Hello, Anton! I run OSPF to get proper IGP. As classical iBGP is working on loopbacks with update-source. So we need to know the path. The way you want point me is to remove OSPF and use next-hop-self, right?
Regards, Boris пн, 27 февр. 2017 г. в 23:31, Anton Danilov <[email protected]>: > Hi, Boris. > Why do you use both protocol (OSPF and BGP) between the core router and > the border router? > I think you have at least two ways to solve your issue. > First way is pretty simple - don't use OSPF between the core router and > the border. It will make your setup is pretty simple: the core router > advertise only aggregated prefix (from static protocol) to the border, then > this prefix is being advertised with eBGP into upstream domain; the border > advertise all prefixes from upstream to the core router; the core router > advertise default router through itself into the OSPF domain. Obviously, > you don't need redistribute the iBGP into OSPF and vice versa. > Second way: you can separate the routes from OSPF and iBGP with different > routing tables (RIB) inside bird, and advertise the aggregated route into > upstream domain with eBGP. Also, you can use the route leaking with the > pipe protocol between RIBs. > > 2017-02-27 15:07 GMT+03:00 Борис Коваленко <[email protected]>: > > Ok, Martin. Let speak in general. The topology is very simple: > There is set of "area" routers speaking with "core" router by ospf. Core > router has "supernet" routed to null (ip route X.X.X.X/20 null0) to avoid > forwading loops to unallocated IPs. And it also injects this route into > ospf and bgp with "redistribute static". When route is injected into bgp it > gets some communities for further processing. Core router is speaking to > border router by ospf and bgp. So on border router we get two routes > X.X.X.X/20 ospf E2 type, and X.X.X.X/20 ibgp. OSPF wins, and route can not > be announced to ebgp peers. What do You suggest? Change preference for OSPF > - so we break igp? > > This is my knowlegde from cisco/quagga world. And it does not work with > bird. Where is my mistake? > > Regards, > Boris > > > пн, 27 февр. 2017 г. в 16:50, Martin Mares <[email protected]>: > > Hello! > > > I'm newbie to bird. Used cisco/quagga before. But filter language of bird > > is very nice, so I want to try it. But I have one big misunderstanding. > > With other vendors each protocol has it own routing table. So OSPF may > work > > only with ospf prefixes, BGP with bgp and so on. If we need protocol to > get > > access to other routing tables there are redistribute XXX commands. > > > > Unfortunatelly in bird there is one "super" table by default. So i get > > sutiation where I have to prefixes on router, one from static protocol, > and > > one from ibgp. Prefix from ibgp has some communities on it, and I use > this > > communities in filters to ebgp. But static prefix always win. By some > > reason I can't remove static prefix and use ibgp prefix and also can't > add > > communities to static prefix as they are changed by other router. > > Generally speaking, what you export to other routers should be a subset of > what you really use for forwarding packets. Otherwise you are inviting > routing > loops and other problems. (There are exceptions to this rule, for example > when > you are running a BGP route reflector, but I suspect it is not your case.) > > From this point of view, it does make much sense to me what you are trying > to accomplish. If you use the static route for forwarding, you should > export > it via eBGP. If the static route is merely a backup for cases when iBGP is > down, adjust its preference so that the iBGP route will be preferred. > > Have a nice fortnight > -- > Martin `MJ' Mares <[email protected]> http://mj.ucw.cz/ > Faculty of Math and Physics, Charles University, Prague, Czech Rep., Earth > "It is easier to port a shell than a shell script." -- Larry Wall > > -- > > С уважением, > Борис Коваленко > > > > > -- > Anton. > -- С уважением, Борис Коваленко
