Glad you figured out your bug. It's worth pointing out that Google's
MapReduce has stricter type signatures than Hadoop does. Hadoop allows the
final (k, v) pairs to be of a different type than the inputs to the reducer,
and the final key associated with a final value does not necessarily need to
be the same as the intermediate key.

i.e.:
map:     (k1, v1) -> list(k2, v2)
reduce: (k2, list(v2)) -> list(k3, v3)

- Aaron

On Sun, Mar 21, 2010 at 10:10 AM, welman Lu <[email protected]> wrote:

> Hi, all
>
> I solved it!
> By through set:
>
>         conf.setMapOutputKeyClass(Text.class);
>         conf.setMapOutputValueClass(BytesWritable.class);
>         conf.setOutputKeyClass(Text.class);
>         conf.setOutputValueClass(Text.class);
>
> So stupid I am! I need to read more documents first!
>
> Best Regards
> Jiamin Lu
>
>
>

Reply via email to