BGP Flapping while OSPF stable

[m-oyegunle]

Hi,

I work in a service provider environment and a customer requested for an Enterprise MPLS network for a customer.

The customer is going to 2 service providers terminating on their network.

the solution is something like this

Customer CPE -CE -P/PE - P/PE -CE -Customer CPE

the protocol between Cusomter CPE - CE is EIGRP, CE-PE is BGP and PE-PE is BGP for customer traffic and IS-IS for the core.

Once the EIGRP is redistributed into the BGP, The BGP begins to flap unless the redistributed prefixes are filtered to only large subnets e.g. /23, /22

Now the second provider has thesame setup with the difference of PE-CE protocol to be OSPF, PE-PE to be OSPF also.

The second providers routes were tagged and those routes were filtered out, however the flaps were still observed.

also noticed was that the routes redistributed from the eigrp into bgp on the CE were advertised via BGP to the PE and those routes were advertised back by the PE to the CE.

We tried changing the PE-CE protocol to OSPF and the link is currently stable

[Nagendra Kumar Nainar]

Hi,

Once the EIGRP is redistributed into the BGP, The BGP begins to flap unless the redistributed prefixes are filtered to only large subnets e.g. /23, /22

- What error are you seeing with BGP flap?.

The second providers routes were tagged and those routes were filtered out, however the flaps were still observed.

- The second provider appears to be running OSPF as PE-CE protocol. Do you mean, the OSPF is flapping?. Or did you meant, the OSPF prefix learnt from CPE where filtered and not redistributed into BGP?.

[m-oyegunle]

There are no errors from the BGP logs;

It only shows the neighbor down, this happens after every 180seconds

Yes, the OSPF prefixes learnt from the second provider were tagged and not redistributed into the BGP

[Nagendra Kumar Nainar]

Hi,

It only shows the neighbor down, this happens after every 180seconds

<Nagendra> Ok. Are you seeing the BGP flapping every 180 seconds due to Hold time expired?. It appears to be MTU/MSS issue.

To confirm that, can you try disabling PMTUD for this neighbor using "neighbor <> transport path-mtu-discovery disable"

-Nagendra

[m-oyegunle]

This has already being disabled.

[Vinit Jain]

How many routes are being received from the CE side. Also, can you perform the below ping test from PE to PE

PE to PE:

ping <remote_peering_ip> sou <local_peering_ip> size <mtu> df-bit rep 100

Also,

show ip bgp nei <nei_ip>

Please share the output in a file.

Thanks

Vinit

[Nagendra Kumar Nainar]

Do  you have it disabled on both side?. Can you share teh configuration and logs from both the neighbors?.

-Nagendra