I have R1 and R2 routers that have VRRP setup on ether1 on which BGP will occur with the ISP. The ISP have initialized a BGP1 session with its own /30 IP range and a secondary BGP2 session with its own separate /30 IP range. How can I setup two separate VRRP interfaces so that on R1 BGP1 is on VRRP1 interface in master mode and BGP2 is on VRRP2 interface in backup mode while on R2 BGP1 is on VRRP1 interface in backup mode and BGP2 is on VRRP2 interface in master mode? That way each router always has an active BGP route table of the internet.
You don’t need VRRP if you already have 2 BGP sessions with your upstream.
Just ditch VRRP on the public side (ether1), and do 1 BGP peering on each of your routers.
BGP will take care of failover, etc.
You can also use BGP MED to control which router will be the “primary” (ie: on which router you will receive incoming traffic).
Take a look on this MUM presentation https://mum.mikrotik.com/presentations/GR15/MUMGreece-Athens-2015-Nikalexis_Nikos.pdf
What about the delays? BGP has significant
On the one hand, using VRRP the way OP described, will not necessarily make failover any faster, if for example the physical link stays up but it doesn’t pass any traffic.
Even if you failover on your end, the other end will still have to timeout before using the routes from the other BGP peer.
On the other hand, you can use BFD (which is supposedly the proper method for fast failure detection).
Beware though, BFD on ROS does not play well (or at all) with other BGP vendors (or even with ROS).
Another option (which is described in the MUM presentation) would be to use VRRP on the “lan” side and use scripting on master/backup states to change the BGP MED so that the upstream will quickly use the routes your the second router even if the BGP peering with the first router hasn’t timed out yet.
Tank you for the feedback. I am probably going to try a few different setups to see what works the best. Unfortunately I am having issues with MTU with VRRP with CHR where you cant set the MTU on the VRRP interface higher than 1500 or on any sub interfaces such as VLANS. Support was able to reproduce the problem so until they have a fix I’ll be putting this project on the side. I look forward to trying the suggestions once they fix the VRRP MTU issue.
Instead of combining VRRP and BGP which normally isn’t recommended, consider using this design with BFD which is simlilar to what you’re asking for
What would be the configuration on mikrotik for it?
What I don’t see with that is gateway redundancy for the subscriber base.. unless that aggregation cloud is where you’ve moved VRRP to.