I'm not opposed to making things easier/safer on the operator but my gut says IDR is going to look at this and say that it is up to the operator to configure bestpath knobs consistently in their AS and that a protocol change is not needed. This is based somewhat on the pushback we received when trying to add something as simple as a capability that allows BGP to tell his peers what his local hostname is. We were able to acquire an open capability code # for hostname exchange but it was not easy (I really thought it would be going in).
I could be completely wrong about how IDR will react to the idea of listing bestpath order as a capability though. I think it would be good to discuss on IDR before committing any code to quagga. My 2c. Daniel On Mon, Dec 14, 2015 at 6:28 AM, Martin Winter <mwinter@opensourcerouting .org> wrote: > On 11 Dec 2015, at 8:25, Donald Sharp wrote: > > I agree with Paul's reasoning here, but I also agree with Daniel's point of >> getting an actual Capability #. We recently went through the same thing >> with the bgp hostname patch. >> > > Personally, I have a hard time to see good reasons why someone wants (or > should) > change the order. But people sometimes have crazy (good or bad) ideas on > abusing BGP. > > The main issue here is the lack of standard. I’m curious what others think. > I don’t think things needs to be approved (as RFC) before added to Quagga, > but > at least been discussed once at IETF would be a good thing to reduce the > risk > of having to change it again within months with something incompatible. > Personally, I would prefer to have this at least once (or better twice) > discussed at the IETF IDR working group. > > But overall, I think the feature is at least interesting to discuss. > > - Martin > > > > On Fri, Dec 11, 2015 at 11:18 AM, Paul Jakma <[email protected]> wrote: >> >> On Fri, 11 Dec 2015, Daniel Walton wrote: >>> >>> They didn't have any issues though. >>> >>>> >>>> >>> Sure, but as per the previous thread on this, mismatches on that could >>> indeed cause instability. >>> >>> I guess I still don't have my head wrapped around the exact problem that >>> >>>> is solved by exchanging the bestpath rules used? >>>> >>>> >>> Making BGP safer to use, and making it harder for operators to >>> accidentally get their network into an "unsafe" state that might be prone >>> to weird instability problems (potentially hard to debug). >>> >>> If it is desirable to have a different selection order (and it seemed >>> Cumulus thought so? And I agree it'd be good to have cluster-list >>> earlier), >>> then we can make bgpd make life easier for operators by making it harder >>> to >>> deploy it in a likely broken config. >>> >>> Why make life harder for operators, when a little bit of code in bgpd can >>> make their life easier? :) >>> >>> That was the use-case basically. >>> >>> regards, >>> -- >>> Paul Jakma [email protected] @pjakma Key ID: 64A2FF6A >>> Fortune: >>> A banker is a fellow who lends you his umbrella when the sun is shining >>> and wants it back the minute it begins to rain. >>> -- Mark Twain >>> >>> >>> _______________________________________________ >>> Quagga-dev mailing list >>> [email protected] >>> https://lists.quagga.net/mailman/listinfo/quagga-dev >>> >>> _______________________________________________ >> Quagga-dev mailing list >> [email protected] >> https://lists.quagga.net/mailman/listinfo/quagga-dev >> >
_______________________________________________ Quagga-dev mailing list [email protected] https://lists.quagga.net/mailman/listinfo/quagga-dev
