[rbak-nsp] BGP - GracefulRestart & BFD - how does it work ?

[Yuri Shefer]

Hi Marcin,

The BFD is running on the line card. If you enable BFD for your eBGP
neighbor, the BGP will become a client for BFD, i.e. BGPd process will be
notified if the BFD session fails.

The config part is very simple:

context bfd-bgp
 router bfd
  interface to-as123
   minimum transmit-interval 100
   minimum receive-interval 100
   detection-multiplier 3

router bgp 123
  neighbor 1.1.1.2 external
   bfd


To verify that the BGPd is correctly registered for receiving BFD events,
you can check "show bfd session details" and under the "Client" you would
see bgp.

On Thu, Jun 13, 2013 at 5:49 AM, Marcin Kuczera <marcin at leon.pl> wrote:

> hello,
>
> I have a customer who has a customer that is very sensitive to BGP
> failovers.
> I mean - holdtime is a problem if there is a real fault between routers
> (not just bgpd restart).
>
> There is quite long timeout - restart 120s + 90s of holdtime..
>
> Ok, we don't have problems everyday - but recently before upgrade to
> 12.1.1.2 we did for a week.
> So I need to find some soft solution for future cases.
>
> Now, I'am wondering what will happen if we start using BFD between our and
> our customer routers.
> Is the BDF for BGP a separate process from bgpd ?
>
> I mean - I would like to keep graceful restart for bgpd crash cases, and
> BFD for linecard crashes case.
> Does anyone has an experience using these features together and could
> share its opinion ?
>
> (and drop a example of BFD config for a BGP peer ;)
>
> --
Best regards,
Yuri
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/redback-nsp/attachments/20130613/d1b30ba8/attachment.html>