1. You can find out why the namenode is in safemode with "hdfs dfsadmin -safemode get"
2. I don't understand-- can you be more detailed? What are you out of, memory or disk space?
3. Is the replication factor greater than the number of datanodes?
4. is the DFS down?