I uploaded bunch .txt data into distributed file system. Cluster summary shows there is enough heap size memory. However, whenever I ran Mahout 0.8 seqdirectory command. The system displayed out of heap memory exception. I shutdown hadoop cluster and allocated more memory to mapred.child.java.opts. I then restarted the hadoop cluster and the log shows that namenode is corrupted. I tried to check the health of file system. However, the connection is refused. It ended out I have to reformat the system. I'm caught between this circle. Any help is appreciated.
