02-20-2009 09:15 AM - edited 03-04-2019 03:39 AM
Hi,
we have a bgp peering session with our isp on ipv4 and are learning our default gateway through it. In parallel we have also an ipv6 peering session with the same ISP which allows us to learn the ipv6 default gateway.
But through this second bgp session we are also learning IPv4 prefixes which is not what we want (roughly 3100 ipv4 entries).
I tryed to implement some prefix-list filters without real success.
Does anyone has some insights on how to achieve this (filtering ipv4 address from ipv6 bgp neighbor annoucments).
Here our config:
router bgp 64555
no bgp log-neighbor-changes
neighbor 2991:f1f1:52:8000::5 remote-as 65186
neighbor 2991:f1f1:52:8000::5 description ISP1-IPV6
neighbor 192.168.1.33 remote-as 786
neighbor 192.168.1.33 description ISP1
!
address-family ipv4
neighbor 2991:f1f1:52:8000::5 activate
neighbor 192.168.1.33 activate
neighbor 192.168.1.33 send-community
neighbor 192.168.1.33 next-hop-self
neighbor 192.168.1.33 soft-reconfiguration inbound
neighbor 192.168.1.33 prefix-list DEFAULT in
neighbor 192.168.1.33 prefix-list CUSTOMER out
neighbor 192.168.1.33 route-map Peer-ISP-In in
neighbor 192.168.1.33 route-map Peer-ISP-Out out
no auto-summary
no synchronization
network 1.2.3.0
exit-address-family
!
address-family ipv6
neighbor 2991:f1f1:52:8000::5 activate
neighbor 2991:f1f1:52:8000::5 send-community
neighbor 2991:f1f1:52:8000::5 next-hop-self
neighbor 2991:f1f1:52:8000::5 soft-reconfiguration inbound
neighbor 2991:f1f1:52:8000::5 prefix-list DEFAULT6 in
neighbor 2991:f1f1:52:8000::5 prefix-list CUSTOMER6 out
neighbor 2991:f1f1:52:8000::5 route-map Peer-ISP6 out
network 2991:f1f1:55::/48
exit-address-family
!
....
ipv6 prefix-list DEFAULT6 seq 5 permit ::/0
!
ipv6 prefix-list CUSTOMER6 seq 5 permit 2991:f1f1:55::/48 le 64
route-map Peer-Janet6-Out permit 10
set community 65186:2
!
route-map Peer-ISP6-In permit 10
set local-preference 90
!
Any idea is welcome and will gratefully appreciated.
seb.
02-20-2009 09:40 AM
Hello Sebastien,
I may be wrong but I would do the following
address-family ipv4
no neighbor 2991:f1f1:52:8000::5 activate
because now you have configured your BGP4+ or MP BGP to talk about ipv4 with the ipv6 neighbor.
there is a communication channel given by the two endpoints and then in the address-families you decide the arguments:
so if you want to receive only ipv6 routes from
neighbor 2991:f1f1:52:8000::5
you don't need to activate it in af ipv4 unicast
the neighbor will know you don't want to receive ipv4 routes during capabilities exchange at session setup
Hope to help
Giuseppe
Hope to help
Giuseppe
02-20-2009 03:34 PM
Giuseppe,
You are correct. The IPv6 neighbor need to be deactivated under ipv4 unicast AF. The issue is that ipv4 unicast AF is activated by default as soon as you configured a neighbor in global mode. To avoid that the next time, you should configured "no bgp default ipv4-unicast", which will force you to activate the ipv4 unicast AF as it is the case with any other AFs.
Regards
02-21-2009 01:26 AM
Of course!!!!
You are completely right this immediatly solves the problem. I was so focused on the IPv6 section that I completely overlooked that line.
Thanks a lot,
Seb
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide