Lost of vertex when running a Bagel based program in non-local mode

[杨强]

Hi.

I wrote a program to implement KMeans algorithm with Bagel.
The program runs OK in local mode.
While when I try to run it in the cluster, it seems that the Bagel.run method
returns only part of vertices. That is, I used test data with 12 nodes to run 
the 
program and only 6 nodes was returned.

I'm using spark-0.8.0. The source file in attached.  Hope you can give me some 
advice.

Thanks,
Qiang Yang

BagelKMeans.scala
Description: Binary data