EBGP Route taking over EIGRP even after Distance (15) command being used

[vishal agavane]

Hi,

We have 5 nos of location which are directly connected through TDM Fiber optics networks, additionally we have lease line to provide backup path in case of Fiber cut in between any of these 5 locations.
We used EGRP as IGP Routing protocol on TDM links connected to our cisco router, we have published all our local routes over EIGRP routing protocol. Our lease line provider is using EBGP routing protocol for backup connectivity and same has been redistributed into EIGRP. Please refer our network topology in the attachment.

 

Recently we have seen that when our R-2 TDM connectivity towards R-3 has lost and restored back our data source from H-2 host at R5 to destination H-1 host at R2 stop working, after few troubleshooting we came to know that R-5 is learning H-2 host subnet as external (AD-170) and R2 is learning H-2 router as internal (AD-15) routes. Please refer below output from R-1 & R2.

R-2 Configuration 
router eigrp 99
network 10.0.0.0
distance 15 10.11.16.11 0.0.0.0
no auto-summary

R-2#sh ip route 10.11.31.0  <--(R-5 Local Subnet)
Routing entry for 10.11.31.0/24
Known via "eigrp 99", distance 15, metric 30720, type internal
Redistributing via eigrp 99
Last update from 10.11.16.11 on FastEthernet0/0, 12:40:01 ago
Routing Descriptor Blocks:
* 10.11.16.11, from 10.11.16.11, 12:40:01 ago, via FastEthernet0/0
Route metric is 30720, traffic share count is 1
Total delay is 200 microseconds, minimum bandwidth is 100000 Kbit
Reliability 255/255, minimum MTU 1500 bytes
Loading 1/255, Hops 1

===============================

R-5 Configuration

router eigrp 99
redistribute bgp 64520 metric 10000 100 255 1 1500
network 10.11.16.11 0.0.0.0
network 10.11.31.0 0.0.0.255
distance 15 10.11.16.4 0.0.0.0
distance 15 10.11.16.5 0.0.0.0
distance 15 10.11.16.1 0.0.0.0
distance 15 10.11.16.2 0.0.0.0
distance 15 10.11.16.3 0.0.0.0
no auto-summary
!
router bgp 64520
no synchronization
bgp log-neighbor-changes
network 10.11.22.0 mask 255.255.255.0
network 10.11.23.0 mask 255.255.255.0
network 10.11.24.0 mask 255.255.255.0
network 10.11.25.0 mask 255.255.255.0
neighbor 10.67.132.161 remote-as 38529
neighbor 10.67.132.161 description IIDCCCR1
neighbor 10.67.132.161 soft-reconfiguration inbound
neighbor 10.67.132.161 route-map LOW out
no auto-summary

R-5#sh ip int route 10.11.22.0 <--(R-2 Local Subnet)
Routing entry for 10.11.22.0/25
Known via "eigrp 99", distance 170, metric 284160
Tag 38529, type external
Redistributing via eigrp 99, bgp 64520
Advertised by bgp 64520
Last update from 10.11.16.11 on FastEthernet0/0, 12:53:05 ago
Routing Descriptor Blocks:
* 10.11.16.11, from 10.11.16.11, 12:53:05 ago, via FastEthernet0/0
Route metric is 284160, traffic share count is 1
Total delay is 1100 microseconds, minimum bandwidth is 10000 Kbit
Reliability 255/255, minimum MTU 1500 bytes
Loading 25/255, Hops 1
Route tag 38529
10.11.16.10, from 10.11.16.10, 12:53:05 ago, via FastEthernet0/0
Route metric is 284160, traffic share count is 1
Total delay is 1100 microseconds, minimum bandwidth is 10000 Kbit
Reliability 255/255, minimum MTU 1500 bytes
Loading 26/255, Hops 1
Route tag 38529

============================

When our communication was lost, we have used tracert from H-2 (R-5) host to H-1 (R-2) host and observe below hops on continuously cycle due to which TTL expires comes.

