[quagga-users 3472] OSPF and VRRP : incompatible?

Wouter de Jong wouter at widexs.nl
Mon Dec 20 20:03:04 GMT 2004


Hello,

I have a set-up that involves 5 machines with Quagga, running ospfd, on
FreeBSD 5.2.1/5.3, of which 3 are LAN-gateways. The 2 other boxes are 
used as 'VPN-gateway' to connect the networks together:

vpn-rtr01
vpn-rtr02
rtr01.office1
rtr02.office1
rtr.office2

The rtr*.office1 and rtr.office2 have 2 OpenVPN (http://opvenvpn.sourceforge.net)
connections: 1 to vpn-rtr01 (cost 100), 1 to vpn-rtr02 (cost 110)
rtr*.office1 exports 192.168.192.0/24 and rtr.office2 exports 192.168.200.0/24

Foreach tunnel, I use a /30 from 192.168.252.0/xx

This part works flawlessly, if one vpn-rtr fails, the other one takes
over.

The .office1 network has 1 DSL-connection (at this time).
The network for my switches, routers, 'n stuff is 192.168.128.0/xx
rtr01 is master VRRP, rtr02 is backup VRRP. When a router is in
backup-state, it's default route is set to a VRRP VIP, 192.168.128.1
Both routers are VLAN-gateway's for the .192. network, with both 1 IP
from this network configured on the vlan-device (.2 and .3). There is also a VRRP
VIP, 192.1/32 (cause BSD interface aliases have to have netmask 255.255.255.255)

Problem is as follows:

When rtr01.office1 goes down, rtr02.office1 takes over all the
networking tasks perfectly. But, when using ospfd, the route of 192.168.192.0 to my 
vlan-device is deleted on rtr02, when it goes from
backup to master status. It DOESN'T happen when rtr01 is down, I reboot
rtr02, and when rtr02 comes up, the routes are saved. Also, when rtr01
comes back up later, and master-status changes from rtr02 to rtr01, the
route get's injected again on rtr02, so it'll work again. Pretty useless
as a backup router then, huh ? :)

I've read something about a VRRP <-> OSPF incompatibility somewhere,
that states that 'nothing' should change interface stuff once OSPF is
initialized on it.

I have one solution, and that is adding a route to the 192.168.192.0/24
network with a destination of my vlan-device.

You might think what's the fuzz about adding 1 extra route per subnet
once in a config file? Well, I don't understand _why_ ospfd gives me this 
behaviour, and I intend to use this setup for more networks than described 
above, so manually fixing this for every subnet is a lot of extra work
eventually.

Anyone who can point me into the good directions for getting this to
work in the way I want ? :)
I would appreciate it very, very much.

Many thanks in advance, and thanks for even reading this.

Kind regards,

-- 
WideXS                          http://www.widexs.nl
Wouter de Jong                  System-Administrator
Tel +31 (0)23 5698070           Fax +31 (0)23 5698099
Bijlmermeerstraat 62,           2131 HG  HOOFDDORP, NL


More information about the Quagga-users mailing list