Hi, What about simple acl on the non-mpls interfaces blocking bgp from loopback of ibgp src -> loopback of ibgp dest? Am I missing the boat completely?
I know you don't want acl's on any core intf's, but if you want funny solutions you might have to do funny stuff... Cheers ;-) -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Oliver Boehmer (oboehmer) Sent: Thursday, October 15, 2009 8:09 AM To: David Freedman Cc: [email protected] Subject: Re: [c-nsp] ibgp TTL > How about explicit path TE with no autoroute announce (and only statics > for these dedicated iBGP loopbacks?) well, if the only path to the destination is through the "non-MPLS part of the network", there will be no TE path available. so the tunnel will go down and the statics go away, and IGP path will be chosen. Well, you could add loating statics to Null0.. but this is certainly not "nice" and requires lot of manual work.. oli > >> More detail: > >> > >> I have a standard IP/MPLS backbone with MP-iBGP between PEs loopbacks > with > >> IS-IS L2 or OSPF area 0 as IGP. > >> > >> This IGP is extended to some non MPLS routers X. > >> > >> In some backbone links failure, IGP allow MP-iBGP to stay UP via X links > >> (non MPLS). > >> > >> This specific IGP design introduce a L3VPN blackhole that can be solved > by > >> IGP prefix filtering or by limiting TTL for MP-iBGP sessions, if possible > :) > > > > Hmm, you could also cause iBGP session to fail if you just add an > interface ACL not allowing iBGP between your PEs across the links not > running MPLS. > > > > Not sure if there is any real solution to this, other than increasing the > link metric towards the "non-MPLS-capble part" so much that MPLS packets > will not cross these links (or turn this part of the network into a stub > area to achieve the same). > > > > oli > > > > > > > >> On Wed, Oct 14, 2009 at 2:10 PM, Oliver Boehmer (oboehmer) > >> <[email protected]> wrote: > >> yes, only supported for ebgp. Would be interested about the "very > >> specific design" and why Manu requires this functionality? > >> > >> oli > >> > >>> AFAIK this command is for eBGP only, no? > >>> > >>> On Tue, Oct 13, 2009 at 10:07 PM, Matlock, Kenneth L > >>> <[email protected]>wrote: > >>> > >>>> Router bgp <AS> > >>>> Neighbor <AS> ttl-security hops <hops> > >>>> > >>>> ? > >>>> > >>>> Ken Matlock > >>>> Network Analyst > >>>> Exempla Healthcare > >>>> (303) 467-4671 > >>>> [email protected] > >>>> > >>>> > >>>> > >>>> -----Original Message----- > >>>> From: [email protected] > >>>> [mailto:[email protected]] On Behalf Of Manu Chao > >>>> Sent: Tuesday, October 13, 2009 4:52 AM > >>>> To: [email protected] > >>>> Subject: [c-nsp] ibgp TTL > >>>> > >>>> For a very specific design, i need to limit TTL value in > >> ibgp-multihop. > >>>> Is it possible to limit iBGP TTL like we do with ebgp-multihop ttl > >>>> command? > >>>> > >>>> Any input appreciated. > >>>> > >>>> Manu > >>>> _______________________________________________ > >>>> cisco-nsp mailing list [email protected] > >>>> https://puck.nether.net/mailman/listinfo/cisco-nsp > >>>> archive at http://puck.nether.net/pipermail/cisco-nsp/ > >>>> > >>> _______________________________________________ > >>> cisco-nsp mailing list [email protected] > >>> https://puck.nether.net/mailman/listinfo/cisco-nsp > >>> archive at http://puck.nether.net/pipermail/cisco-nsp/ > > > > _______________________________________________ > > cisco-nsp mailing list [email protected] > > https://puck.nether.net/mailman/listinfo/cisco-nsp > > archive at http://puck.nether.net/pipermail/cisco-nsp/ > > _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/ _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
