OT: SR/TLB Revised - discoveries..! (longish)

From: Nigel Taylor (nigel_taylor@xxxxxxxxxxx)
Date: Sun Nov 25 2001 - 18:36:47 GMT-3


   
John,
     I got to looking at your SR/TLB scanerio and it got my attention
as to the config you had didn't work once you removed the command
"bridge X bitswap-layer3-addresses". So I test out a few things and
this is what I discovered. Basically, I got this working with the posted
config.
I used everything based on your posted config except I just used different
IPs.
 A few things to keep us aware of what's going on and the config that allows
this SR/TLB to work.

R1's to0 MAC - 0000.308c.6b20 / 0000.0C31.D604 (bit-swapped)
R1's config

interface TokenRing0
 ip address 192.168.2.1 255.255.255.0
 no ip directed-broadcast
 ring-speed 16
 multiring ip

-----------------

R2's to0 MAC - 0006.3a3a.0707 / 0060.5C5C.E0E0 (bit-swapped0
R2's config

no ip routing
source-bridge ring-group 200
source-bridge transparent 200 300 1 20

interface Serial1
 no ip address
 no ip route-cache
 clockrate 64000
 bridge-group 20
!
interface TokenRing0
 no ip address
 no ip route-cache
 ring-speed 16
 source-bridge 10 1 200

bridge 20 protocol ieee
bridge 20 bitswap-layer3-addresses

-------------------

R3's e0 MAC - 0010.7b35.d05d / 0008.DEAC.0BBA (bit-swapped)
R3's config

no ip routing
interface Ethernet0
 no ip address
 no ip directed-broadcast
 no ip route-cache
 bridge-group 1
!
interface Serial0
 no ip address
 no ip directed-broadcast
 no ip route-cache
 no ip mroute-cache
 no fair-queue
 bridge-group 1
!
bridge 1 protocol ieee

---------------------

R4's e0 MAC - 00e0.1e80.3499 / 0007.7801.2C99 (bit-swapped)
R4's config

interface Ethernet0
 ip address 192.168.2.50 255.255.255.0
 no ip directed-broadcast

---------------------

Additional test points
PC1 connected on the same e0 segment of R4
e0 MAC - 0000.21c9.d79f / 0000.8493.EBF9 (bit-swapped)

PC2 connected on the same token ring segment of R1
to MAC - 0000.83ac.70e7 / 0000.C135.0EE7 (bit-swapped)

Ok, let's take this from the top.. R1, this config is pretty basic
except for the use of the "multring ip" command which To enable
collection and use of RIF information.

R2, the interesting router.. After going through everything on CCO
and in all my books, white papers etc.. they all suggest that in SR/TLB
that each domain sees the other as another SRB/TB and vise versa. However,
I could not get this to work unless the "bridge X bitswap-layer3-addresses"
comamnd was added to R2.

R2 after clearing rif-cache, bridge, and source-bridge..
R2#sh bri

Total of 300 station blocks, 294 free
Codes: P - permanent, S - self

Bridge Group 20:

    Address Action Interface Age RX count TX count
0000.c165.100e forward RingGroup200 0 2 0
0000.c165.2015 forward RingGroup200 0 4 0
0000.21c9.d79f forward Serial1 0 24 0
00e0.1e80.3499 forward Serial1 0 1 0(R4)**
0000.0c31.d604 forward RingGroup200 0 2 0(R1)**
0000.c135.0ee7 forward RingGroup200 0 67 0

R2#sh source-bridge

Local Interfaces: receive transmit
             srn bn trn r p s n max hops cnt cnt drops
To0 10 1 200 * f 7 7 7 94 46 0

Global RSRB Parameters:
 TCP Queue Length maximum: 100

Ring Group 200:
  No TCP peername set, TCP transport disabled
   Maximum output TCP queue length, per peer: 100
  Rings:
   bn: 1 rn: 10 local ma: 4006.3a3a.0707 TokenRing0 fwd: 1
   bn: 1 rn: 300 locvrt ma: 4006.3a3a.0747 Bridge-group 20 fwd: 1

