From: John Neiberger (neiby@xxxxxxxxxx)
Date: Mon Nov 26 2001 - 13:40:20 GMT-3
Thanks for that link, it explains the reasoning behind the behavior. I
wanted to note that in my configuration I was not using ebgp-multihop since
these were iBGP peers.
The rationale appears to be the same in both cases: a 0/0 route will not
suffice if that is the only route available to reach a BGP peer.
Regards,
John
On Mon, 26 Nov 2001 10:00:52 +0100, Rivron Francois wrote:
| FYI :
|
|
http://www.cisco.com/univercd/cc/td/doc/product/software/ios121/121cgcr/ip_r
| /iprprt2/1rdbgp.htm#xtocid142341
|
| To prevent the creation of loops through oscillating routes, the multihop
| will not be established
| if the only route to the multihop peer is the default route (0.0.0.0).
| From :
|
|
|
| > -----Message d'origine-----
| > De: John Neiberger [SMTP:neiby@excite.com]
| > Date: lundi 26 novembre 2001 05:33
| > @: Matt Smith; ccielab@groupstudy.com
| > Objet: Re: Interesting BGP problem
| >
| > I just set this up and was able to verify it in 12.1(11). Without at
| > least
| > a specific classfull route in the routing table the neighbors would
never
| > get past ACTIVE. Here is the output from debug ip bgp:
| >
| > 09:24:03: BGP: 172.16.1.1 multihop open delayed 10064ms (no route)
| > 09:24:13: BGP: 172.16.1.1 multihop open delayed 13920ms (no route)
| > 09:24:14: BGP: Import timer expired. Walking from 1 to 1
| > 09:24:27: BGP: 172.16.1.1 multihop open delayed 14560ms (no route)
| > 09:24:29: BGP: Import timer expired. Walking from 1 to 1
| > 09:24:35: BGP: compute bestpath
| > 09:24:41: BGP: 172.16.1.1 multihop open delayed 14720ms (no route)
| > 09:24:44: BGP: Import timer expired. Walking from 1 to 1
| > 09:24:56: BGP: 172.16.1.1 multihop open delayed 10144ms (no route)
| >
| > On this router I then added a static route to 172.16.0.0 and the
neighbors
| > came up. Very interesting! This definitely falls into the "Good to
Know"
| > category.
| >
| > Thanks,
| > John
| >
| > On Sun, 25 Nov 2001 20:02:24 -0500, Matt Smith wrote:
| >
| > | Hey all,
| > | I just got done doing some last minute BGP practice scenarios in
| > preperation
| > | for my upcoming date with fate on 11/30. At any rate I was making
some
| > quick
| > | and dirty lab scenarios to test some varios configurations and I
found
| > the
| > | following
| > |
| > | R1 <----->R2<------->R3
| > |
| > | R1 is configured with a 0.0.0.0 route to R2 and R3 with a 0.0.0.0
route
| > to R2
| > | as well. I pinged from R1 to R3 and communicationed worked fine.
| > |
| > | Now I issued router bgp 100 on both R1 and R2 and defined the neigh
bor
| > | statements for iBGP.
| > |
| > | The BGP relationship never came up. AS#s were correct on both
router
| > | processes and neighbor statements and the IP addresses were correct
as
| > well.
| > |
| > | show ip bgp neighbor reports that neither router has sent or
recieved
| > and
| > BGP
| > | messages and a deb ip packet verifies that this is true.
| > |
| > | What was the culprit? well........... Make a guess then read on
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > |
| > | I issued a static route for the specific address of the neighbor on
| > each
| > | router (R1 and R3) pointing to R2 and the relationship came up.
| > Apparently
| > | BGP requires a route in the routing table for the nieghbor address
and
| > will
| > | not use the 0.0.0.0 route. IP classless enabled on both routers and
as
| > I
| > | stated both routers could ping one another. Has anyone ever read
and
| > | documentation supporting this to be true? I have never seen any
rules
| > | pertaining to this issue so I thouht I would share it with the
group.
| > Just a
| > | quick tidbit. Hope it helps someone someday.
| > |
| > | Luck to All
| > |
| > | Matt Smith
This archive was generated by hypermail 2.1.4 : Fri Jun 21 2002 - 06:45:23 GMT-3