[c-nsp] bgp vpls rfc4761 - no failover at multihomed ce site
Aaron
aaron1 at gvtc.com
Tue Apr 29 18:21:55 EDT 2014
ugh! 15.3(3)S2 didn't help. and 15.4.2(S) didn't help either. ... what
am I doing wrong ? when I bring down 3600-2 g0/10, I can't reach devices
behind 3600-1 g0/10, and it appears that 9k still have pw and bgp nlri for
3600-2. So it's like the ac on 3600-2 (g0/10) going down does not propagate
into bgp to cause 9k to see 3600-2 nlri as unreachable and 9k still sees
3600-2 as best path. 9k does see both 3600's. ONLY when I remove the vfi
context completely from 3600-2 does the 9k failover to 3600-1 and traffic is
restored. See previous email in thread about bgp l2vpn vpls table on 9k.
Both me's pretty much look the same...
l2vpn vfi context v45
vpn id 45
autodiscovery bgp signaling bgp
ve id 2
interface Vlan45
no ip address
member vfi v45
bridge-domain 45
member GigabitEthernet0/10 service-instance 45
interface GigabitEthernet0/10
switchport trunk allowed vlan none
switchport mode trunk
load-interval 30
service instance 45 ethernet
encapsulation dot1q 45
rewrite ingress tag pop 1 symmetric
9k is...
l2vpn
bridge group v45
bridge-domain v45
interface GigabitEthernet0/0/0/3
!
vfi v45
vpn-id 45
autodiscovery bgp
rd 64512:45
route-target 64512:45
signaling-protocol bgp
ve-id 1
!
!
!
routed interface BVI45
-----Original Message-----
From: Jason Lixfeld [mailto:jason at lixfeld.ca]
Sent: Tuesday, April 29, 2014 12:30 PM
To: Aaron
Cc: <cisco-nsp at puck.nether.net>
Subject: Re: [c-nsp] bgp vpls rfc4761 - no failover at multihomed ce site
There are BGP signaled, BGP AD VPLS issues like that, that I found and
reported. They were resolved in 15.3(3). I'd get on to 15.3(3)S2 if you
can.
Sent from my iPhone
> On Apr 29, 2014, at 1:06 PM, "Aaron" <aaron1 at gvtc.com> wrote:
>
> Hi All, it doesn't seem that the AC's going down cause a bgp
reconvergence.
>
>
>
> my test architecture is now, a single homed ce site off a asr9k (4.1.2)
and
> a second site dual-homed ce off (2) me3600's (15.3(2)S2)....
>
>
>
> when i shutdown the (ac, attachment circuit) pe-ce port on the 3600-2 the
> traffic doesn't failover to 3600-1. the only way i can get it to failover
> is to completely remove the vfi context in 3600-2. then i see the 9k fire
> off a pw to 3600-1 and traffic starts to flow again to site 2.
>
>
>
>
> |----eng-lab-3600-1-----occam----- site2-pc2
>
>
> | |
>
>
> | |
>
>
> | |
>
> site1-pc---eng-lab-9k-1----mplscloud----|
> occam----- site2-pc1
>
>
> | |
>
>
> | |
>
>
> | |
>
>
> |----eng-lab-3600-2-----occam
>
>
>
> 3600-1 is 10.101.12.251
>
> 3600-2 is 10.101.12.250
>
>
>
>
>
>
>
> *** before i delete vfi context from 3600-2....
>
>
>
> RP/0/RSP0/CPU0:eng-lab-9k-1#sh bgp l2vpn vpl rd 64512:45
>
> ...
>
> Network Next Hop Rcvd Label Local Label
>
> Route Distinguisher: 64512:45 (default for vrf v45:v45)
>
> *> 1:1/32 0.0.0.0 nolabel 16315
>
> *>i2:1/32 10.101.12.250 18 nolabel
>
> * i 10.101.12.251 307 nolabel
>
>
>
>
>
> *** after i delete vfi context from 3600-2....
>
>
>
> RP/0/RSP0/CPU0:eng-lab-9k-1#sh bgp l2vpn vpl rd 64512:45
>
> Network Next Hop Rcvd Label Local Label
>
> Route Distinguisher: 64512:45 (default for vrf v45:v45)
>
> *> 1:1/32 0.0.0.0 nolabel 16315
>
> *>i2:1/32 10.101.12.251 307 nolabel
>
>
>
>
>
> Aaron
>
> _______________________________________________
> cisco-nsp mailing list cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
More information about the cisco-nsp
mailing list