EIGRP in OSPF Redistribution

[DenyAnyAny]

Hello everybody,

assume a router running EIGRP and OSPF. EIGRP is redistributed into OSPF Process, fine. Router learns two equal cost routes for a subnet via EIGRP, but redistributes only one into OSPF. Any idea why?

sh ip rou 10.x.155.0

Routing entry for 10.x.155.0/24

  Known via "eigrp 100", distance 170, metric 13059072, type external

  Redistributing via eigrp 100, ospf 100

  Advertised by ospf 100 subnets route-map eigrp_to_ospf

  Last update from 10.y.31.30 on Vlan1, 10:03:18 ago

  Routing Descriptor Blocks:

  * 10.y.31.31, from 10.y.31.31, 10:03:18 ago, via Vlan1

      Route metric is 13059072, traffic share count is 1

      Total delay is 500120 microseconds, minimum bandwidth is 10000 Kbit

      Reliability 255/255, minimum MTU 1476 bytes

      Loading 1/255, Hops 3

    10.y.31.30, from 10.y.31.30, 10:03:18 ago, via Vlan1

      Route metric is 13059072, traffic share count is 1

      Total delay is 500120 microseconds, minimum bandwidth is 10000 Kbit

      Reliability 255/255, minimum MTU 1476 bytes

      Loading 1/255, Hops 3

sh ip os da ex 10.x.155.0

            OSPF Router with ID (10.z.245.74) (Process ID 100)

                Type-5 AS External Link States

  LS age: 11

  Options: (No TOS-capability, DC)

  LS Type: AS External Link

  Link State ID: 10.x.155.0 (External Network Number )

  Advertising Router: 10.z.245.74

  LS Seq Number: 80000016

  Checksum: 0xD7BE

  Length: 36

  Network Mask: /24

        Metric Type: 1 (Comparable directly to link state metric)

        TOS: 0

        Metric: 100

        Forward Address: 10.y.31.31

        External Route Tag: 188

[DenyAnyAny]

Hi Varma,

this is not the only thing that makes me wonder in this topology. I noticed that only one router is generating an LSA type 5 for a specific EIGRP route, i.e. either R3 or R4, although redistribution is identically configured on both routers. Example: R3 redistributes a route for 10.x.155.0/24 into OSPF with R1 as forward address and generates for that an LSA type 5, but R4 does not. Consequence: R3 keeps its two equal cost EIGRP routes for 10.x.155.0/24 since it does not receive any LSA for it from R4, but R4 (and all routers in OSPF like R5) installs the external OSPF route received from R3. Any idea why R4 does not generate an LSA type 5 for that subnet?

@Kishore: not so easy to change the AD.

Regards

[Vaibhava Varma]

Hi Mail

In my opinion this is what is happening and this is again due to feedback routing and the pretty not so good design:

R4 has two equal cost EIGRP External routes for any destination X . R4 redistributes the route into OSPF Process and since OSPF has a better AD than EIGRP External Route the OSPF External Route wins and the EIGRP route is removed from RIB but the moment this happens the redistribution which happened in first place is not valid anymore and the redistribution stops and the OSPF route is removed and suppose same time with exactly same process happening on R3 it installed the OSPF External route as well as propogated to R4. Now depending upon the RIB convergence the R3' OSPF External Route was received at the moment when EIGRP routes were back in RIB and now OSPF External route from R3 wins coz of AD and momentarily the destination X is placed via R3' OSPF.

But again in sometime same process will repeat on R3 as happened on R4 as explained above and the OSPF External Route will be removed from RIB of R3 as the EIGRP routes being redistributed no more exist and consequently from R4 and this leads to big feedback Routing Loop.

This is my understanding whats happening and really this is not a good thing to do 🙂

Regards

Varma

[DenyAnyAny]

Hi Varma,

now I see my error in reasoning. I was wondering why R4 was not redistributing the EIGRP routes into OSPF. But actually the EIGRP external routes disappears from the routing table as soon as they are overwritten by OSPF. The fact that they still exist under EIGRP database was misleading me. Thanks for the hint.

Regards

[Kishore Chennupati]

Hi Varma.

..... and the OSPF route is removed 

I beg to differ here. Allow me to explain

In his topology , he is redistributing from EIGRP into OSPF on both R3 and R4.

1. Lets take R3 first. It has 2 ECMP eigrp routes and redistributes them into OSPF. Now, why its not installing those 2 routes was the original question. Anyway, lets say its installed one route as E1 since he is using that metric. It will advertise this out to R4 and R5 and they update their databases and RIB's

2. Now coming to R4. It does the same redistribution from EIGRP to OSPF. But if it has received the E1 route before it could redistribute EIGRP into OSPF then it will install that route and redistribution doesnt take place anymore.

The above process can happen in either direction Either from R3-R4 or R4-R3. At any stage the ospf route will not be removed from the RIB.

Regards,

Kishore

[Vaibhava Varma]

HI Kishore

Yes I totally agree it can happen in both directions but what I have analysed in the T0 ie the time of start events for all this is not 0 and have thought of one such hypothetical moment when all this can happen. This does not mean necessarily it will always happen but there's a fair probability of this happening when T0 is not equal to  0.

Hope this clarifies my viewpoint.

Regards

Varma

[Kishore Chennupati]

Hi Maik,

Did you get a chance to fix the issue?

Regards, Kishore

[DenyAnyAny]

Hi Kishore,

Actually I am just trying to analyse the situation before taking an action plan, because in reality the topology is even worse. I think the easiest way would be to configure R3 and R4 to deny OSPF routes originating from EIGRP, so that both external EIGRP routes remains in the RIB.

Regards