From: Theodore TZEVELEKIS (theodore_tzevelekis@xxxxxxxxx)
Date: Thu May 17 2001 - 05:39:28 GMT-3
If the R1-R2 link is down, no info can be exchanged between non-backbone
areas necause the virtual link is down. It will also never come back up
again while the link is down, becase there's no neibhoship between R1 and
R2. The v-l will not run over the ISDN:
1. Because it has to be directly between R1 and R2
2. Because As long as R2 does not have a way out of area2, it will never be
able to reach R1 to establish the v-l.
about the bgp config, well.. ahem.. actually I removed that config :-) .
However if and when I find some time, I will reconfigure it and send it. The
general idea, however, was that R3 would generate, through bGP a default
route to R6 with a very high metric, thus creating a dynamic floating route
on R6. When R6 lost its ospf learnt route (The default from R3 is an IBGP
route thus with a very high Administrative Distance thus gets overriden by
OSPF) it takes up the BGP one that points to R3.
It would look similar to this (don't take my word for it though)
R3:
nei 6.6.6.6 default-info route-map gen-def
route-map gen-def
set metric 5000
set metric-type 1
access-list permit 0.0.0.0 0.0.0.0
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
Roman Rodichev
Sent: Thursday, May 17, 2001 3:18 AM
To: ccielab@groupstudy.com
Subject: RE: Question: OSPF, DDR, backup routes
you got a problem in your first paragraph. You can exchange routes between
non-backbone areas. Area1 will never send routes to Area2, they will first
be sent to area 0
and by they way! Now that I think about it, even if you run ospf demand
circuit between R1 and R3, ISDN will always stay up. That's because you will
have a virtual link between Area1 and Area0 traversing it.
Can you post your configs? I want to look at your BGP configs... just
curious about your "dirty solution"
>From: "Theodore TZEVELEKIS" <theodore_tzevelekis@yahoo.com>
>To: "Roman Rodichev" <rodic000@hotmail.com>, <ccielab@groupstudy.com>
>Subject: RE: Question: OSPF, DDR, backup routes
>Date: Thu, 17 May 2001 02:59:58 +0200
>
>I agree. With the virtual-link down, you have no intra-area route exchange.
>However, R2 also belongs to area 2 so it can still exchange routes with R3.
>One can imagine that we may be running some other protocol beyond R2-e3...
>that's another story. The issue here is how will a default route get
>installed in R2's table.
>
>ospf demand-circuit is probably THE solution. I mean I don't see any way it
>shouldn't work. I've done setups with OSPF-DDR before and they work fien
>(except when combined to multiple dialer profies between two same peers
>used
>as backups... but that's another story ;-) But! I don't want to run ospf
>over the ISDN. That would mean extending area0 or area 2 over the ISDN.
>Which does not fall within the specs that I have defined for myself ;-)
>
>I just want to try to find a solution with this data. So far I haven't been
>able to. And I'm beginning to lose hope :-(
>
>The only workaround I found was to use my BGP to inject the route (yes I'm
>running BGP over the whole thing as well) but
>1. This is not a proper solution (I want to use my OSPF network to do it).
>2. Last but not least: As soon as R2 gets a route from BGP, it will
>regerate
>a default route back into R3, thus overriding the floating default pointing
>to the ISDN and eclosing it into a loop (R2->R3 and R3->R2).
>
>In this scenario, the only solution I found is to generate the default
>route
>on R1, then just advertise it to R3 through R2 (thus not generate it
>locally
>on R2). This way, when the link goes down, R2 loses its own default, gets
>one from the BGP pointing to R3 and does not regerate one.
>
>But as I said, this is a "dirty" solution. I don't like it.
>
>Any comments?
>
>-----Original Message-----
>From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
>Roman Rodichev
>Sent: Thursday, May 17, 2001 2:36 AM
>To: ccielab@groupstudy.com
>Subject: Re: Question: OSPF, DDR, backup routes
>
>
>Very nice! I like it. But you can't do it.
>
>As soon as R1-R2 connections is gone, Area 1 and Area 2 need some way of
>connecting to area 0. You can only do it with a virtual link. But virtual
>link will never run over the interface that is not running OSPF :)
>
>try enabling ospf demand-circuit on the bri and tell us how it works out
>
>
> >From: "Theodore TZEVELEKIS" <theodore_tzevelekis@yahoo.com>
> >Reply-To: "Theodore TZEVELEKIS" <theodore_tzevelekis@yahoo.com>
> >To: "ccie-groupstudy" <ccielab@groupstudy.com>
> >Subject: Question: OSPF, DDR, backup routes
> >Date: Thu, 17 May 2001 01:43:37 +0200
> >
> >Hi all,
> >
> >I've been trying to find a way to solve a very specific problem...
>wihtout
> >any luck:
> >
> >You have 3 routers running OSPF:
> >R1 (e0-area0, e1-area1, bri1 -> R3, virtual-link to R2)
> >R2 (e0-area1, e1-area2(nssa), e3-area1, virtual-link to R1)
> >R3 (e0-area2(nssa), BRI1 -> R1)
> >
> >OSPF is NOT running on ISDN.
> >I inject a default route from R2 into area 3 therefore to R3, pointing to
> >itself (R2) - default-information yada yada.
> >
> >Now what I want, is with just one default on R3, and statics on R1
> >(floating
> >in both occasions) to be able to access R2-e3 through the ISDN in the
>event
> >of (R1-e1)<->(R2-e0) failure.
> >
> >What this means is that I have to somehow inject a default route into R2
> >when the ISDN comes up (Reminder: in normal circumstabnce, R2 is
>injecting
> >a
> >default route into R3). Furthermore, OSPF does NOT redistribute static
> >default routes. The only way I have found so far is to actually generate
>a
> >default-info yada yada on R3 when the link comes up. BUT! as soon as R2
> >receives this, it, in turn, generates a new default-info and injects it
> >back
> >into R3.
> >
> >Also, as the virtual link is down, I have to find a way to inject that
> >route
> >into the OSPF neighbors in area 1 hanging out of R2-e3.
> >
> >Any ideas?
> >
> >Thanks
> >
> >Theo
> >
> >
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:30:43 GMT-3