bgp rib failures, eigrp routes installed into bgp

[DARYLE DIANIS]

Hi again, there is a lot about bgp to know.  I have 2 routers, I looked at both show ip bgp and show ip route on both routers. In one router all the routes were installed by bgp.  In the second router, almost all the routes were installed by eigrp and there are many rib-failures on this one.  

This is the bgp section for each, they look similar (I didn't code this, still learning), yet they behave differently.  Both have a eigrp redistribution.  What would cause the difference in the routing table?

thanks in advance,,,,

This router installs bgp routes:

router bgp 65002
 bgp log-neighbor-changes
 network 10.210.100.0 mask 255.255.255.252
 redistribute static
 redistribute eigrp 1 route-map EIGRP2BGP
 neighbor 10.210.100.1 remote-as 1
 neighbor 10.210.100.1 default-originate
 neighbor 10.210.100.1 soft-reconfiguration inbound
 neighbor 10.210.100.1 route-map LOCALPREF in
 neighbor 192.168.255.4 remote-as 65002
 neighbor 192.168.255.4 update-source Loopback0

This router installs eigrp routes:

router bgp 65002
 no synchronization
 bgp log-neighbor-changes
 network 10.210.100.8 mask 255.255.255.252
 redistribute eigrp 1 route-map EIGRP2BGP
 neighbor 10.210.100.9 remote-as 1
 neighbor 10.210.100.9 soft-reconfiguration inbound
 neighbor 10.210.100.9 route-map AS-PREPEND out
 neighbor 192.168.255.148 remote-as 65002
 neighbor 192.168.255.148 update-source Loopback0
 no auto-summary
 

[milan.kulik]

Hi,

it's difficult to guess without more details.

Are you talking about the same prefixes received by EIGRP from the LAN on both routers?

Could you paste some sh route, sh ip bgp ..., sh ip eigrp topo ... outputs for an example prefix?

I can imagine following scenario, e.g., when a redistribution is involved:

Let's say both routers are receiving the prefixes from the LAN originally via EIGRP.

Both put them to their RIB and redistribute into BGP.

Both are receiving the prefixes from the other router via iBGP (through the LAN) and also from the ISP router via WAN.

But the prefix redistributed by the router itself is the best from BGP point of view (redistributed with Weight=32768 attribute by default).

Then the LAN interface on router A goes Down for some time.

Router A  removes the EIGRP prefix from the RIB and also withdraws the redistributed prefix from his BGP table.

The prefix received by iBGP is also withdrawn as iBGP session went down.

So the only remaining routing info for the subnet is the prefix received via eBGP which is installed into the RIB now.

Then the LAN port goes Up again.

Router A is receiving the EIGRP prefixes again.

Will he put them to the RIB though?

No! Because it has got the same prefix within BGP table and as it was received via eBGP, the Administrative Distance is 20, much better then EIGRP AD=90.

So that could be the situation you see.

In fact, after a while the iBGP session should establish again and the prefix received by iBGP should beat the prefix received by eBGP (shorter AS_PATH). And EIGRP prefix should beat the iBGP then due to better AD.

BUT I see  neighbor 10.210.100.1 route-map LOCALPREF in in your config.

So it's possible the Local Preference attribute is changed for eBGP prefixes to beat the iBGP ones and EIGRP never wins consequently?

I know all of this is just a speculation, I'm trying to show what can happen while a redistribution plus eBGP/iBGP peerings are involved at the same time.

If you provide more details, we can discuss them.

Best regards,

Milan

[paul driver]

Hello

In one router all the routes were installed by bgp.  In the second router, almost all the routes werenstalled by eigrp and there are many rib-failures on this one. - Your possible rib failures can be due to the redistribution and eigrp already prefixes in the rib of the router with better admin distance which when BGP is also running will show the bgp routes as rib failures in the bgp table.

res

Paul