diff --git a/tracker/rabit_hadoop.py b/tracker/rabit_hadoop.py
index a283e0ca7..20bb47e7f 100755
--- a/tracker/rabit_hadoop.py
+++ b/tracker/rabit_hadoop.py
@@ -50,7 +50,7 @@ parser.add_argument('-f', '--files', nargs = '*',
                         ' You can also use it to manually cache files when auto_file_cache is off')
 parser.add_argument('--jobname', default='auto', help = 'customize jobname in tracker')
 parser.add_argument('--timeout', default=600000000, type=int,
-                    help = 'timeout of each mapper job, automatically set to a very long time,'\
+                    help = 'timeout ((in milli seconds)) of each mapper job, automatically set to a very long time,'\
                         'normally you do not need to set this ')
 parser.add_argument('-m', '--memory_mb', default=-1, type=int,
                     help = 'maximum memory used by the process, Guide: set it large (near mapred.cluster.max.map.memory.mb)'\
@@ -74,7 +74,7 @@ parser.add_argument('command', nargs='+',
 args = parser.parse_args()
 
 if args.jobname == 'auto':
-    args.jobname = ('Rabit(nworker=%d):' % args.nworker) + args.command[0].split('/')[-1];
+    args.jobname = ('Rabit[nworker=%d]:' % args.nworker) + args.command[0].split('/')[-1];
 
 def hadoop_streaming(nworker, worker_args):
     fset = set()