From: John Neiberger (neiby@xxxxxxxxxx)
Date: Thu Mar 14 2002 - 12:29:08 GMT-3
As I see it, though, from the perspective of the token-ring-
only router, the ethernet domain is a different ring. So, from
it's perspective, there are two rings. It's my (limited)
understanding that without multiring ip, the TR-only router
would only speak IP to devices on the same ring.
Is that correct or not? Admittedly, SRB isn't my strong point.
John
---- On Thu, 14 Mar 2002, steven.j.nelson@bt.com
(steven.j.nelson@bt.com) wrote:
> All
>
> As far as I understand this, Multiring IP is not required in
this
> sceanrio.
> Multiring IP is used on a TR interface to enable the use and
the caching
> of
> RIF information when SRB to another ring or srb domain.
>
> If no rif is found in the packet that hits the ring with
Multiring IP
> configured then the router will act as an SRB end station and
send out a
> test frame (explorer) to locate the destination and then pass
the packet
> back to the source host with the RIF information so the host
can then
> send
> an XID to the destination.
>
> Not required when SRB/TLB between ring and ethernet.
>
> Just my two cents and probably wrong !!!
>
> Ta
>
> Steve
>
> -----Original Message-----
> From: RSiddappa@NECBNS.com [mailto:RSiddappa@NECBNS.com]
> Sent: 14 March 2002 06:42
> To: neiby@ureach.com; ccielab@groupstudy.com
> Subject: RE: SR/TLB for IP Discoveries (LONG)
>
>
> John,
>
> This is absolutely good and workign config. But I am stll not
sure do u
> need
> that multiring Ip on the token ring interface fo the router
doign
> Translational bridging.
>
> I do no think. I had it workign with out that few days back.
>
> R.
>
>
>
> -----Original Message-----
> From: John Neiberger [mailto:neiby@ureach.com]
> Sent: Thursday, March 14, 2002 12:17 AM
> To: ccielab@groupstudy.com
> Subject: SR/TLB for IP Discoveries (LONG)
>
>
> Since I was temporarily fed up with staring at ipsec configs
I
> thought I'd take a crack at the SR/TLB stuff going on today.
I
> wasn't able to make this work using the configs posted
today.
> In fact, I can't figure out why the posted configs were
working
> at all. :-)
>
> I printed out three different docs from CCO and cross-
> referenced that with examples from the archives to come up
with
> a working solution. Here is the scenario:
>
> R2 --- (eth) --- R4 --- (tr) --- R3
>
> Let's start with R4 since it's the most complex. The first
> step is to configure SRB and transparent bridging:
>
> no ip routing
> !
> source-bridge ring-group 100
> !
> int to0
> source-bridge 10 5 100
> !
> int e0
> bridge-group 1
> !
> bridge 1 protocol ieee
>
> Then, configure SR/TLB to bridge between the two interfaces:
>
> source-bridge transparent 100 200 5 1
>
> In that config, the first number is from the source-bridge
ring-
> group statement. The second number is the virtual ring
number
> that identifies the transparent bridging domain to the SRB
> domain. The third number is the bridge number taken from the
> SRB config on the token ring interface. Finally, the fourth
> number is the bridge-group number taken from the ethernet
> interface config.
>
> At this point things won't be working just yet. On R2, the
> ethernet-only router, we simply need to configure an IP
> address. Then, we need to add an IP address to R3 that's in
> the same subnet. Go ahead, try to ping...it won't work yet,
> and here's where it starts to get a little tricky if you're
> doing this with IP.
>
> On R4 you've configured SR/TLB which handles layer-two
traffic
> just fine but it doesn't know how to handle embedded layer
> three addresses. To fix this, add the following:
>
> bridge 1 bitswap-layer3-addresses
>
> That's fairly straightforward, but things aren't going to
work
> yet. If you try to ping R2 from R3, the initial ARP will
> fail. Why? Because the router needs to send an explorer and
> it's not configured to do so. So, go to R3 and add the
> following:
>
> int to0
> source-bridge 10 4 200 (the 4 and 200 are irrelevant)
> source-bridge spanning
> multiring ip <------ important!
>
> Pings still won't work until you go back to the SR/TLB router
> and add 'source-bridge spanning' and 'multiring ip'. At that
> point, pings should succeed.
>
> With this solution there are no guarantees that all IP
traffic
> will work. In fact, this is configuration is probably a
really
> bad idea if you're running IP, but it at least sort of
works.
>
> I broke this down into sections to show the steps I went
> through to figure this out and make it work. Here are the
> final working configs:
>
> R2 (ethernet):
>
> int ethernet0
> ip address 172.16.43.2 255.255.255.0
>
> R3 (token ring):
>
> int tokenring0
> ip add 172.16.43.3 255.255.255.0
> source-bridge 10 4 200
> source-bridge spanning
> multiring ip
>
> R4 (SR/TLB):
>
> no ip routing
> !
> source-bridge ring-group 100
> source-bridge transparent 100 200 5 1
> !
> int ethernet 0
> bridge-group 1
> !
> int tokenring0
> source-bridge 10 50 100
> source-bridge spanning
> multiring ip
> !
> bridge 1 protocol ieee
> bridge 1 bitswap-layer3-addresses
>
> I currently have these exact configs on my routers, and here
is
> the result:
>
> R3#
> R3#sho arp
> Protocol Address Age (min) Hardware Addr Type
> Interface
> Internet 172.16.43.3 - 0000.30e2.cc0c SNAP
> TokenRing0
> R3#ping 172.16.43.2
>
> Type escape sequence to abort.
> Sending 5, 100-byte ICMP Echos to 172.16.43.2, timeout is 2
> seconds:
> .!!!!
> Success rate is 80 percent (4/5), round-trip min/avg/max =
> 8/8/8 ms
> R3#sho arp
> Protocol Address Age (min) Hardware Addr Type
> Interface
> Internet 172.16.43.3 - 0000.30e2.cc0c SNAP
> TokenRing0
> Internet 172.16.43.2 0 0000.30b1.4bc1 SNAP
> TokenRing0
> R3#
>
> I sure hope I don't have to mess with this in the real lab,
but
> this little adventure was very enlightening. At least I
think
> I could get it working if I run into it when it counts. :-)
>
> Regards,
> John
>
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:57:08 GMT-3