----- Original Message ----- From: "Henning Brauer" <[EMAIL PROTECTED]> To: <[email protected]> Sent: Wednesday, August 10, 2005 6:54 PM Subject: Re: bgpd and two CARPed routers
> well, your description is not clear. Sorry, let me try and expand. We have two providers, each giving us two feeds/sessions - these four sessions and their respective /30 subnets terminate on our external IX switches, which both of the routers have one interface connected to. In short meaning the following (ignoring the basic internal facing stuff); Provider 1 neighbours - 1.1.1.1 & 1.1.1.5 Provider 2 neighbours - 2.2.2.1 & 2.2.2.5 /etc/hostname.fxp0 inet 3.3.3.1 255.255.255.248 NONE /etc/hostname.carp0 inet 1.1.1.2 255.255.255.248 1.1.1.3 carpdev fxp0 vhid 1 pass provider1a /etc/hostname.carp1 inet 1.1.1.6 255.255.255.248 1.1.1.7 carpdev fxp0 vhid 2 pass provider1b /etc/hostname.carp2 inet 2.2.2.2 255.255.255.248 2.2.2.3 carpdev fxp0 vhid 3 pass provider2a /etc/hostname.carp3 inet 2.2.2.5 255.255.255.248 2.2.2.7 carpdev fxp0 vhid 4 pass provider2b iBGP always remains on fxp0/3.3.3.x, both machines. eBGP sessions are brought up over all four CARP devices of the active router and remain in IDLE state on the backup router. Up to three sessions could flap as they wish, without causing interuption of service. Then should the master router itself fail, the CARP groups would move to the backup and the eBGP sessions get re-initiated accordingly? I hope that's a bit easier to understand. I can't find many real-world examples of implementing OpenBGP, so would simply appreciate a sanity check that it all sounds correct. Thanks,