(Trace route output)

10.11.31.252. -- (R-5 Local Vlan-2 Interface ip)
10.67.132.161 -- (R-5 LL Provider Edge router Ip)
116.50.125.198 -- (LL provider cloud network)
10.67.8.17 -- (Local R-1 LL Provider Edge router Ip)
10.67.8.18 -- (Local R-1 LL Client Edge router Ip)
10.11.16.11 -- (R-5 TDM Fiber network interface IP)
-----
10.67.132.161 -- (Local R-5 LL Provider Edge router Ip)
116.50.125.198 -- (LL provider cloud network)
10.67.8.17 -- (Local R-1 LL Provider Edge router Ip)
10.67.8.18 -- (Local R-1 LL Client Edge router Ip)
10.11.16.11 -- (R-5 TDM Fiber network interface IP)
------

As this was happened in production network so to make this up we have used below static route on R-2 for single local host (10.11.22.1) afterward communication established. Please check below routes and show ip route output on R-5 after putting these below routes on R-2.

R2(Config)#ip route 10.11.22.1 255.255.255.255 Vlan2 (This is local subnet for R-2 even though we have tried this and communication was established.)

After putting this routes on R-2 we have seen R-1 started learning R-2 routes through internal EIGRP however eigrp AD value show 90 but I think it should be 15 as per EIGRP configuration.

R-5#sh ip route 10.11.22.0 (For entire R-2 subnet it's shows external route)
Routing entry for 10.11.22.0/25
Known via "eigrp 99", distance 170, metric 284160
Tag 38529, type external
Redistributing via eigrp 99, bgp 64520
Advertised by bgp 64520
Last update from 10.11.16.11 on FastEthernet0/0, 12:53:05 ago
Routing Descriptor Blocks:
* 10.11.16.11, from 10.11.16.11, 12:53:05 ago, via FastEthernet0/0
Route metric is 284160, traffic share count is 1
Total delay is 1100 microseconds, minimum bandwidth is 10000 Kbit
Reliability 255/255, minimum MTU 1500 bytes
Loading 25/255, Hops 1
Route tag 38529
10.11.16.10, from 10.11.16.10, 12:53:05 ago, via FastEthernet0/0
Route metric is 284160, traffic share count is 1
Total delay is 1100 microseconds, minimum bandwidth is 10000 Kbit
Reliability 255/255, minimum MTU 1500 bytes
Loading 26/255, Hops 1
Route tag 38529

R-5#sh ip route 10.11.22.1 (After applying static route for 22.1 on R2 its show internal route with metric 90)
Routing entry for 10.11.22.1/32
Known via "eigrp 99", distance 90, metric 25628160, type internal
Redistributing via eigrp 99, bgp 64520
Advertised by bgp 64520
Last update from 10.11.16.2 on FastEthernet0/0, 17:26:42 ago
Routing Descriptor Blocks:
* 10.11.16.2, from 10.11.16.2, 17:26:42 ago, via FastEthernet0/0
Route metric is 25628160, traffic share count is 1
Total delay is 1000100 microseconds, minimum bandwidth is 100000 Kbit
Reliability 255/255, minimum MTU 1500 bytes
Loading 26/255, Hops 1

Any idea on this?

Thanks,

[Giuseppe Larosa]

Hello @vishal agavane ,

your fixup static route is configured in the following way:

>> ip route 10.11.22.1 255.255.255.255 Vlan2

you have specified an exit interface and not a next-hop in this case EIGRP network command creates a local internal route for prefix 10.11.22.1/32 on R2.

on R2 the route has AD like a connected route.

On R-5 the route should be seen with AD 15 because it is coming from 10.11.16.2 that is in the list of neighbors with reduced AD.

But it shows AD 90 the default value.

General considerations: if you have mutual redistributions of EIGRP into BGP and BGP into EIGRP in two different routers you should do it in a controlled way using route-maps . You can use route tags on EIGRP to mark prefixes learned via BGP and you can use BGP community to mark prefixes learned from EIGRP into BGP.

Hope to help

Giuseppe

[vishal agavane]

Understood, will try this an update you on this.

[paul driver]

Hello

@vishal agavane wrote:

After putting this routes on R-2 we have seen R-1 started learning R-2 routes through internal EIGRP however eigrp AD value show 90 but I think it should be 15 as per EIGRP configuration.

FYI -  the eigrp distance is only locally significant so you shouldn't see R2 set distance on R1 and would suggest specify R3 rid not R5

Also I would say if you append specific networks to the routing stanza of R2 then you may see different result.

R2
router eigrp 99
no network 10.0.0.0
no distance 15 10.11.16.11 0.0.0.0
distance 15 10.11.16.3 0.0.0.0
network 10.11.22.254 0.0.0.0
network 10.11.16.2 0.0.0
etc...

