[c-nsp] IBGP route stays stuck

[Pasquino Andrea]

I should set a lab as I cannot debug easily on prod devices.

Anyway PE1 after does advertise the route to EBGP peer even after the route is back again in the external AS, here’s an excerpt of the advertised routes

     Network          Next Hop            Metric LocPrf Weight Path
*>i 7.7.7.7/32       X.X.X.2                0     33      0 ?

The EBGP peer, which I do not control, for some reason discards the update that it receives from PE1. As a matter of fact the loopback 7.7.7.7 is reachable correctly across the external AS

Regarding PE1, I think it discards the update coming from the external AS: I have no clue, afaik it should prefer EBGP information coming with LP=222).

Andrea

From: Pete Lumbis [mailto:alumbis at gmail.com]
Sent: giovedì 23 ottobre 2014 19:14
To: Pasquino Andrea
Cc: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] IBGP route stays stuck

Maybe the PE1 eBGP peer is also preferring the route via PE2, so the announcement to PE1 contains it's own AS Number?

On Thu, Oct 23, 2014 at 11:47 AM, Pasquino Andrea <andrea.pasquino at eng.it<mailto:andrea.pasquino at eng.it>> wrote:
Pete,

here you are steady state: dummy route 7.7.7.7/32<http://7.7.7.7/32> comes from EBGP

PE1:
     Network          Next Hop            Metric LocPrf Weight Path
 *>  7.7.7.7/32<http://7.7.7.7/32>       X.X.X.93                    222      0   XXX9 XXX7 65000 ?
PE2:
   Network            Next Hop            Metric LocPrf Weight Path
 *>i 7.7.7.7/32<http://7.7.7.7/32>       X.X.X.8                0    222      0   XXX9 XXX7 65000 ?


ip route vrf xxxx 7.7.7.7 255.255.255.255 Dialer319 172.19.9.2 240
!
router bgp XXXX
!
address-family ipv4 vrf xxxx
  redistribute static route-map TER
!
route-map TER permit 10
 set local-preference 33
 set weight 0

+++++++++++++++++++++++++++++++++++++++++++++

Now I stop announcing the route from EBGP, the floating static kicks in:

PE1:
PSMDCI11ME1#sh ip bgp vpnv4 vrf sumitomo
*>i 7.7.7.7/32<http://7.7.7.7/32>       X.X.X.2                0     33      0 ?
PE2:
   Network          Next Hop            Metric LocPrf Weight Path
*> 7.7.7.7/32<http://7.7.7.7/32>       X.X.X.2               0     33      0 ?

+++++++++++++++++++++++++++++++++++++++++++++

When I restart announcing the route from EBGP nothing changes...
I have to delete the static route on PE2 and putting it back again.

Andrea


From: Pete Lumbis [mailto:alumbis at gmail.com<mailto:alumbis at gmail.com>]
Sent: giovedì 23 ottobre 2014 17:15
To: Pasquino Andrea
Cc: cisco-nsp at puck.nether.net<mailto:cisco-nsp at puck.nether.net>
Subject: Re: [c-nsp] IBGP route stays stuck

Do you happen to have the output of the BGP tables (specifically the a.b.c.d route) from PE1 and PE2 during the failure? My guess is that it's "expected" (in the sense BGP is doing what it's supposed to, even if undesired) and seeing it in the broken state should make it easy to reverse engineer why it happens.

-Pete

On Thu, Oct 23, 2014 at 9:59 AM, Pasquino Andrea <andrea.pasquino at eng.it<mailto:andrea.pasquino at eng.it>> wrote:
Pete,

Actually the floating static on PE2 is redistributed  with weight 0 by default.
I have configured on PE2 a route-map in order to redistribute into BGP the routes coming from MPBGP with a higher weight, no luck. I think I’m going to check with TAC…

Andrea

From: Pete Lumbis [mailto:alumbis at gmail.com<mailto:alumbis at gmail.com>]
Sent: mercoledì 22 ottobre 2014 15:44
To: Pasquino Andrea
Cc: cisco-nsp at puck.nether.net<mailto:cisco-nsp at puck.nether.net>
Subject: Re: [c-nsp] IBGP route stays stuck

Perhaps when the floating static kicks in on PE2 the weight will be 32768, most likely making it the best route in the BGP table, causing it to advertise to PE1? I've seen similar configurations run into this. If this is your issue the solution is a route-map on the redistribution statement setting the weight to 0

On Wed, Oct 22, 2014 at 2:30 AM, Pasquino Andrea <andrea.pasquino at eng.it<mailto:andrea.pasquino at eng.it><mailto:andrea.pasquino at eng.it<mailto:andrea.pasquino at eng.it>>> wrote:
Hello,

This is the scenario:

I have PE1 and PE2 in AS100
PE1 has an EBGP peering with AS200 and receives a certain prefix a.b.c.d with LP=222
The same prefix a.b.c.d is redistributed on PE2 with LP=33 from a floating static route, and serves as a back-up route

Now here's what happens: if for some reason AS200 stops announcing prefix a.b.c.d, then the announcement from PE2 kicks and we are happy (back-up runs fine)

When  AS200 starts again announcing prefix a.b.c.d, then the announcement from PE2 does not get flushed from PE1 and PE2. We have to erase the "ip route vrf blahblah" command and rewrite it

Have you any clues ?

Andrea

_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net<mailto:cisco-nsp at puck.nether.net><mailto:cisco-nsp at puck.nether.net<mailto:cisco-nsp at puck.nether.net>>
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net<mailto:cisco-nsp at puck.nether.net>
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/