[Lustre-discuss] operation X on unconnected MDS / MGS

[Thomas Roth]

Hi all,

for some time already I keep getting error messages on my MGS/MDS about 
operations on unconnected MGS or MDS.

 > Aug  1 18:08:53  kernel: LustreError: 
14182:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:09:19  kernel: LustreError: 
14182:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:09:57  kernel: LustreError: 
14182:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:12:48  kernel: LustreError: 
14182:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:13:12  kernel: LustreError: 
14183:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:13:53  kernel: LustreError: 
14182:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:18:17  kernel: LustreError: 
16000:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:21:32  kernel: LustreError: 
14183:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:25:33  kernel: LustreError: 
14323:0:(handler.c:1518:mds_handle()) operation 400 on unconnected MDS 
from 12345-IP at tcp
 > Aug  1 18:25:59  kernel: LustreError: 
14299:0:(handler.c:1518:mds_handle()) operation 400 on unconnected MDS 
from 12345-IP at tcp
 > Aug  1 18:26:37  kernel: LustreError: 
16000:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:29:33  kernel: LustreError: 
14182:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 101 on 
unconnected MGS
 > Aug  1 18:30:09  kernel: LustreError: 
14312:0:(handler.c:1518:mds_handle()) operation 400 on unconnected MDS 
from 12345-IP at tcp
 > Aug  1 18:30:33  kernel: LustreError: 
10570:0:(handler.c:1518:mds_handle()) operation 35 on unconnected MDS 
from 12345-IP1 at tcp
 > Aug  1 18:33:53  kernel: LustreError: 
14182:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:33:53  kernel: LustreError: 
14300:0:(handler.c:1518:mds_handle()) operation 400 on unconnected MDS 
from 12345-IP2 at tcp
 > Aug  1 18:38:12  kernel: LustreError: 
14182:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:38:53  kernel: LustreError: 
14321:0:(handler.c:1518:mds_handle()) operation 400 on unconnected MDS 
from 12345-IP1 at tcp
 > Aug  1 18:50:33  kernel: LustreError: 
14299:0:(handler.c:1518:mds_handle()) operation 400 on unconnected MDS 
from 12345-140.181.106.156 at tcp
 > Aug  1 18:50:59  kernel: LustreError: 
14308:0:(handler.c:1518:mds_handle()) operation 400 on unconnected MDS 
from 12345-IP at tcp
 > Aug  1 18:51:38  kernel: LustreError: 
14182:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS
 > Aug  1 18:58:53  kernel: LustreError: 
14305:0:(handler.c:1518:mds_handle()) operation 400 on unconnected MDS 
from 12345-IP2 at tcp
 > Aug  1 18:59:13  kernel: LustreError: 
14326:0:(handler.c:1518:mds_handle()) operation 41 on unconnected MDS 
from 12345-IP at tcp
 > Aug  1 18:59:38  kernel: LustreError: 
14302:0:(handler.c:1518:mds_handle()) operation 101 on unconnected MDS 
from 12345-IP1 at tcp
 > Aug  1 19:03:12  kernel: LustreError: 
14183:0:(mgs_handler.c:538:mgs_handle()) lustre_mgs: operation 400 on 
unconnected MGS


According to the Change Log 1.6.4.3->1.6.5, at least  there should be no 
more  'operation 101 on unconnected MGS', since the servers are all 
running v 1.6.5

As far as I can tell, both MGS and MDS where perfectly available to 
whole time.
As I'm still relying on Google, and searching for these messages has 
brought up nearly nothing: what do they mean? Is this something to worry 
about?

Thanks,
Thomas