You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-user@hadoop.apache.org by Robert Grandl <rg...@student.ethz.ch> on 2011/02/14 10:34:35 UTC

Hadoop start scripts modification

Hi all,

I am trying to modify hadoop starting scripts such that each tasktracker 
from the conf/slaves to take his own configuration from a directory 
conf_addressTracker.

Can somebody point me out in which files I have to modify 
HADOOP_CONF_DIR variable such that the trackers to load their own 
configurations directories ?

I was trying to modify bin/hadoop-daemon.sh and replace HADOOP_CONF_DIR 
with the HADOOP_CONF_DIR_tracker.

However, if I am stopping a tracker and trying to start it again from 
the job tracker with a command like this:
ssh rgrandl@tracker_host 
'/home/rgrandl/School/Project/hadoop_compile/hadoop-0.20.2/bin/hadoop-daemon.sh 
--config conf_tracker_host start tasktracker'

I got an error like: Error: JAVA_HOME is not set.

I don't know why I get this as JAVA_HOME points to the right location to 
all conf/hadoop-env.sh files.

Can somebody help me out with some clues on the right way to solve my 
problem ?

Thanks,
Robert