CEF and host route

[verma-rohit]

Hi,

R1-------------R2

R1:

lo0: 1.1.1.1/32

R2:

lo02.2.2.2/32

I have MP-BGP configured between R1 and R2.

Everything works fine, however on R1 i summarized loopback in ospf as 1.1.0.0/16. R2 routing table has 1.1.1.1/32 route  and has 1.1.0.0/16 as configured. However i noticed CEF did install 1.1.1.1/32 host route which i find strange and I believe it has to do with MP-BGP configuration.

When i remove MP-BGP and runs plain ospf between two routers, host route 1.1.1.1/32 no longer exits. So why Mp-BGP is injecting a host route  when route is summarized?

R2: When MP-BGP is in operation

R2#show ip cef
Prefix Next Hop Interface
0.0.0.0/0 no route
0.0.0.0/8 drop
0.0.0.0/32 receive
1.1.0.0/16 192.168.1.1 Ethernet0/0
1.1.1.1/32 192.168.1.1 Ethernet0/0
2.2.2.2/32 receive Loopback0
127.0.0.0/8 drop
192.168.1.0/24 attached Ethernet0/0
192.168.1.0/32 receive Ethernet0/0
192.168.1.1/32 attached Ethernet0/0
192.168.1.2/32 receive Ethernet0/0
192.168.1.255/32 receive Ethernet0/0
224.0.0.0/4 drop
224.0.0.0/24 receive
240.0.0.0/4 drop
255.255.255.255/32 receive

R2: When MP-BGP is removed      

R2#show ip cef
Prefix Next Hop Interface
0.0.0.0/0 no route
0.0.0.0/8 drop
0.0.0.0/32 receive
1.1.0.0/16 192.168.1.1 Ethernet0/0
2.2.2.2/32 receive Loopback0
127.0.0.0/8 drop
192.168.1.0/24 attached Ethernet0/0
192.168.1.0/32 receive Ethernet0/0
192.168.1.1/32 attached Ethernet0/0
192.168.1.2/32 receive Ethernet0/0
192.168.1.255/32 receive Ethernet0/0
224.0.0.0/4 drop
224.0.0.0/24 receive
240.0.0.0/4 drop
255.255.255.255/32 receive

[Georg Pauwen]

Hello,

share the running configs of both routers...

[verma-rohit]

1.Configuration files are attached.
2. On R1, I have intentionally summarized lo0 as 1.1.0.0/16.

3. On R2: Why CEF has host route for R1 lo0 even though it is summarized on R1. RIB does not have host route.

  R2#show ip cef
Prefix Next Hop Interface
0.0.0.0/0 no route
0.0.0.0/8 drop
0.0.0.0/32 receive
1.1.0.0/16 192.168.1.1 Ethernet0/0
1.1.1.1/32 192.168.1.1 Ethernet0/0
2.2.2.2/32 receive Loopback0
127.0.0.0/8 drop
192.168.1.0/24 attached Ethernet0/0
192.168.1.0/32 receive Ethernet0/0
192.168.1.1/32 attached Ethernet0/0
192.168.1.2/32 receive Ethernet0/0
192.168.1.255/32 receive Ethernet0/0
224.0.0.0/4 drop
224.0.0.0/24 receive
240.0.0.0/4 drop
255.255.255.255/32 receive 

R2#show ip cef 1.1.1.1 detail
1.1.1.1/32, epoch 0
1 RR source [active source]
Dependent covered prefix type rr, cover 1.1.0.0/16
recursive via 1.1.0.0/16
nexthop 192.168.1.1 Ethernet0/0

R2: RIB

1.0.0.0/16 is subnetted, 1 subnets
O IA 1.1.0.0 [110/11] via 192.168.1.1, 00:14:15, Ethernet0/0
2.0.0.0/32 is subnetted, 1 subnets
C 2.2.2.2 is directly connected, Loopback0
192.168.1.0/24 is variably subnetted, 2 subnets, 2 masks
C 192.168.1.0/24 is directly connected, Ethernet0/0
L 192.168.1.2/32 is directly connected, Ethernet0/0

Note: If i change R1 bgp router-id to some other loopback ip, host route for 1.1.1.1/32 goes away on R2.

[Georg Pauwen]

Hello,

Iabbed this in GNS3. The problem is the redistribution of connected routes:

address-family ipv4 vrf A
redistribute connected
exit-address-family

If you remove that, the host route in CEF on R2 is gone.

