Hi we are seeing exactly the same issue where I work:
Sat Sep 17 2016 20:02:50 +01:00 [kern_ndmpd:info:4503] [ 4503] INFO: Starting ndmp for client [10.54.48.51].61726
0000001c.0002a6b5 050f8ad1 Sat Sep 17 2016 20:24:54 +01:00 [kern_ndmpd:info:4503] [ 4503] INFO: Starting ndmp for client [10.54.48.51].63553
0000001c.0002a6b6 050f8ad1 Sat Sep 17 2016 20:24:54 +01:00 [kern_ndmpd:info:4503] [ 4503] INFO: Starting ndmp for client [10.54.48.51].63556
0000001c.0002a6b7 050f8b29 Sat Sep 17 2016 20:25:02 +01:00 [kern_ndmpd:info:4503] [14875] DEBUG: NDMP Connect: Successfully connected to [10.54.48.51].63617, using local address [10.54.48.58].1035 (-0:00:01)
0000001c.0002a6b8 050f8b29 Sat Sep 17 2016 20:25:02 +01:00 [kern_ndmpd:info:4503] [14875] NOTICE: BASE_DATE '0' => L=0 T=0.0 level=0
0000001c.0002a6b9 050f8b29 Sat Sep 17 2016 20:25:02 +01:00 [kern_ndmpd:info:4503] [14875] INFO: START BACKUP Doing a LEVEL 0 backup
0000001c.0002a6bb 050f8b29 Sat Sep 17 2016 20:25:02 +01:00 [kern_ndmpd:info:4503] [14875] INFO: DATA: performing token full backup
0000001c.0002a6bc 050f8b29 Sat Sep 17 2016 20:25:03 +01:00 [kern_ndmpd:info:4503] [14875] INFO: DATA notification state 1 (SESS_STATE: LISTENING) event 10 (EVENT: STATE CHANGE) (-0:00:01)
0000001c.0002a6bd 050f8b29 Sat Sep 17 2016 20:25:03 +01:00 [kern_ndmpd:info:4503] [14875] INFO: DATA notification state 4 (SESS_STATE: ACTIVE) event 10 (EVENT: STATE CHANGE) (-0:00:01)
0000001c.0002a6be 050f8be5 Sat Sep 17 2016 20:25:22 +01:00 [kern_ndmpd:info:4503] [14875] INFO: buffer full for ndmp_io_process (fd=8) count total 35876 (5) (-0:00:01)
0000001c.0002a6bf 050f8c7b Sat Sep 17 2016 20:25:37 +01:00 [kern_ndmpd:info:4503] [14875] INFO: buffer full for ndmp_io_process (fd=8) count total 35876 (20) (-0:00:01)
0000001c.0002a6c0 050f8d15 Sat Sep 17 2016 20:25:51 +01:00 [kern_ndmpd:info:4503] [14875] INFO: buffer full for ndmp_io_process (fd=8) count total 35876 (35) (-0:00:01)
0000001c.0002a6c2 050f8da7 Sat Sep 17 2016 20:26:06 +01:00 [kern_ndmpd:info:4503] [14875] INFO: buffer full for ndmp_io_process (fd=8) count total 35876 (50) (-0:00:01)
0000001c.0002a6c3 050f8e40 Sat Sep 17 2016 20:26:22 +01:00 [kern_ndmpd:info:4503] [14875] INFO: buffer full for ndmp_io_process (fd=8) count total 35876 (65) (-0:00:01)
Our media servers are able to connect to other filers over other networks, but this filer is on our new NGNI network which is the only one we are seeing these issues.
I was expecting switch port configuration because I was unable to telnet to the NDMP port on the filer from the same subnet as the media server and yet I am able to connect to the NDMP port from a server on the same subnet as the filer!?
Obviously from what has been posted previously it can't be that as they managed to backup a small amount of data, it seems to fail when open for a longer period.
I will let you know if I find anything out from our network team.