Monday, September 23, 2019

question about OSPF and VRRP

Hey guys,

I'm using FRRouting 7.1 on Debian with kernel 5.2.2 and have a question about using vrrp with ospf.

I have 2 routers (R1 and R2) which connected to isp (ISP1) with BGP and internal to a third-party router (R3).

BPG and the external connections works well. Both Routers (R1 and R2) connected through OSPF on the internal interfaces (eno2) and a configured loopback (lo). This will work also well without VRRP. Check the interface config below.

Router 1 Interface Status VRF Addresses --------- ------ --- --------- eno1 up default xxx.xxx.xxx.xxx/29 eno2 up default 198.18.0.2/28 eno3 up default 172.16.198.21/24 eno4 down default lo up default 10.0.0.1/32 Router 2 Interface Status VRF Addresses --------- ------ --- --------- eno1 up default 172.16.198.22/24 eno2 down default enp4s0f0 up default xxx.xxx.xxx.xxx/29 enp4s0f1 up default 198.18.0.3/28 lo up default 10.0.0.2/32 

Router 3 has a configured interface with the IP 198.18.0.10/28 and is directly connected through a switch to router 1 and 2. Router 3 should not be a part of the OSPF area.

Then I wanted to configure VRRP on R1 and R2, that R3 can reach one of both (the master) router through HA.

In reason of this, I installed keepalived and configured vrrp. This works also from the vrrp perspective.
But the OSPF will fails if VRRP is running. After shutddown, OSPF comes back.

VRRP will add the virtual IP as additional IP on en2 Interface with 198.18.0.1/32 for R1.

In the log I got the notification, that the NetworkMask mismatch on enp4s0f1:198.18.0.3 (configured prefix length is 28, but hello packet indicates 32)
.

And in the OSPF data package, I have seen that the VRRP IP is redistribute as a new Stub Network with 198.18.0.1/32

I have already tried to deny the redistribution or block through a access-list filter, but does not helped.
The vrrp IP is not for interessting for my OSPF config, only for the backend router R3.

Any thoughts how I can fix this in my OSPF config?

Thanks Henry



No comments:

Post a Comment