How to weight redistributed BGP routes

Answered Question
Jan 5th, 2009

For the two routers(located in different location), they are learning the same BGP routes, but I'd like to redistribute BGP into the EIGRP so that router B redistributed BGP is always preferred over Router A redistributed BGP routes. Does this metric configuration look ok? should this work? Thanks.

//Router A

router eigrp 109

redistribute static

redistribute bgp 64806 metric 1024 1000 250 1 1500

no auto-summary

network 172.20.0.0

network 172.24.0.0

network 176.20.0.0

network 192.168.0.0

network 192.168.10.0

//Router B

router eigrp 109

redistribute static

redistribute bgp 64808 metric 2048 1000 255 1 1500

no auto-summary

network 172.20.0.0

network 172.24.0.0

network 176.20.0.0

network 192.168.0.0

network 192.168.1.0

I have this problem too.
0 votes
Correct Answer by lejoe.thomas about 7 years 11 months ago

hi Shawkat,

To achieve the objective it's better you keep the bandwidth same on both router and just modify delay. In the above configuration, the advertised distance from Router A will be much larger that it will never be considered as a feasible successor in EIGRP.

So keep that bandwidth same on both routers and just modify the delay by 1 or 2.

eg: 999 on Router B and 1000 on Router A

RTR B

redistribute bgp 64808 metric 2048 999 255 1 1500

RTR A

redistribute bgp 64808 metric 2048 1000 255 1 1500

HTH

Lejoe

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 4.5 (2 ratings)
Loading.
Correct Answer
lejoe.thomas Mon, 01/05/2009 - 23:44

hi Shawkat,

To achieve the objective it's better you keep the bandwidth same on both router and just modify delay. In the above configuration, the advertised distance from Router A will be much larger that it will never be considered as a feasible successor in EIGRP.

So keep that bandwidth same on both routers and just modify the delay by 1 or 2.

eg: 999 on Router B and 1000 on Router A

RTR B

redistribute bgp 64808 metric 2048 999 255 1 1500

RTR A

redistribute bgp 64808 metric 2048 1000 255 1 1500

HTH

Lejoe

shawkatalvi Mon, 02/23/2009 - 22:01

There is a problem again. When we lost the RTR B, all traffic switched through RTR A(exactly as we wanted), but when RTR B was restored, routes were still travelling via RTR A. I had to manually stop BGP routing on RTR A to get the reroutes working properly.

How can this be achieve automatically when it detects that default path is restored?

milan.kulik Tue, 02/24/2009 - 00:23

Hi,

after the router B restore, did you see the BGP prefixes redistributed to the EIGRP topology on router B?

Are the router A and B receiving the BGP prefixes from the same router in a different AS? Or from different neighbors? Is there any BGP session between routers A and B?

BR,

Milan

royalblues Tue, 02/24/2009 - 00:29

Are you by any chance redistibuting EIGRP back into BGP as well?

Narayan

shawkatalvi Tue, 02/24/2009 - 04:10

Yes I am redistributing EIGRP into BGP as well.

Both router has:

redistribute eigrp 109 metric 1

do you think I should change the metric?

shawkatalvi Tue, 02/24/2009 - 04:02

Hi,

//after the router B restore, did you see the BGP prefixes redistributed to the EIGRP topology on router B?

I did not see that. Any specific command?

//Are the router A and B receiving the BGP prefixes from the same router in a different AS? Or from different neighbors? Is there any BGP session between routers A and B?

Router A and B receiving the BGP prefixes from different routers with different AS (provider's router). But they advertise same routes. There is no BGP session between A and B. It's all EIGRP.

BR,

Milan

milan.kulik Tue, 02/24/2009 - 07:48

Hi,

command to check EIGRP:

show ip eigrp topology net mask

See http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094613.shtml#rdc

To understand completely:

Are there any other routers involved in EIGRP?

At the time of trouble:

Your internal subnets were not advertised via BGP from router B? (I.e., all traffis from the provider was still sent to A.)

Or the prefixes received from the provider were not redistributed to EIGRP on router B? (I.e., all traffic from your sites was sent to the providers via A.)

Or both?

Generally, mutual redistribution between two routing protocols is not recommended.

See http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a008009487e.shtml

e.g.

IMHO, you should make sure only correct subnets being redistributed, see

http://www.cisco.com/en/US/docs/ios/11_1/np1/configuration/guide/4ciprout.html#wp6060

as an example.

It's also possible that you are advertising the same (EIGRP learnt) subnets from both routers A and B (i.e., different originating AS from BGP point of view).

BR,

Milan

royalblues Tue, 02/24/2009 - 10:30

This is a case of mutual redistribution problem

So when you are redistributing EIGRP back to BGP you need to apply the filters as below

route-map eigrp2bgp permit 10

match ip address 1

set local-preference 90

set weight 0

access-list 1 permit .. which are suppsoed to be learnt from BGP

router A

router bgp 64808

redistribute eigrp 109 metric 1 route-map eigrp2bgp

HTH

Narayan

shawkatalvi Mon, 03/02/2009 - 13:52

Thanks Guys. Looks like problem is resolved now. Even though EIGRP and BGP are being injected into each other, I am using prefix list not to accept certain routes.

Even if I use route-map to filter out certain routes, How can I achieve redundancy? RTR A and B are supposed to be acting as complete redundancy if one fails. So limiting certain routes means during failover certain routes are missing. Any thoughts?

milan.kulik Mon, 03/02/2009 - 14:09

Hi,

just guessing without knowing details:

IMHO, while RTA A and B are exchanging routing info via EIGRP, they shoud use the same AS number for BGP.

It's possible you might not need prefix filtering anymore then, as BGP would refuse prefixes originated from the same AS automatically.

BR,

Milan

shawkatalvi Mon, 03/02/2009 - 14:24

Hi Milan.

EIGRP AS is the same but BGP AS is different.

So on two routers, Each BGP AS is different so they have prefix-list in place.

milan.kulik Tue, 03/03/2009 - 08:10

Hi,

yes, that's what I meant:

If you were using the same BGP AS number in both sites, the BGP process on router A would refuse the prefixes received via BGP from the provider but avertised by router B originally (and possibly redistributed from router B EIGRP) and vice versa.

That might prevent some routing loops in a case of router the other router fail/fix.

BR,

Milan

Actions

This Discussion