Problem in a lot of small jobs.

[Buyung Bahari]

Dear hadoop expert,

I want to ask, i have a lot of small job, i sending all in parallel, so 
the job queue in hadoop,  and the batch process every 15 minutes.

After 15 minutes, i erase all the output and the data. Why after a few 
days run, the hdfs run out of space? I try to run balancer and erase the 
temporary job, but dont help.

And second question, how to manage job.xml di hadoop logs directory?

I'm using Hadoop 0.18.3 and Pig for map and reduce.

Thanks for the advice.




Best Regards


Buyung Bahari