
    connected to work, and at the same time working
    with M/R jobs on my Mac. It occurred to me that
    maybe Hadoop was binding to the wrong IP (the IP
    given to you after connecting through VPN),
    bottom line, I disconnect from the VPN, and the M/R job
    finished as expected after that.
  • Marcos Ortiz at Apr 26, 2012 at 12:24 pm ⇧

    On 04/26/2012 01:49 AM, john cohen wrote:
    I had the same issue. My problem was the use of VPN
    connected to work, and at the same time working
    with M/R jobs on my Mac. It occurred to me that
    maybe Hadoop was binding to the wrong IP (the IP
    given to you after connecting through VPN),
    bottom line, I disconnect from the VPN, and the M/R job
    finished as expected after that.
    This is logic because, after that you configure to connect to the VPN,
    your machines have
    other IPs, based on the request of the private network. You can test,
    changing the IPs for the
    new ones based on the VPN request.

    --
    Marcos Luis Ortíz Valmaseda (@marcosluis2186)
    Data Engineer at UCI
    http://marcosluis2186.posterous.com



    10mo. ANIVERSARIO DE LA CREACION DE LA UNIVERSIDAD DE LAS CIENCIAS INFORMATICAS...
    CONECTADOS AL FUTURO, CONECTADOS A LA REVOLUCION

    http://www.uci.cu
    http://www.facebook.com/universidad.uci
    http://www.flickr.com/photos/universidad_uci

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupcommon-user @
categorieshadoop
postedAug 5, '11 at 8:56a
activeApr 26, '12 at 12:24p
posts5
users5
websitehadoop.apache.org...
irc#hadoop

5 users in discussion

Manish: 1 post John cohen: 1 post Madhu phatak: 1 post Marcos Ortiz: 1 post Harsh J: 1 post

Content

People

Support

Translate

site design / logo © 2022 Grokbase