ebgp+ label

[sachin30720041]

Hello All,

I have configured following setup. Ebgp + label is configured on  all 3 routers.

11.11.11.11/32 is advertised from R1 in bgp.

33.33.33.33/32 is advertised from R3 in bgp.

R1 received 33.33.33.33/32 through bgp.

R1# show bgp ipv4 unicast 33.33.33.33
BGP routing table entry for 33.33.33.33/32, version 4
Paths: (1 available, best #1, table default)
Not advertised to any peer
Refresh Epoch 1
2 3
2.2.2.2 (metric 2) from 2.2.2.2 (22.22.22.22)
Origin IGP, localpref 100, valid, external, best
mpls labels in/out nolabel/17

R3 received 11.11.11.11/32 through bgp.

R3# show bgp ipv4 unicast 11.11.11.11
BGP routing table entry for 11.11.11.11/32, version 2
Paths: (1 available, best #1, table default)
Not advertised to any peer
Refresh Epoch 1
2 1
2.2.2.2 (metric 2) from 2.2.2.2 (22.22.22.22)
Origin IGP, localpref 100, valid, external, best
mpls labels in/out nolabel/16

But on R1, ping from 11.11.11.11 to 33.33.33.33 does not work.

R1# ping 33.33.33.33 source 11.11.11.11

.........

I do not know why ping is not working.

R1-----R2----R3

R1

------

int gi1
ip address 100.10.20.1 255.255.255.0
ip ospf 1 area 0
no sh

int lo0
ip address 1.1.1.1 255.255.255.255
ip ospf 1 area 0

int lo1
ip address 11.11.11.11 255.255.255.255
exit

router ospf 1
router-id 1.1.1.1

router bgp 1
template peer-session session1
remote-as 2
update-source lo0
ebgp-multihop 255
exit
neighbor 2.2.2.2 inherit peer-session session1
address-family ipv4 unicast
network 11.11.11.11 mask 255.255.255.255
neighbor 2.2.2.2 activate
neighbor 2.2.2.2 send-label

R2

----------------
int gi1
ip address 100.10.20.2 255.255.255.0
ip ospf 1 area 0
no sh

int gi2
ip address 100.20.30.2 255.255.255.0
ip ospf 1 area 0
no sh

int lo0
ip address 2.2.2.2 255.255.255.255
ip ospf 1 area 0

int lo1
ip address 22.22.22.22 255.255.255.255

router ospf 1
router-id 2.2.2.2

router bgp 2
template peer-session session1
remote-as 1
update-source lo0
ebgp-multihop 255
exit
template peer-session session2
remote-as 3
update-source lo0
ebgp-multihop 255
exit
neighbor 1.1.1.1 inherit peer-session session1
neighbor 3.3.3.3 inherit peer-session session2
address-family ipv4 unicast
network 22.22.22.22 mask 255.255.255.255
neighbor 3.3.3.3 activate
neighbor 3.3.3.3 send-label
neighbor 1.1.1.1 activate
neighbor 1.1.1.1 send-label

R3

----------------------
int gi2
ip address 100.20.30.3 255.255.255.0
ip ospf 1 area 0
no sh

int lo0
ip address 3.3.3.3 255.255.255.255
ip ospf 1 area 0

int lo1
ip address 33.33.33.33 255.255.255.255

router ospf 1
router-id 3.3.3.3

router bgp 3
template peer-session session1
remote-as 2
update-source lo0
ebgp-multihop 255
exit
neighbor 2.2.2.2 inherit peer-session session1
address-family ipv4 unicast
network 33.33.33.33 mask 255.255.255.255
neighbor 2.2.2.2 activate
neighbor 2.2.2.2 send-label

[AnthonySylvester]

What does the routing table on R1 and R3 look like? You might have received the BGP routes, but if they aren't in the RIB, then the router can't route the ping traffic.

[sachin30720041]

R1#show ip route bgp
Gateway of last resort is not set

22.0.0.0/32 is subnetted, 1 subnets
B 22.22.22.22 [20/0] via 2.2.2.2, 00:49:30
33.0.0.0/32 is subnetted, 1 subnets
B 33.33.33.33 [20/0] via 2.2.2.2, 00:48:59

R3#show ip route bgp
Gateway of last resort is not set

11.0.0.0/32 is subnetted, 1 subnets
B 11.11.11.11 [20/0] via 2.2.2.2, 00:49:12
22.0.0.0/32 is subnetted, 1 subnets
B 22.22.22.22 [20/0] via 2.2.2.2, 00:49:12

[AnthonySylvester]

I just labbed up your scenario. Looks like no label is present because the entry is a recursive route without a label. You can check CEF to see this:

R3#sh ip cef 11.11.11.11 detail 
11.11.11.11/32, epoch 0, flags [rib defined all labels]
  recursive via 2.2.2.2 label 16
    recursive via 100.20.30.2 unusable: no label

If you add a static route to R1's or R3's loopback and send it to R2 you should be fine. Here is the output after I add the static routes on R1 and R3 for each other (on R3: ip route 11.11.11.11 255.255.255.255 100.20.30.2):

R3#sh ip cef 11.11.11.11
11.11.11.11/32
  nexthop 100.20.30.2 GigabitEthernet0/2


R3#ping 11.11.11.11 source 33.33.33.33
Type escape sequence to abort.
Sending 5, 100-byte ICMP Echos to 11.11.11.11, timeout is 2 seconds:
Packet sent with a source address of 33.33.33.33 
!!!!!
Success rate is 100 percent (5/5), round-trip min/avg/max = 2/3/4 ms

[MHM Cisco World]

this is Lab I do and success to ping, 

but 

please note you must use 

ping mpls not only ping command.

[AnthonySylvester]

Agreed. That'll setup LDP neighborships and allow the LSP end to end.

[Harold Ritter]

Hi @AnthonySylvester ,

You don't need the LDP neighborship if you configure the BGP sessions using the directly connected interfaces instead of the loopback interfaces.

Regards,

[MHM Cisco World]

sorry but R2 need to LDP to forward packet depend on label not on RIB table.
that why I config the LDP.

[Harold Ritter]

Hi @MHM Cisco World ,

R2 will be able to forward MPLS traffic if the BGP sessions are established using the directly connected interfaces. This is how large data centre customers are running MPLS in their infrastructure without requiring any LDP or IGP, just BGP ipv4 + label. This scales much better.

Regards,

[MHM Cisco World]

one Question here, is the packet is label or not 
if it label then R2 need to know the label of 11.11.11.11/33.33.33.33 
if it not label then R2 need to know how to traffic the 11.11.11.11/33.33.33.33
am I right ?

for direct connection Yes we don't need to config LDP in interface connect both BGP peer (like the case of MPLS inter-AS)
for in-direct connection we need make middle Hop to know how handle traffic.

that what I understand.

[Harold Ritter]

Hi @MHM Cisco World ,

Yes, packets will be labelled. R2 learns the labels for 11.11.11.11/32 and 33.33.33.33/32 via eBGP IPv4 + Label. As you mentioned, LDP + IGP is also a valid scenario, but for large scale DC, it does not scale as well.

Regards,

[Harold Ritter · Accepted solution]

Hi @sachin30720041 ,

If you want traffic to be label switched, you need to establish the BGP sessions using the directly connected interfaces rather than the loopback address.

Regards,