Hi,

Your failure seems to be in the task-side. I suspect a mix of libraries.

What version of Hadoop are you *deploying* across all nodes?

On Mon, Sep 10, 2012 at 3:56 PM, Li Li <[EMAIL PROTECTED]> wrote:
> hi all,
>     I am trying an example from an tutorial for version 0.19 by using
> hadoop 1.0.3.

Why do that when 1.0.3 has its own examples and tutorials at
http://hadoop.apache.org/common/docs/stable/?

>     and it compiles but fail when running in pseudo distributed mode.

I packaged your code from the below link and am able to run it just
fine on 1.0.3. and a bunch of other versions I have on my machine.

Could you share your specific error please?

>     I found a question in stackoverflow
> http://stackoverflow.com/questions/11961517/new-to-hadoop-mapreduce-getting-a-nosuchmethodexception-on-mapred-reducer-init
>     if I don't use older version, I can't run older codes?

The answer there is bogus. There is no such issue that makes you not
use mapred.* API over mapreduce.*. The problem was something else that
should've taken some more debugging to find out, but the asker now
believes it was the use of the stable API.

--
Harsh J