From: Joe Chang (changjoe@earthlink.net)
Date: Thu Mar 27 2003 - 14:35:04 GMT-3
----- Original Message -----
From: "David Ham" <ccieau@yahoo.com>
To: <ccielab@groupstudy.com>
Sent: Thursday, March 27, 2003 3:26 AM
Subject: some DLSW circuit drop on MSFC
> Hi all,
>
> I have urgent question to ask. I am having trouble to
> keep the number of dlsw circuits to be stable in
> production area.
>
> Senario is like this
>
> <--- 6509 MSFC --- trunk --- 4006
>
> Since 4006 is not capable of handling DLSW, I have set
> up a ISL trunk to 6509 and creat vlans and bridge the
> vlans into DLSW on 6509 MSFC.
>
> Objective is that I need get about 20 vlans across and
> bridge them into DLSW.
>
> So far I can see DLSW peer is established and can see
> some dlsw circuits that are connected. But the issue
> is that I can see only about 10 circuits are stable
> and rest of them are repeatly droped out after having
> about 10 sec connection and so on.
>
>
> Have you guys seen something like that??
> Please let me know if you have any idea.
>
> Regards,
>
> David Ham
Are each of the SNA terminals on a separate VLAN on the 4006? If so, try
moving them all on the same VLAN. Here's a possible reason your scenario is
not working.
A logical view of your network:
<--- 6509 MSFC -- vlan x --- 4006 --- terminal x
-- vlan y --- --- terminal y
-- vlan z --- --- terminal z
When the first explorer is received by the 6509 (say for terminal x) the
reachability caches will be empty. 6509 has to send a SNA test frame out
each of its dlsw enabled LANs, in this case the VLAN interfaces. The source
mac of the replicated test frames will not be that of the vlan, but of the
head-end host that is initiating all this. The 4006 will receive one copy of
the test frame, associate the SMAC with the vlan in its CAM table. Then the
switch may receive another copy of the frame, rewrite the CAM so that the
SMAC is associated with another vlan, and so on. Receiving the test frame,
terminal x will respond with its own frame, which the 6509 may or may not
receive from the correct vlan depending on the state of the CAM at that
moment. A circuit will be established between the DLSW peers, but the 6509
probably will have an incorrect local reachability table. Subsequent test
frames for the other terminals will then add to this confusion.
This archive was generated by hypermail 2.1.4 : Sat Apr 05 2003 - 08:51:45 GMT-3