dlsw circuits random disconnect
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
03-24-2006 02:12 PM
Hello!!
Could somebody help me with this?
We have a dlsw network of almost 1600 remote dlsw peers.
We have six central peer routers.
Sometimes we saw some SNA PU´s down, and after 1 to 3 min return to "CONNECTED" (dlsw circuit). We dont´n see any dlsw peer down, only dlsw circuit falls.
On branch office router we saw by the command "show dlsw ciscuits history detail", that the event before disconnect is
Event
WAN halt-dl
What does this mean?
It is possibly a problem with the Central Peer Router?
The Carrier confirm us no problem with the WAN Network.
The Central Routers are Cisco7200VXR NPE400 512DRAM and IOS 12.4(4)T1
The branch office routers are Cisco1760 IOS 12.3(9).
Thanks for help!!
Pedro.
- Labels:
-
Server Networking
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
03-26-2006 02:48 AM
I'd look into monitoring the CPU usage (as well as the amount of memory used by DLSw Process) of the 1700 series branch routers. It may be that these TCP based connections from the 6 central routers are overwhelming them.
(i.e., http://www.cisco.com/warp/public/63/highcpu_processes.html#tcp_timer)
Especially if the remote routers are pushing a fair amount of traffic with process switching turned on the intfs (i.e., ACLs with the 'log' option on) or if you have stuff like SNMP polls from network mgmt software going.
Also...................
"Disconnectphase. When an LLC end station wishes
to terminate one of its existing connections, it sends
a disconnect (DISC) frame to the other end station,
which responds with UA. The same frame types are
defined for SDLC. With DLSW, these frames are reflected
between partners using the SSP messages
halt-dl (halt data link) and dl-halted (data link halted)
and the circuit is then disconnected. Another DLSW
disconnect scenario is the loss of a transport connection
due to intermediate router failure. When
a DLSw node detects such a failure, it performs a
local disconnect of all connected data links that
were using the failed transport connection."
from: http://www.research.ibm.com/journal/sj/343/gayek.pdf
**Hope this helps**
Based on your earlier statement...
"On branch office router we saw by the command "show dlsw ciscuits history detail", that the event before disconnect is
Event
WAN halt-dl
What does this mean? "
I'd look into the process utilization on the Branch routers
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
04-26-2006 04:05 PM
Thanks for help Donald!!
I presume that the problem is on Central Routers.
Maybe some variable like memory, cpu etc is exceding a umbral.
I don´t think the the problems comes from de remote side, because we don´t see any bad indication.
What does "intermediate router failure" means? Did you now some escenarios? The Carrier tell us that they don´t have any failure on the MPLS cloud.
Regards..
