[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference irocz::common_brouters

Title:Digital Brouters Conference
Notice:New common-code brouter family: RouteAbout, DECswitch 900
Moderator:MARVIN::HARTLL
Created:Mon Jul 17 1995
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:929
Total number of notes:3736

863.0. "Redistribution between OSPF and RIP" by NNTPD::"[email protected]" (Feynman Lo) Thu Apr 17 1997 07:45

Can I redistributing routing information between OSPF and RIP
using DECswitch 900ETs, like what Cisco does?

The customer network config (simplified):

                   
                         [cisco router network]
                            /           \
  running         [Cisco router A]     [Cisco router B]
  OSPF                 /                    \
          10.91  (Token ring A)       (Token ring B)   10.92
                    /                        \
                 [900ET]                  [900ET]
                   |                          |
      10.170     ----------------------------------  
                    Alphaservers and workstations (DUnix 3.2g)   
                     (not support OSPF)

10.170 is a newly installed ip network.
The customer requires resilience between path A and B.
As Digital Unix 3.2G does not support OSPF by default, the customer
is trying to add two DECswitch 900ETs between the OSPF network and
10.170. 

We need to confirm  whether we can config the 900ETs
to run OSPF on one side and RIP on the 10.170 side, and
that routing information can be re-distributed between
OSPF and RIP sides so that the AlphaServers could listen
for RIP broadcasts from 900ET and automatic failover between
path A and path B.
                     
If yes, what specials should be aware of?
If no, any suggestion in this circumstance?

Regard,
- feynman

                
                     





[Posted by WWW Notes gateway]
T.RTitleUserPersonal
Name		DateLines
863.1IROCZ::GUNNERThu Apr 17 1997 10:5826
Yes you can redistribute routes between OSPF and RIP and you can
run OSPF on the 10.9x side and RIP on the 10.170 side. You need to
enable redistribution by using the command:-

OSPF Config> set comparison

You probably want to set this to "1" which makes RIP routes have the
same preference as OSPF External Type 1 routes. With this setting
all OSPF External Type 1 routes will be distributed into RIP (and
vice versa, but in your case there will be no RIP routes in your
routing table).

Since the UNIX systems will not store multiple equal cost RIP routes,
at any time its likely that all systems on 10.170 will use the
same router to route off the LAN. One possible problem is with
ICMP Redirects - its possibe that during a loss of reachability
through one router, it may send a host a Redirect to the other router.
This is OK, but by default the Redirect may not be timed out ever by
the host so that if you subsequently got a reverse failure (the
other router loses reachability, but the first one is now OK) then the
redirected routes would not get replaced. I think its possible to
run the public domain gated program to get Redirect timeouts to
occur - or it may be that the hosts you have already support Redirect
timeout. I hope somebody else can add some more clarification here !

Chris
863.2MARVIN::HARTTony Hart, InterNetworking Prod. Eng. GroupThu Apr 17 1997 14:509
RIP will advertise any OSPF routes it finds in the table by default (unless its
the default route but you can change as well).

You need to make sure that your 10.170 net is advertised in the OSPF network,
either (as Chris says) by having OSPF on the  DECswitch import RIP routes (in
which case you need to configure it as an ASBR in the OSPF menu) or else by
running OSPF on the 10.170 interface as well.

Tony
863.3Thanks...will try on thisNNTPD::"[email protected]"Feynman LoSat Apr 19 1997 09:011
[Posted by WWW Notes gateway]
863.4fail to redistribute Inter-area routesNNTPD::"[email protected]"Feynman LoTue Apr 22 1997 13:5464
Hi,

I got quite some problems configuring the environment

Problem 1:

We encountered problem redistributing inter-area routes
to RIP by the 900ETs.

The current configuration is as follows:

- ospf enabled on token ring interface  (10.91.91.44)
- area=0.0.0.0, set comparison = 1, ASBR enabled to import Direct route
- disable sending and receiving all rip routes on Token ring interface
- sending all rip routes over Ethernet interface (10.170.170.46)

OSPF>dump  does contain the SPF and SPIA route entries.

But when I do a netstat -rn on the Digital Unix I only find
routes belonging to area 0.

Are there somethings I have missed?