[vishal agavane]

Sure will check this and update.

[Elliot Dierksen]

The route you have installed was a /25. I am guessing based on what I see that the BGP routes might be /24's. Admin distance decides which route is installed when the length of the routes is the same. You could have a /25 route installed for both 10.11.22.0/25 and 10.11.22.128/25 as well as a route for 10.11.22.0/24. The /24 route would never get used because the /25 is a longer match.

[vishal agavane]

On R2 we have Vlan-x with interface ip 10.11.22.100/24 and under EIGRP we have advertise 10.0.0.0 network. We haven't use /25 route anywhere on R-2, however on R5 10.11.22.0/25 route being learned as external routes? is it because of LL provider used /25 mask for 10.11.22.0? and due to which /25 route being preferred over /24 route?

if this is true then how we are going to check what mask is being used and propagated by LL ISP? any debug command?

[Giuseppe Larosa]

Hello @vishal agavane ,

talk wiith the provider however being external route they come from BGP,

but begin more specific then the real one can create issues

@Elliot Dierksen is on the right track to help you.

show  ip bgp 10.11.22.0

Hope to help

Giuseppe

[Elliot Dierksen]

The mask in the network statement is used to identify interfaces. Only interfaces, and nothing more. The route that will be advertised is based on the netmask assigned to the interface. If you had two /25 interfaces that are part of the same /24 at a site and wanted to advertise a /24 to the rest of the network (which would make sense), you would have to do that with a summary. It is a common misconception that you have to put in a "network" statement to make an interior gateway protocol aka IGP (RIP, EIGRP OSPF) pass a route, but that is not the case. The network statement for IGP's is only to select interfaces. That isn't true with an exterior gateway protocol aka EGP like BGP.

[vishal agavane]

Thanks Elliot for your explanation.

As mentioned above we dont have interface with /25 mask, we only have single Vlan interface with /24 mask however ip route shows 10.11.22.0/25 as external route advertised by bgp. Does this mean ISP advertising this route with /25 mask?

R5 have working EIGRP neighbourship with R2 and R2 has propagated 10.11.22.0 network to R5, in this case R5 should show 10.11.22.0/24 as internal route however R5 is showing 10.11.22.0/25 external route in the routing table? any suggestion to resolve this issue?

R-5#sh ip route 10.11.22.0 (For entire R-2 subnet it's shows external route)
Routing entry for 10.11.22.0/25
Known via "eigrp 99", distance 170, metric 284160
Tag 38529, type external
Redistributing via eigrp 99, bgp 11111
Advertised by bgp 11111

If you could check traceroute output it shows that packet targeting 10.11.22.0 network from R5 is getting loop between R5 & R1. When we used ip route on R1 & R5 it show below route in its routing table.

R-5#sh ip route

B 10.11.22.0/25 [20/0] via 10.67.132.161, 20:12:39
D 10.11.22.0/24 [15/25628160] via 10.11.16.2, 17:03:56, FastEthernet0/0

B 10.11.22.128/25 [20/0] via 10.67.132.161, 20:12:40

R-1#sh ip route

D EX 10.11.22.0/25 [170/284160] via 10.11.16.11, 12:53:40, FastEthernet0/0
[170/284160] via 10.11.16.10, 12:53:40, FastEthernet0/0
D 10.11.22.0/24 [90/25628160] via 10.11.16.2, 17:27:15, FastEthernet0/0

D EX 10.11.22.128/25
[170/284160] via 10.11.16.11, 12:53:41, FastEthernet0/0
[170/284160] via 10.11.16.10, 12:53:41, FastEthernet0/0

Please suggest.

[Elliot Dierksen]

The "type external" and admin distance of 170 points to it being a route that was redistributed into EIGRP somewhere and then learned by this router. The /25 BGP route with an admin distance of 20 says you learned it from an eBGP peer. Perhaps they mistyped something when they created the route? It is hard to say, but you are definitely getting a /25 from your eBGP peer.

I may be waxing philosophically here, but mutual redistribution is a very dicey proposition. I would encourage you to think about it in a different way. Does a spoke router need to learn every single route? No. Does a spoke router need to be able to reach every single route? Most likely yes. Think of areas where you can segment things and only provide a default or perhaps a summary to the sections of your network that only have one outbound path. You'll save memory and CPU in the routers, plus you will avoid the chainsaw juggling that is mutual redistribution.