Hi Antonio,
Are you sure you copied the right hadoop-env.sh to all cluster nodes and
restarted the MapReduce framework (with stop-mapred.sh and start-mapred.sh)?
Also you should try first running the row counter as a Driver like so:
./bin/hadoop <path-to-hbase>/hbase-0.20.2jar rowcounter mytable colfam1
colfam2 ...
It needs to have some columns to scan or else it will report a NPE later
in the TableInputFormat.
Lars
Antonio D'Ettole schrieb:
Hi folks,
I was trying to run the RowCounter example on my hadoop cluster. I followed
the instructions at
http://hadoop.apache.org/hbase/docs/current/api/org/apache/hadoop/hbase/mapreduce/package-summary.htmland
modified my HADOOP_CLASSPATH . I can tell this works because by
invoking
./bin/hadoop org.apache.hadoop.hbase.mapreduce.RowCounter mytable
the rowcounter example starts and the job is submitted to the jobtracker.
Alas, all task attempts exit with this exception:
Error: java.lang.ClassNotFoundException: org.apache.zookeeper.Watcher
at java.net.URLClassLoader$1.run(URLClassLoader.java:217)
at java.security.AccessController.doPrivileged(Native Method)
[............]
I'm _sure_ I added the right zookeper library to the path because when I
remove it the process exits immediately with the exception
Exception in thread "main" java.lang.NoClassDefFoundError:
org/apache/zookeeper/Watcher
and the process isn't even launched. By putting the jar file back into the
classpath, the job is submitted and started but i'm back to the first
exception. Does anyone have an idea what I could be doing wrong?
Thanks
Antonio
