[c-nsp] Possible split horizon issue with bgp signalled vpls

[Jason Lixfeld]

I dunno.  All I know is that after upgrading to S1a, I saw some odd L2 issues.  The one I was able to track down was the ME3600 was unable to resolve ARP for hosts on a VLAN behind a directly connected port-channel.  The other one that I didn't have time to track down was a host in a VFI was unable to reach hosts in the same VPLS instance on a remote VFI.

It's possible that I mis-read the email, but I thought I was told it was due to CSCuh05321.  Maybe what the email meant was that the fix for CSCuh05321 didn't work?  Maybe the fix for CSCuh05321 caused another issue?  All I know is both nodes where I tested this release had to be reverted.

On Nov 19, 2013, at 7:33 AM, Adam Vitkovsky <adam.vitkovsky at swan.sk> wrote:

> That's what I was about to ask as the CSCuh05321 is actually listed under
> 15.3(3)S caveats not under 15.3(3)S1a so I'd assume it is resolved in S1a
> already right? 
> 
> adam
> -----Original Message-----
> From: cisco-nsp [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Nick
> Ryce
> Sent: Tuesday, November 19, 2013 12:14 PM
> To: Jason Lixfeld
> Cc: <cisco-nsp at puck.nether.net>
> Subject: Re: [c-nsp] Possible split horizon issue with bgp signalled vpls
> 
> Hi Jason,
> 
> CSCuh05321 says it is fixed in the S1 release so would that not mean that it
> is also in S1a?
> 
> Nick
> 
> 
> On 19 Nov 2013, at 03:50, Jason Lixfeld
> <jason at lixfeld.ca<mailto:jason at lixfeld.ca>> wrote:
> 
> Just be mindful of CSCuh05321 if you are going to try S1a.  If you think you
> might hit that, I'd suggest moving to static vpls and skip that release
> until it's fixed.
> 
> Also, with S1a, look at CSCtl54835 and verify or else your isis adjacencies
> will break.
> 
> Sent from my iPhone
> 
> On Nov 18, 2013, at 10:08 PM, Nick Ryce
> <nick at fluency.net.uk<mailto:nick at fluency.net.uk>> wrote:
> 
> Just found 1 switch on 15.3(2)S so may be worth a punt and upgrade
> 
> Nick
> On 19 Nov 2013, at 02:02, Jason Lixfeld
> <jason at lixfeld.ca<mailto:jason at lixfeld.ca>> wrote:
> 
> Issues I was having with BGP signalled VPLS a couple of months ago in
> 15.3(2)S1 resulted in filing CSCui46390.  I'd otherwise suggest trying
> 15.3(3)S1a to see fix works, but that version seems to have introduced
> CSCuh05321, so I think that might end badly for you; it did for me :(
> 
> On Nov 18, 2013, at 3:05 PM, Nick Ryce
> <nick at fluency.net.uk<mailto:nick at fluency.net.uk>> wrote:
> 
> Hi,
> 
> I'm tearing my hair out with this one and can't figure out how to resolve
> it.
> 
> I have 3 switches that have a BGP signalled VPLS with customer routers
> hanging off the end of all 3 ( one switch has 2 cpe )
> 
> All have the RD 56595:4 and RT 56595:4
> 
> All pseudo wires are up between the switches with config snippets below:-
> 
> Switch 1
> 
> VFI name: FLVPLS004, state: up, type: multipoint, signaling: BGP VPN ID: 4,
> VE-ID: 3, VE-SIZE: 10
> RD: 56595:4, RT: 56595:4, 56595:4
> Bridge-Domain 903 attachment circuits:
>  Vlan903
> Neighbors connected via pseudowires:
> Interface          Peer Address    VE-ID  Local Label  Remote Label    S
> pseudowire100033   46.226.0.9      2      402          39              Y
> pseudowire100037   46.226.0.14     1      401          49              Y
> 
> This switch has 1 cpe with any vlan tags stripped and can ping all devices
> on the other switches
> 
> 
> Switch 2
> 
> VFI name: FLVPLS004, state: up, type: multipoint, signaling: BGP VPN ID: 4,
> VE-ID: 1, VE-SIZE: 10
> RD: 56595:4, RT: 56595:4, 56595:4
> Bridge-Domain 11 attachment circuits:
>  Vlan11
> Neighbors connected via pseudowires:
> Interface          Peer Address    VE-ID  Local Label  Remote Label    S
> pseudowire100015   46.226.0.12     3      49           401             Y
> pseudowire100018   46.226.0.9      2      48           37              Y
> 
> This switch has 2 cpe's with any vlan tags stripped.  They can ping each
> other and the device connected to switch 1 but cannot ping device on switch
> 3
> 
> Switch 3
> 
> VFI name: FLVPLS004, state: up, type: multipoint, signaling: BGP VPN ID: 4,
> VE-ID: 2, VE-SIZE: 10
> RD: 56595:4, RT: 56595:4, 56595:4
> Bridge-Domain 4 attachment circuits:
>  Vlan4
> Neighbors connected via pseudowires:
> Interface          Peer Address    VE-ID  Local Label  Remote Label    S
> pseudowire100028   46.226.0.12     3      39           402             Y
> pseudowire100027   46.226.0.14     1      37           28              Y
> 
> This switch has 1 cpe device connected with any vlan tags stripped and can
> only ping the device on switch 1
> 
> 
> Each switch can see all the correct mac addresses.
> 
> It sounds like split horizon but I assumed this was only to do with the
> local switch?
> 
> All devices are running 15.3(3)S
> 
> Any help much appreciated.
> 
> Nick
> 
> 
> 
> 
> _______________________________________________
> 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/
> 
> 
> 
> _______________________________________________
> 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/
>