From: Tony Hanks (jhconsulting2001@xxxxxxxxx)
Date: Mon Jan 07 2002 - 06:54:09 GMT-3
EA,
dude, did you ever get this resolved? I used R1 as the Route-reflector and
have tried redistributing BGP into OSPF first on R1, then on R6. I think
the routes BGP validated as best was like 2 to 3 routes when redistribution
took place on R1. It looked fine when it's configured on R6, and hence AS1
(R8) gets the routes that it needs from OSPF...
Also, what about the BGP route that originates from AS3 (R7)? I can't get
the 172.168.70.0/24 route in the R6's BGP table to show as validated.
Hence, AS1 (R8) can never get that route. In the answers, R8's routing
table doesn't include this route, while the question (task 2, item 19)
states that R7 & R8 should receive both the 172.168.70.0/24 &
172.168.80.0/24 routes. Can you tell me what you did and whether you got
the 172.168.70.0 route over to R8? Thanks in advance!
Tony Hanks
MCSE+I, CCNP
Network Infrastructure Engineer
J & H Consulting Inc.
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
EA Louie
Sent: Thursday, December 27, 2001 10:18 AM
To: ccielab@groupstudy.com
Subject: Re: BGP and IGP redistribution - ccbootcamp8
Here are the conditions:
route reflection where the router reflector is not synced, but the 2 clients
are synced
when IGP->BGP redist is peformed at the route reflector, the two clients get
the bgp routes, but any routes that are not directly connected show up as
not sync'ed, and therefore not advertised to eBGP peers.
when IBP->BGP redist is performed at one of the two clients, that client and
the route reflector propagate BGP routes.
when I change the BGP RID on the route reflector (which is not synced) to
match RID from an intermediate (non-BGP) OSPF router and redistribute from a
router that is synced, the problem disappears. :-) You can add that to
your list of fixes.
-e-
----- Original Message -----
From: "Stephen C. Feldberg" <scfeldberg@hotmail.com>
To: "EA Louie" <elouie@yahoo.com>; <ccielab@groupstudy.com>
Sent: Wednesday, December 26, 2001 6:52 AM
Subject: Re: BGP and IGP redistribution - ccbootcamp8
> Any route reflection configured in the transit AS? Rember that OSPF and
BGP
> have issues when sync is left enabled and the OSPF route needs to be
> propagated between route reflector clients. The "fix" is to move the
> redistribution point, or implement another IGP.
>
> Steve
> ----- Original Message -----
> From: "EA Louie" <elouie@yahoo.com>
> To: <ccielab@groupstudy.com>
> Sent: Tuesday, December 25, 2001 3:54 PM
> Subject: BGP and IGP redistribution - ccbootcamp8
>
>
> > Findings from ccbootcamp 8 that I wanted to share with you and get some
> > opinions about.
> >
> > The scenario has 3 BGP AS'es with one acting as a transit AS. The rules
> > called for synchronization on 3 of the BGP routers.
> >
> > I found that redistributing OSPF into BGP on a router that is not
> synchronized
> > did not allow the BGP routes to be sync'ed at a downstream iBGP
neighbor,
> > which prevented those routes from being propagated to the next eBGP
> neighbor.
> > The exact error message was found in deb ip bgp updates, which said "No
> valid
> > path for a.b.c.d", and also from 'show ip bgp a.b.c.d' which would
> indicate
> > the route was not sync'ed. Can anyone explain to me why, when the bgp
> routes
> > are in the ip routing table, I'd get these messages, which essentially
> prevent
> > an iBGP to advertise routes to a eBGP neighbor? It looks like it has
> > something to do with bgp next-hop addresses being the same in the IP RT
> and
> > the BGP table, but I can't be sure, and wouldn't know how to make them
the
> > same even if I were sure.
> >
> > Now, if there's another way to solve this problem, I'd love to hear
about
> it,
> > but all I did to solve it was move the OSPF --> BGP redistribution point
> to R6
> > (for those familiar with this lab). I had originally done this at R1 (a
> route
> > reflector), where 'no sync' was allowed.
> >
> > I also found that redistributing the eBGP routes into OSPF (or other
IGP)
> at
> > the iBGP entry points is a good practice, and only requires a small
> > access-list and route-map to accomplish.
> >
> > -e-
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:56:18 GMT-3