I have hadoop up and running, processing jobs, both in a local and in a
clustered configuration. It works fine.
I now have a simple pig script written, which works fine in local mode. But when
run on hadoop in a single node mode, it hangs at the first reduce. The single
node config is exactly as per the hadoop docs - hdfs on localhost:9000, job
tracker at 9001. Java version 1.6.0_21_b06 on Unbuntu 10.4.
Hadoop lives in ~/work/hadoop-0.20.2 and pig lives in ~/work/pig-0.7.0, with the
same config files in the respective conf directories (otherwise pig just runs in
Is there some sort of configuration magic I need to do here? I read the pig
wiki, the hadoop docs and wiki, and could not find anything on it, and anything
google turns up is from 2008 and some 2009.
I also tried pigpen, and this seems to insist on the deprecated hadoop-site.xml
file, and even when it finds it, nothing other than the graphic representation
of the script and the editor seem to work - everything else causes it to throw a
null pointer exception. This is with Eclipse Helios, so I suppose I'd need to go
back to Galileo or something. Is this correct?
Thanks in advance for your help!