I just configured a Debian in pseudo distributed mode, Java 7 and Hadoop 2.7.2. I'm running the experiments, and when I run a big pi calculation hadoop dies (everything... datanode, nodemanager, namenode, etc), all of them receive a SIGTERM 15. But not only hadoop, even the connection to hadoopuser is disconnected. It is like the operating system (or something) is killing everything open by hadoopuser when the map-reduce starts.
Anyone know what can be happening?
