From: Albert Lu (albert_ccie@xxxxxxxxx)
Date: Tue Nov 20 2001 - 03:41:16 GMT-3
Brian,
I'm not sure what you meant by the requirement to have an intra or inter
area to a non-zero forwarding address. Could you please elaborate, or
explain.
Thanks
Albert
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
Brian Hescock
Sent: Tuesday, November 20, 2001 2:34 AM
To: albert_ccie@yahoo.com
Cc: ccielab@groupstudy.com
Subject: Re: update: redistribution of directly connected networks
Albert,
Sounds correct. But I tend to avoid "redist connected" because it
basically gives you a noose to hang yourself with, particularly with ospf.
Read up on the ospf "fowarding address", as seen in "show ip ospf database
external", and the requirement to have an intra or inter area route to a
non-zero forwarding address.
B.
Albert Lu wrote:
> Brian,
>
> I know that this is an old thread, but I just wanted to share my
experiences
> on redistributing connected.
>
> The scenario I was working on was with RIP, IGRP, and connected
> redistributing into OSPF. From the config, you can see that the networks
of
> RIP and IGRP overlap each other, since they can only define classful
> networks.
>
> Redistributing one routing process at a time, starting with RIP and
tagging
> them. It can be seen that all directly connected networks that cover the
> 150.50.0.0 boundary that is not part of OSPF is redistributed.
>
> After that, IGRP is redistributed into OSPF, and it can be seen that IGRP
> has taken over the directly connected routes that RIP has redistributed.
> >From what Doyle says, it is because IGRP has a lower admin distance than
> RIP.
>
> Finally, redistributing connected with a tag, it can be seen that it
> redistributes "ALL" directly connected networks of the router that is not
> part of OSPF. I believe that it is because connected has the lowest admin
> distance out of all of them.
>
> I'm not sure whether this is correct behavior, my explanation sounds
> logical. I just wanted to make sure, and see what you can add from your
> findings.
>
> Thanks
>
> Albert
>
> Here is a section of my config:
>
> interface Loopback0
> ip address 200.0.0.2 255.255.255.255
> !
> interface Loopback1
> ip address 172.16.0.1 255.255.0.0
> !
> interface Ethernet0
> ip address 150.50.17.2 255.255.255.0
> !
> interface Serial0
> no ip address
> shutdown
> no fair-queue
> !
> interface Serial1
> no ip address
> encapsulation frame-relay
> !
> interface Serial1.1 point-to-point
> ip address 150.50.24.2 255.255.255.0
> frame-relay interface-dlci 104
> !
> interface Serial1.2 multipoint
> ip address 150.50.100.2 255.255.255.224
> ip ospf network point-to-multipoint
> frame-relay interface-dlci 105
> frame-relay interface-dlci 106
> !
> interface BRI0
> ip address 150.50.9.2 255.255.255.192
> shutdown
> isdn x25 static-tei 0
> !
> router ospf 10
> log-adjacency-changes
> redistribute connected metric-type 1 subnets tag 3
> redistribute rip metric 100 metric-type 1 subnets tag 1
> redistribute igrp 10 metric 10 metric-type 1 subnets tag 2
> network 150.50.100.0 0.0.0.31 area 0
> !
> router rip
> passive-interface default
> no passive-interface Ethernet0
> network 150.50.0.0
> !
> router igrp 10
> passive-interface default
> no passive-interface Serial1.1
> network 150.50.0.0
> !
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
> Brian Hescock
> Sent: Wednesday, October 24, 2001 6:55 AM
> To: ccielab@groupstudy.com
> Subject: update: redistribution of directly connected networks
>
> I've spoken with several Cisco Development Engineers and here's the
> correct behavior regarding whether we redistribute directly connected
> networks:
>
> The correct behavior is to redistribute directly connected networks as
> long as the network is covered by a network statement in the protocol to
> be redistributed. It's a bug if the directly connected network isn't
> redistribued in that situation. So the statement "we only redistribute
> from the routing table" is incomplete, we redistribute routes from the
> routing table and those directly connected networks covered by a network
> statement in the protocol to be redistributed.
>
> Example:
>
> int e 0
> ip add 10.1.1.1 255.255.255.0
> int e 1
> ip add 10.2.2.1 255.255.255.0
>
> router rip
> network 10.0.0.0
>
> router ospf 1
> network 10.2.2.1 0.0.0.0 area 0
> redistribute rip metric 100 subnets
>
> 10.1.1.0 /24 is known as a directly connected route. But it's included
> under the 10.0.0.0 network statement under router rip so the network is
> flagged by rip and ospf will then redistribute the directly connected
> network. You do not need to use "redistributed connected" unless you
> don't have a network statement that doesn't cover the network on the
> interface. So if ethernet 2 was 9.1.1.1, it wouldn't be redistributed
> unless you use "redistribute connected" since 9.1.1.1 doesn't fall under
> the 10.0.0.0 network statement. But it would be redistributed if you
> had "network 9.0.0.0" under router rip also.
>
> So the answer is any behaviour you see in earlier code where it doesn't
> redistribute in this situaiton is a bug and where it does redistribute
> in 12.1 and 12.2 is correct. This wasn't a change in behavior, just
> bugs in earlier code where you saw the problem.
>
> Please let me know if I've made a typo that changes the meaning and I'll
> send out a correction, thanks.
>
> Brian
This archive was generated by hypermail 2.1.4 : Fri Jun 21 2002 - 06:45:18 GMT-3