[verma-rohit]

Redistribution is for connected routes in vrf A where as lo0 is in R1 global routing. Lo0 is not part of any vrf

[Georg Pauwen]

Hello,

have you tried to remove the 'redistribute connected' ? What is the result ?

[verma-rohit]

Yes I did and it did remove cef entry for host route but I am unable to get my head around it. 

Redistribute connected was for vrf A where as host route entry was for loopback 0 which was not in vrf A. I am assuming , when redistribution was in place , lo1 was advertised as vpnv4 route to R2 with its lo0 as  source. Now as lo0  is in global table, it got redistributed. 

[Harold Ritter]

It is recommended not to summarize the loopback address used as the destination for the L3VPN service.

In your case, the /32 is automatically generated in the CEF table for the next hop of the VPNv4 prefixes, hence you only see it if the redistributed connected is configured. This /32 only exist in the CEF table, not in the RIB.

Regards,

[verma-rohit]

To Add it further:



(PE)R1--------R2(PE)





1. R1 lo0 is 1.1.1.1 and it is summarized as 1.1.0.0/16. R1 will assign label to both routes and advertise it to R2.
2. R2 RIB will have 1.1.0.0/16 and NO 1.1.1.1/32. ldp binding received is for 1.1.0.0/16 and 1.1.1.1/32.Now R2 CEF has a route for 1.1.1.1/32 because it the update source on R1 was lo0 but mpls forwarding table has no entry for 1.1.1.1/32 with a local assigned label.



On R2, doesn't mpls forwarding table should have a route with a label for 1.1.1.1/32. The exact route is CEF. There is broken LSP as mpls ping results in all Q but I am unable to get my head around it.

[Harold Ritter · Accepted solution]

This should work, as 1.1.1.1/32 will be recursively resolved via 1.1.0.0/16 and will use its label. I just tried it and it works as expected.

R2#sh ip cef 1.1.1.1 det
1.1.1.1/32, epoch 0
1 RR source [active source]
Dependent covered prefix type rr, cover 1.1.0.0/16
recursive via 1.1.0.0/16
nexthop 192.168.24.2 GigabitEthernet0/1 label 16-(local:17)
R2#

[verma-rohit]

Hi Harold,

Yes it indeed worked. However, when i issue mpls ping from R1 ( PE ) to R2(PE) it fails. I get response as request not sent ( QQQ). Now there is  end to end label , not for 1.1.1.1/32 but for 1.1.0.0/16. 

Any reason why LSP is broken ? 

[Harold Ritter]

If R1 is the one summarizing, the ping not working should be in the other direction (R2 to R1). It should work from R2 to R1, if you use "ping mpls ipv4 1.1.0.0/16". No reason for it not to work from R1 to R2 though.

Regards

[verma-rohit]

Thanks Harold fr the reply. Please bear with me for some more time. I have injected 11.11.11.11/32 on R1. Output below is on R2

R2#show ip route vrf A 11.11.11.11

Routing Table: A
Routing entry for 11.11.11.0/24
Known via "bgp 100", distance 200, metric 0, type internal
Last update from 1.1.1.1 00:07:25 ago
Routing Descriptor Blocks:
* 1.1.1.1 (default), from 1.1.1.1, 00:07:25 ago
Route metric is 0, traffic share count is 1
AS Hops 0
MPLS label: 17
MPLS Flags: MPLS Required

Here next HOP is 1.1.1.1. So i looked into mpls forwarding table for 1.1.1.1
R2#
R2#show mpls for
R2#show mpls forwarding-table 1.1.1.1
Local       Outgoing      Prefix            Bytes Label            Outgoing              Next Hop
Label       Label           or Tunnel Id    Switched               interface
None      No Label        1.1.1.1/32        0                          Et0/0           192.168.1.1

There is no label for 1.1.1.1 in LFIB. I agree it is  borrowing label from 1.1.0.0/16 

R2#show ip cef 1.1.1.1 detail
1.1.1.1/32, epoch 0
1 RR source [active source]
Dependent covered prefix type rr, cover 1.1.0.0/16
recursive via 1.1.0.0/16
nexthop 192.168.1.1 Ethernet0/0

I understand now that label swap  from R2 to R1 for R1 lo0 is for network 1.1.0.0/16 but what is bothering me that next hop is 1.1.1.1 and mpls forwarding table for the said next hop ( 1.1.1.1) shows no label. 

What am i missing ?