Routing Table Eligibility

Question

I have an architecture which is that each remote site has 2 Dialer interfaces dialling into a virtual-template core which authenticates the sessions and applies attributes from a RADIUS server. OSPF runs on the dialer interfaces (on remote sites) and spawned Virtual-Access interfaces (from the virtual-template on the core router) and OSPF neighborship comes up just fine:

remote_router#show ip ospf neighbor

Neighbor ID     Pri   State           Dead Time   Address         Interface
10.200.0.6        0   FULL/  -        00:00:30    10.200.0.6      Dialer1
10.200.0.6        0   FULL/  -        00:00:24    10.200.0.6      Dialer2

however there is no ECMP operational:

    remote_router#show ip os database external 0.0.0.0

            OSPF Router with ID (10.200.0.32) (Process ID 2002)

                Type-5 AS External Link States

  LS age: 1521
  Options: (No TOS-capability, DC, Upward)
  LS Type: AS External Link
  Link State ID: 0.0.0.0 (External Network Number )
  Advertising Router: 10.200.0.6
  LS Seq Number: 800032A2
  Checksum: 0x525
  Length: 36
  Network Mask: /0
        Metric Type: 2 (Larger than any link state path)
        MTID: 0
        Metric: 1
        Forward Address: 0.0.0.0
        External Route Tag: 3489700465

you can see here that the metric for both interfaces is the same:

Dialer1 is up, line protocol is up (spoofing)
  Internet Address 10.200.0.53/32, Area 0.0.0.0, Attached via Interface Enable
  Process ID 2002, Router ID 10.200.0.32, Network Type POINT_TO_POINT, Cost: 1562
  Topology-MTID    Cost    Disabled    Shutdown      Topology Name
        0           1562      no          no            Base
  Enabled by interface config, including secondary ip addresses
  Transmit Delay is 1 sec, State POINT_TO_POINT
  Timer intervals configured, Hello 10, Dead 40, Wait 40, Retransmit 5
    oob-resync timeout 40
    Hello due in 00:00:07
  Supports Link-local Signaling (LLS)
  Cisco NSF helper support enabled
  IETF NSF helper support enabled
  Index 1/1/5, flood queue length 0
  Next 0x0(0)/0x0(0)/0x0(0)
  Last flood scan length is 1, maximum is 5
  Last flood scan time is 0 msec, maximum is 0 msec
  Neighbor Count is 1, Adjacent neighbor count is 1
    Adjacent with neighbor 10.200.0.6
  Suppress hello for 0 neighbor(s)
  Simple password authentication enabled
Dialer2 is up, line protocol is up (spoofing)
  Internet Address 10.200.0.57/32, Area 0.0.0.0, Attached via Interface Enable
  Process ID 2002, Router ID 10.200.0.32, Network Type POINT_TO_POINT, Cost: 1562
  Topology-MTID    Cost    Disabled    Shutdown      Topology Name
        0           1562      no          no            Base
  Enabled by interface config, including secondary ip addresses
  Transmit Delay is 1 sec, State POINT_TO_POINT
  Timer intervals configured, Hello 10, Dead 40, Wait 40, Retransmit 5
    oob-resync timeout 40
    Hello due in 00:00:07
  Supports Link-local Signaling (LLS)
  Cisco NSF helper support enabled
  IETF NSF helper support enabled
  Index 1/2/6, flood queue length 0
  Next 0x0(0)/0x0(0)/0x0(0)
  Last flood scan length is 1, maximum is 5
  Last flood scan time is 0 msec, maximum is 0 msec
  Neighbor Count is 1, Adjacent neighbor count is 1
    Adjacent with neighbor 10.200.0.6
  Suppress hello for 0 neighbor(s)
  Simple password authentication enabled

What I think is happening is that because the advertising router (10.200.0.6) is available over Dialer1 and Dialer2, OSPF is unable to ECMP over the two interfaces as it sees the route as the same and therefore will only install one route. I can't find any documentation or debugs to prove this is the case though, if someone could help me I'd appreciate it.

Kind Regards,

Gareth

Answer 1

If I understand your question correctly (sorry can't comment to ask questions until I get more rep), OSPF will actually install both routes and will load balance traffic across them because they are equal cost paths. If you wanted to test your theory about it being an equal cost thing, then you could add an interface cost to one of them.

Answer 2

In the end I created a new Loopback at the tunnel gateway and used that as the address for tunnel 2. I now have load balancing across the 2 tunnel gateway addresses. I can only assume that the router was considering the single IP address only once in the routing table.

G