What Anil says. Sounds like your job is launched with default configs -- which are for local mode. You need to point it at your distributed cluster install.
For MapReduce jobs, HADOOP_CLASSPATH needs to be set appropriately. On Thursday, May 26, 2016, anil gupta <[email protected]> wrote: > Hi, > > It seems like your classpath is not setup correctly. /etc/hadoop/conf and > /etc/hbase/conf needs to be in MapReduce classpath. Are you able to run > row counter job of hbase in distributed cluster? What version of Hadoop you > are using? Did you use Ambari or Clodura Manager to install the cluster? > > Thanks, > Anil Gupta > > On Thu, May 26, 2016 at 7:19 AM, Lucie Michaud < > [email protected] > <javascript:_e(%7B%7D,'cvml','[email protected]');>> > wrote: > >> Hello everybody, >> >> >> >> For a few days I developed a MapReduce code to insert values in HBase >> with Phoenix. But the code runs only in local and overcharge the machine. >> >> Whatever changes I make I observe that the mapred.LocalJobRunner class is >> systematically used. >> >> >> >> Do you have an idea of the problem? >> >> >> >> I attached to this post the execution logs of my program. >> >> >> >> Thank you in advance for your help. :) >> >> Feel free to ask me for more details if it can help. >> > > > > -- > Thanks & Regards, > Anil Gupta >