Problem 2 :

In order to workaround this problem, I tried to configure
the 900ETs to claim themselves as default on the Ethernet
side. But I am confused about what is the correct way to
do so. I did the job by 
  IP-CONFIG> add route 0.0.0.0 0.0.0.0  10.170.170.46

to add a static(default) route to the routing table.
Then Digital Unix does report it learned the default route
(either of the two 900ETs). But then on the 900ETs, the
original default routes learned from the OSPF side is
replaced by this manually added one.

I think I have done it improperly. What is the proper steps
(I can't deduce it from manual) to make  the router to claim itself
as default router on RIP side?

Problem 3:
Few weeks ago when I configured a 900ET in switch only mode, I used
the available HST service to configure the router's IP
for management.

This time I configure it in routing mode and also configure
a separate IP address (same subnet as the Ethernet interface) for HST,
but then I fail   to connect to the HST addreess. I can only
telnet to the Ethernet interface.

Is HST only for switch-only mode?


Any suggestions?

( p.s. we will make use of the other available Token ring
  port on 900ET to connect it to the other Token ring, in
  order to maximize the redundancy)

- feynman

[Posted by WWW Notes gateway]
863.5MARVIN::CLEVELANDThu Apr 24 1997 07:4913
    
>Problem 2 :
>
>In order to workaround this problem, I tried to configure
>the 900ETs to claim themselves as default on the Ethernet
>side. But I am confused about what is the correct way to
>do so. I did the job by 
>  IP-CONFIG> add route 0.0.0.0 0.0.0.0  10.170.170.46
    
    No, that isn't the way to do it.  What you want to do is configure RIP
    to send default route.  In IP config, 'enable send default'.
    
    Tim
863.6MARVIN::HARTTony Hart, InterNetworking Prod. Eng. GroupThu Apr 24 1997 07:5554
>- ospf enabled on token ring interface  (10.91.91.44)
>- area=0.0.0.0, set comparison = 1, ASBR enabled to import Direct route
>- disable sending and receiving all rip routes on Token ring interface
>- sending all rip routes over Ethernet interface (10.170.170.46)
>
>OSPF>dump  does contain the SPF and SPIA route entries.
>
>But when I do a netstat -rn on the Digital Unix I only find
>routes belonging to area 0.

	The problem may be the cost of the OSPF routes, unfortunately if the
OSPF cost is greater than 15, RIP will announce the routes at cost infinity
(i.e. 16).  This is probably something that ought to be changed.

	


>Problem 2 :
>
>In order to workaround this problem, I tried to configure
>the 900ETs to claim themselves as default on the Ethernet
>side. But I am confused about what is the correct way to
>do so. I did the job by 
>  IP-CONFIG> add route 0.0.0.0 0.0.0.0  10.170.170.46
	
	The way to do that is to use the SET ORIGINATE ... command, your static
route won't do what you expect (the CLI should really have caught this and
flagged it as an error).

	Note however that if there is a default route aleady present in the
routing table then that route will be announced in which case you may well run
into Problem 1 again.

	I think you need to add a static default route with the Cisco as the
next hop and get RIP to announce it.  If you're running OSPF as an AS boundary
router then use the OSPF propagation controls to make sure it doesn't import
that default route (since you don't want to announce it into the OSPF network).

>Problem 3:
>Few weeks ago when I configured a 900ET in switch only mode, I used
>the available HST service to configure the router's IP
>for management.
>
>This time I configure it in routing mode and also configure
>a separate IP address (same subnet as the Ethernet interface) for HST,
>but then I fail   to connect to the HST addreess. I can only
>telnet to the Ethernet interface.
>
>Is HST only for switch-only mode?

	Yes, the router uses either the HST stack or the Routing stack but not
both.  But there's no reason why the HST address can't be the same as the
routing interface address (in fact thats what most people do).
863.7MARVIN::HARTTony Hart, InterNetworking Prod. Eng. GroupThu Apr 24 1997 07:5713
Notes clash...

    
>    No, that isn't the way to do it.  What you want to do is configure RIP
>    to send default route.  In IP config, 'enable send default'.
>    
>    Tim

	Tim's right you need to enable sending the default on the interface,
but you also need the SET ORIGINATE ... command if you want to generate a
default route even if there isn't one in the routing table.

Tony