Explorers: ------- input ------- ------- output -------
         spanning all-rings total spanning all-rings total
To0 52 41 93 45 0 45

  Explorer fastswitching enabled
  Local switched: 59 flushed 0 max Bps 38400

         rings inputs bursts throttles output drops
           To0 52 0 0 0

R2#sh rif
Codes: * interface, - static, + remote

Dst HW Addr Src HW Addr How Idle(min) Vlan Routing Information Field
0006.3a3a.0707 N/A To0 * - -
0000.83a6.0870 N/A BG20 3 - 0890.00A1.0C81.12C0
0000.83a6.04a8 N/A BG20 0 - 0890.00A1.0C81.12C0
0000.83ac.70e7 N/A BG20 0 - 0890.00A1.0C81.12C0
0000.308c.6b20 N/A BG20 2 - 0890.00A1.0C81.12C0(R1)**
R2#

A ping from R1 to R4...

R1#p 192.168.2.50

Type escape sequence to abort.
Sending 5, 100-byte ICMP Echos to 192.168.2.50, timeout is 2 seconds:
.....
Success rate is 0 percent (0/5)

and the arp of R1 shows.. it does it that the ARP request did not find a
host/MAC
address mapping of R4.

R1#sh arp
Protocol Address Age (min) Hardware Addr Type Interface
Internet 192.168.2.50 0 Incomplete SNAP
Internet 192.168.2.1 - 0000.308c.6b20 SNAP TokenRing0
R1#

Now if we look at R2, the "sh bri ver" shows that the TB knows about R4
through serial1,
also the TX count = 0.

R2#sh bri ver

Total of 300 station blocks, 294 free
Codes: P - permanent, S - self

BG Hash Address Action Interface VC Age RX count TX
count
20 1E/0 0000.c165.100e forward RingGroup200 - 1 1
0
20 35/0 0000.c165.2015 forward RingGroup200 - 0 18
0
20 48/0 0000.21c9.d79f forward Serial1 - 0 477
6
20 AD/0 00e0.1e80.3499 forward Serial1 - 0 11
0(R4)**
20 D2/0 0000.0c31.d604 forward RingGroup200 - 0 12
0(R1)**
20 E9/0 0000.c135.0ee7 forward RingGroup200 - 0 187
7

Flood ports (BG 20) RX count TX count
Serial1 364 110
RingGroup200 110 364

Now in doing a "sh rif" on R2 we get this.. notice the SRB domain has no
knowledge of
R4 and does not have a cached entry for it's bit-swapped
MAC(0007.7801.2C99).

R2#sh rif
Codes: * interface, - static, + remote

Dst HW Addr Src HW Addr How Idle(min) Vlan Routing Information Field
0006.3a3a.0707 N/A To0 * - -
0000.83a6.0870 N/A BG20 2 - 0890.00A1.0C81.12C0
0000.83a6.04a8 N/A BG20 0 - 0890.00A1.0C81.12C0
0000.83ac.70e7 N/A BG20 0 - 0890.00A1.0C81.12C0
0000.308c.6b20 N/A BG20 0 - 0890.00A1.0C81.12C0(R1)**
R2#

now in using the "bridge 20 bitswap-layer3-addresses" command on R2, notice
the changes on
R1..

R1#sh arp
Protocol Address Age (min) Hardware Addr Type Interface
Internet 192.168.2.47 0 0000.8493.ebf9 SNAP TokenRing0(PC1)
Internet 192.168.2.50 4 0007.7801.2c99 SNAP TokenRing0(R4)
Internet 192.168.2.1 - 0000.308c.6b20 SNAP TokenRing0

This doesn't seem to be performing SR/TLB to me because of the use of the
bitswap command.
It would seem that the SR/TLB conversion isn't being performed. The use of
the "bitswap" command
allows the conversion to be made through the BG as in a TB bridged domain.
I've got to put my sniffer on this to get a better idea of what's going on
here. Also I would like kto test this out with a PC on the token segment(of
R1) instead .. This is interesting to say the least.

thoughts anyone...!

-Nigel



This archive was generated by hypermail 2.1.4 : Fri Jun 21 2002 - 06:45:22 GMT-3