A few tweaks to conform with the new packaging.

hadoop/README.txt

@@ -6,8 +6,8 @@ Note: You may want to do all of this from the machine where you plan
 to launch the hadoop jar job from.  Otherwise you will end up having
 to copy files around.
 
-(If you grabbed a prebuilt h2o_hadoop.zip file, copy it to a hadoop 
-machine and skip to the PREPARE section below.)
+(If you grabbed a prebuilt h2o-*.zip file, copy it to a hadoop machine
+and skip to the PREPARE section below.)
 
 
 GET H2O TREE FROM GIT
@@ -34,6 +34,7 @@ PREPARE JOB INPUT ON HADOOP NODE
 
 $ unzip h2o-*.zip
 $ cd h2o-*
+$ cd hadoop
 
 Create flatfile.txt.
 
@@ -57,7 +58,7 @@ Java exceptions if DNS and DHCP are not fully configured.
 This generator tool is still experimental, please double check the 
 output yourself before relying on it.
 
-$ hadoop jar hadoop/h2odriver_cdh4.jar water.hadoop.gen_flatfile -jt <jobtracker:port> > flatfile.txt
+$ hadoop jar h2odriver_cdh4.jar water.hadoop.gen_flatfile -jt <jobtracker:port> > flatfile.txt
 
 (Note: Make sure to use the h2odriver flavor for the correct version
        of hadoop!  We recommend running the hadoop command from a
@@ -70,7 +71,7 @@ $ hadoop jar hadoop/h2odriver_cdh4.jar water.hadoop.gen_flatfile -jt <jobtracker
 RUN JOB
 -------
 
-$ hadoop jar hadoop/h2odriver_cdh4.jar water.hadoop.h2odriver [-jt <jobtracker:port>] -files flatfile.txt -libjars h2o.jar -mapperXmx 1g -nodes 1 -output hdfsOutputDirName
+$ hadoop jar h2odriver_cdh4.jar water.hadoop.h2odriver [-jt <jobtracker:port>] -files flatfile.txt -libjars ../h2o.jar -mapperXmx 1g -nodes 1 -output hdfsOutputDirName
 
 (Note: -nodes refers to H2O nodes.  This may be less than or equal to
        the number of hadoop machines running TaskTrackers where hadoop