On Tue, Oct 29, 2013 at 11:16:20AM +0000, Andy wrote: > Yea its 24.. Would even be happy to offer some champers.. > > I think this is more of a Maudite crowd.. Connoisseurs on here... > > As I understand it you would need to write a small daemon to do the BFD > state monitoring for the transmission and reception of the heartbeats > with various peers. The protocol is fairly simple so for an experienced > dev this should be easy. > > Then in OpenBGPD you would need to have a way of gracefully and > forcefully immediately shutting down the BGP neighbor that matches the > BFD peer. This could be achieved by simply having the BFD daemon call > 'bgpctl neighbor $bfdpeer down' > > It is not so important for OSPF as that already has fast convergence > time with fast hello's etc.. But for BGP this would make a world of > difference to remove the BGP routes immediately (in less than a second) > as soon as the BGP neighbor goes down/becomes unreachable (even if not a > direct link (multi-hop etc)). >
BFD should be in kernel and it should change the linkstate like the GRE keepalive protocol does. Everything else is pretty much madness and somewhat impossible to do. PS: I think a I have a tree somewhere hiding with some bits added but I never cared enough to move one. So no beer for me (even though I'm just getting free belgium beer). -- :wq Claudio > On 28/10/13 21:10, Dan Farrell wrote: > > I'm not sure how much a crate is, but if it's a case (24 bottles), > > then I'll throw in a case as well for this work. > > Blanche de Chambly, anyone? Or is this more a Maudite crowd? > > > > > > Sincerely, > > > > Dan Farrell > > > > > > On Mon, Oct 28, 2013 at 12:54 PM, Andy <a...@brandwatch.com > > <mailto:a...@brandwatch.com>> wrote: > > > > Hi all, > > > > Would any of the esteemed OpenBSD developers be interested in > > adding support for BFD (Bidirectional Forward Detection) to OpenBSD. > > > > The protocol itself seems pretty simple and provides a sub-second > > keep-alive mechanism to monitor links for routes. E.g. Upon BFD > > failure BGP or OSPF can be torn down etc thus allowing for > > sub-second re-convergence of i/eBGP! > > > > I can only offer a crate of beer to anyone who has the skills and > > is willing :) > > > > '+1's welcome from others who would be interested to show signs of > > support/interest.. > > > > Cheers, Andy.
