Question about OSPF and not routing thru trunk ports.

[pugster73]

Hi,

I seem to have a question about trunking issues with ospf which discovered wrong routes for some reason.  So we have 3 sites, redundant vpls links between the 3 sites.  Site 1 and Site 2 located here in in the US and both sites have 2 switches trunked to each other, and each switch is connected to one of the vpls link to the VPLS cloud.  Site 3 is located abroad and has only 1 switch connected to both vpls providers.

So on VPLS cloud 1 setting, site1switcha ip is 172.20.1.254, site2switcha ip is 172.20.1.50, and site3switch ip is 172.20.1.30

On VPLS cloud 2 setting site2switchb ip is 172.20.2.254, site2switchb ip is 172.20.2.50, and site3switch is 172.20.2.30

When I do a show ip route, it is showing the wrong route and not routing to its trunk port, on site 1 and site 2, but instead try to route site 3 abroad!  So when I do a show ip route:

site1switcha:

172.20.1.254/32, ubest/mbest: 1/0, attached
*via 172.20.1.254, Eth3/43, [0/0], 2y22w, local
172.20.2.0/24, ubest/mbest: 1/0
*via 172.20.1.30, Eth3/43, [110/45], 16:59:39, ospf-1, intra

site1switchb

172.20.1.0/24, ubest/mbest: 2/0
*via 172.20.2.30, Eth3/43, [110/20], 17:07:37, ospf-1, intra
172.20.2.0/24, ubest/mbest: 1/0, attached
*via 172.20.2.254, Eth3/43, [0/0], 2y36w, direct

site2switcha

172.20.0.0/16 is variably subnetted, 3 subnets, 2 masks
C 172.20.1.0/24 is directly connected, GigabitEthernet1/0/48
L 172.20.1.50/32 is directly connected, GigabitEthernet1/0/48
O 172.20.2.0/24
[110/15] via 172.20.1.30, 17:09:52, GigabitEthernet1/0/48

site2switchb

O IA 172.16.1.0 [110/2] via 172.20.2.10, 18:33:42, GigabitEthernet1/0/48
172.20.0.0/16 is variably subnetted, 3 subnets, 2 masks
O 172.20.1.0/24
[110/11] via 172.20.2.30, 17:13:04, GigabitEthernet1/0/48
C 172.20.2.0/24 is directly connected, GigabitEthernet1/0/48
L 172.20.2.50/32 is directly connected, GigabitEthernet1/0/48

site3switch

172.20.0.0/16 is variably subnetted, 4 subnets, 2 masks
C 172.20.1.0/24 is directly connected, GigabitEthernet3/0/47
L 172.20.1.30/32 is directly connected, GigabitEthernet3/0/47
C 172.20.2.0/24 is directly connected, GigabitEthernet3/0/48
L 172.20.2.30/32 is directly connected, GigabitEthernet3/0/48

As you can see, the routing issue is with site 1 and site 2 when trying to route packets to the other network, it goes site 3 located abroad which has high latency as a result.  Not sure what I can do to fix this problem.  Thanks.

[Philip D'Ath]

On the interface that you would prefer not to use, add an "ip ospf cost xxx".  Keep increasing the cost until it does what you want.

Alternatively, put an "ip ospf cost" on every trunk.  Make your preferred trunk cost=100, next preferred cost=110, next preferred cost=120, etc.

[pugster73]

Here's the cost when I do a show run on that interface: (I know it is probably not configured correctly.)

site1switcha - No cost configured.

site1switchb - ip ospf cost 10

site2switcha - ip ospf cost 10

site2switchb - ip ospf cost 1

site3switch - VPLS 1 - ip ospf cost 10

VPLS 2 - ip ospf cost 5

I know it is a dumb question.  But I would prefer to use both vpls circuits at the same time.  Is this possible?  Thanks.

[Philip D'Ath]

Generally I would prefer one VPLS circuit and keep the other as a backup.  Then you get nice symmetric traffic flows.  But if you need it, configure interfaces with equal cost that you would like to load balance across.