Hi,
I would say the approach you have suggested violates the idea of an
OutputFormat. The OutputFormat will simply define how the <key,value> pairs
are written to disk. To achieve what you want I feel like it should be a
separate MapReduce phase that comes after your original one.
Thank you,
Saliya
On Mon, Apr 19, 2010 at 7:45 AM, Utku Can Topçu <u...@topcu.gen.tr> wrote:
> Hello All,
>
> I've been looking for the documentation and the API docs,
> however I couldn't find an OutPutFormat that will go through all the reduce
> emits as key:value pairs by running an execute method for each pair.
>
> Say, I've reduced hostnames as keys, and want to ping every hostname that
> I've reduced as key and store the results of pings in a MySQL table or so.
>
> So basically,
> What I'm thinking of was such a method like:
>
> outputExec(K key, V value){
> ...
> pingResult = ping(key)
> writetoMysql(value,pingResult)
> ....
> }
>
> Is there such an OutPutFormat implementation currently in Hadoop? If not
> so, I'd like to start doing something that fits my needs and try to
> contribute in to the project, but I think I'll need some guidance beforehand
> dealing with the OutPutFormat.
>
> Best Regards,
> Utku
>
>
>
>
>
--
Saliya Ekanayake
http://www.esaliya.blogspot.com
http://www.esaliya.wordpress.com
