Estoy tratando de ejecutar un nodo de dos clústeres en hadoop 0.9.8

Master is DUT5 and Slave is DUT4..

I am trying to run a two cluster node on hadoop 0.9.8. While running start-dfs, it is showing fatal error:

[root@dut5 bin]# ./start-dfs.sh
starting namenode, logging to /hadoop-rdma-0.9.8/libexec/../logs/hadoop-root-namenode-dut5.out
dut5: ssh: relocation error: /lib64/libkrb5.so.3: symbol k5_path_join, version krb5support_0_MIT not defined in file libkrb5support.so.0 with link time reference
dut4: datanode running as process 11910. Stop it first.
dut5: ssh: relocation error: /lib64/libkrb5.so.3: symbol k5_path_join, version krb5support_0_MIT not defined in file libkrb5support.so.0 with link time reference

and hadoop log shows

Java Runtime Environment ha detectado un error fatal:
SIGSEGV (0xb) at pc=0x000000335ba4812c, pid=4368, tid=140618348889856
Marco problemático:
dut5: # C [libc.so.6+0x4812c] _IO_vfprintf+0x3e5c

resuelto

The problem was with the configuration file.. disabling roce made hadoop working