hadoop-common-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Harsh J <qwertyman...@gmail.com>
Subject Re: Running JobTracker on a separate machine
Date Tue, 22 Mar 2011 04:36:19 GMT
start-all.sh will attempt to launch a JobTracker on the same machine
it was executed on (which would fail since mapred.job.tracker would
point to an unavailable binding address). The right way is to launch
these individually (or perhaps writing a custom/modified script to do
what you need).

On Tue, Mar 22, 2011 at 12:00 AM, modemide <modemide@gmail.com> wrote:
> Hello all,
> I was wondering if someone could help me with this issue.
> I was trying to configure the JobTracker to run on a separate machine
> as the Namenode (per the documentation's recommendation).
>
> When I try to configure it and run ./start-all.sh , the NameNode tries
> to start the JobTracker on itself and fails with error.  However, it
> starts all other necessary daemons.  This is the error on the
> NameNode:
> Problem binding to JOBTRACKER_HOSTNAME/10.1.10.2:54311
>
>
> I am able to start the jobtracker individually on the jobtracker
> machine that I want to use and everything works fine.  But, I was
> under the impression that I could use the ./start-all.sh script on the
> NameNode to start all the machines in the cluster.
>
> Any help would be greatly appreciated!
>
> Thank You,
>



-- 
Harsh J
http://harshj.com

Mime
View raw message