I have successfully set up OSPF with dual OpenVPN tunnels in a Multi-WAN client/server setup.
Failover works well, I can down the active WAN, OSPF kicks in and routes OpenVPN traffic over the secondary WAN. I can happily flip between the two WAN's all day long with no problems at all.
My problem is with load balancing. Both of my OpenVPN connections are assigned to interfaces which are then assigned in OSPFd. I have given both the interfaces a metric of 10. When I start two simultaneous pings (via two seperate PC's) from Site B to Site A, traffic graphs only show one WAN being utilised.
Here's what my OSPF route table looks like:
Destination Nexthop Path Type Type Cost Uptime
0.0.0.1 10.0.9.5 Intra-Area Router 10 00:05:36
0.0.0.1 10.0.8.5 Intra-Area Router 10 00:05:29
10.0.8.6/32 10.0.9.5 Intra-Area Network 20 00:05:19
10.0.8.6/32 10.0.8.5 Intra-Area Network 20 00:05:19
10.0.9.6/32 10.0.9.5 Intra-Area Network 20 00:05:36
10.0.9.6/32 10.0.8.5 Intra-Area Network 20 00:05:29
192.168.2.0/24 10.0.9.5 Type 1 ext Network 110 00:05:36
192.168.2.0/24 10.0.8.5 Type 1 ext Network 110 00:05:29
192.168.2.1/32 10.0.9.5 Type 1 ext Network 110 00:05:36
192.168.2.1/32 10.0.8.5 Type 1 ext Network 110 00:05:29
The destination subnet i'm trying to reach from the client is 192.168.2.0/24 and you can see there are two possible routes to this (10.0.8.5 and 10.0.9.5), these correspond two the two ovpnc tunnels which i've setup on the client.
Any ideas of where this is going wrong?