I have a WAN with two hub sites and a dozen spokes with one frame pvc to one hub and a backup pvc to the second hub site. Each hub site has a link to a data processing provider which each remote site communicates to through a telnet session. The provider is using PAT on each of their two routers that connect to our hub sites.
I am using EIGRP with summarized default routes to the branch routers. I have set the bandwidth on all of the serial interfaces to the actual CIR so that the primary link is prefered over the 0 CIR link which I assigned a bandwith of 56k. This gives me a successor to the appropriate hub site.
The issue I am having is that although I have a much better FD through the primary circuit to hub site #1, some packets now an then actually exit out of the second serial interface which will then break the telnet session since it eventually opens a second PAT translation on the provider router #2. During this time the topology table and routing table remain unchanged.
The topology table shows as follows:
P 0.0.0.0/0, 1 successors, FD is 5537536
via 222.192.82.2 (5537536/281600), Serial0.1
via 222.192.83.2 (284982016/281600), Serial0.2
P 172.20.1.0/24, 1 successors, FD is 281600
via Connected, Ethernet0
P 222.192.82.144/28, 1 successors, FD is 281600
via Connected, Ethernet0
and the routing table:
222.192.82.0/28 is subnetted, 1 subnets
C 222.192.82.144 is directly connected, Ethernet0
172.20.0.0/24 is subnetted, 1 subnets
C 172.20.1.0 is directly connected, Ethernet0
D* 0.0.0.0/0 [90/5537536] via 222.192.82.2, 00:00:28, Serial0.1
I guess what I am trying to accomplish is to have the branch routers only send traffic to the host through the primary pvc only unless the primary pvc goes down. I know that EIGRP is supposed to send traffic out of two succesor interfaces with equal costs. I need to know whether this is normal for unequal cost paths. Does it send the traffic based on the ratio of the two path costs?
By the way, these branch routers are running 11.2.
Javier Oropesa
