From: Brian McGahan (brian@cyscoexpert.com)
Date: Fri Oct 25 2002 - 16:46:32 GMT-3
Joe,
From the way I'm understanding this scenario, this is the
default behavior of BGP. Check the BGP table of Rtr_A. I'm willing to
bet that for all except those 20600 prefixes, Rrt_A is routing
everything through Rtr_B.
A BGP router only advertises its best routes. What is happening
is that Rtr_A and Rtr_B both receive full tables from their providers,
and then advertise full tables to each other. After convergence, Rtr_A
realizes that Rtr_B has a better route to the majority of prefixes.
These are marked as best routes in the BGP table, and the matching
prefixes learned from ISP_A are withdrawn from Rtr_A's advertisement to
Rtr_B.
Is this what is happening?
See this previous thread for more detail:
http://www.groupstudy.com/archives/ccielab/200202/msg00366.html
HTH
Brian McGahan, CCIE #8593
Director of Design and Implementation
brian@cyscoexpert.com
CyscoExpert Corporation
Internetwork Consulting & Training
Voice: 847.674.3392
Fax: 847.674.2625
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf
Of
> Joe A
> Sent: Thursday, October 24, 2002 8:35 PM
> To: ccielab@groupstudy.com
> Subject: Peculiar BGP Behavior
>
> Can anyone think of a reason why a router would take in a full table
> from an iBGP peer and then reduce the table down to about 20600
routes?
> I have configued both iBP peers to accept full tables.
>
> The scenario is this:
>
> ISP_A (AS X) ISP_B (AS Y)
> | |
> | |
> eBGP eBGP
> | |
> | |
> RTR_A-------------iBGP (AS Z)-------------RTR_B
>
> RTR_A accepts a full table from ISP_A, and RTR_B accepts a full table
> from ISP_B. RTR_A and RTR_B then advertise the full table to each
> other.
>
> RTR_A shows that it received approximately 110,000 routes from both
> peers.
>
> RTR_B shows that it received the same number of routes from ISP_B, and
> it will BRIEFLY show that it received the same number form RTR_A
('show
> ip bgp summary'). If I clear the iBGP neighbors, I can see RTR_B
> receive the full table from RTR_A (through repeated 'show ip bgp
> summary', but as soon as RTR_B stops receiving, it begins decreasing
the
> number of routes from RTR_A until it gets down to about 20600 routes.
I
> can't debug easily because the debug command with a full table just
> crushes the router.
>
> If I 'show ip bgp <RTR_B> advertised' on RTR_A, RTR_A shows that it
sent
> the full table; if it were following up with withdrawal messages, I'd
> expect that RTR_A would then recount what it advertized, and 'show ip
> bgp <RTR_B> advertised' would not show that the full table was sent.
> Therefore I think I can rule out that RTR_B drops the routes due to
> withdrawal messages from RTR_A, so I can't figure out what's going on.
>
> I have also triple-checked my route filters, and that's not it.
>
> Any ideas?
>
> Joe
>
> [GroupStudy.com removed an attachment of type image/jpeg which had a
name
> of Glacier Bkgrd.jpg]
This archive was generated by hypermail 2.1.4 : Tue Nov 05 2002 - 08:35:57 GMT-3