[ocata-master-414]ERB: Inter-VN routing between BMSs occurs at the Spine and not at the leaf

[Ganesha HV]

1]. Created a topo with :

a). contrail-cluster

nodek10(10.204.216.230) - config/ctrl/openstack
nodek8(10.204.216.228) - compute1
nodek9(10.204.216.229) - compute2
nodei30(10.204.217.142) - contrail-command

b). QFX10002(10.204.216.155) as Spine.

c). QFX5110(10.204.216.131) as leaf, with 2 BMSs attached.
    BMS1 - nodej4(10.204.217.161)
    BMS2 - nodeg11(10.204.217.51)

2]. Created 2 VNs:
 vn1 - 10.10.10.0/24
 vn2 - 20.20.20.0/24

3]. Launched a VM in each of them.

4]. Enabled VxLAN routing in the project.

5]. Created a LR and attached the two networks. Extended this LR to the Spine QFX.

6]. Added the two BMSs to the Leaf QFX.

7]. Leaf QFX was assigned a ERB-UCAST-Gateway role while the Spine QFX was assigned a CRB-MCAST-Gateway role.

8]. After the configs were pushed, I was able to see traffic across the BMS reaching the spine and back to the leaf which is not the desired behaviour. We would need the traffic to be routed locally at the leaf.

For this, I went thru the configs pushed on both the devices and found that the routing-instance config is missing in the Leaf devices. The IRBs are created, but no VGW config is seen on them.

Output of traceroute from BMS1 to BMS2:

root@nodej4:~# traceroute 10.10.10.10
traceroute to 10.10.10.10 (10.10.10.10), 30 hops max, 60 byte packets
 1 20.20.20.6 (20.20.20.6) 0.736 ms 0.957 ms 1.306 ms . <<< Spine's IRB
 2 10.10.10.10 (10.10.10.10) 0.290 ms 0.296 ms 0.291 ms
root@nodej4:~#

root@nodeg11:~# traceroute 20.20.20.20
traceroute to 20.20.20.20 (20.20.20.20), 30 hops max, 60 byte packets
 1 10.10.10.6 (10.10.10.6) 1.040 ms 1.214 ms 1.447 ms <<< Spine's IRB
 2 20.20.20.20 (20.20.20.20) 0.202 ms 0.202 ms 0.465 ms
root@nodeg11:~#

{master:0}[edit]
root@bng-contrail-qfx-1k2# run show interfaces terse irb
Interface Admin Link Proto Local Remote
irb up up
irb.5 up up inet 10.10.10.1/24
                                            10.10.10.6/24
irb.6 up up inet 20.20.20.1/24
                                            20.20.20.6/24

{master:0}[edit]
root@bng-contrail-qfx-1k2#

Output of traceroute from BMS1 to VM1 and VM2:

root@nodej4:~# traceroute 10.10.10.3
traceroute to 10.10.10.3 (10.10.10.3), 30 hops max, 60 byte packets
 1 20.20.20.6 (20.20.20.6) 0.795 ms 0.792 ms 0.785 ms <<< Spine's IRB
 2 * * *
 3 10.10.10.3 (10.10.10.3) 1.094 ms 1.207 ms 1.322 ms
root@nodej4:~#

root@nodej4:~# traceroute 20.20.20.3
traceroute to 20.20.20.3 (20.20.20.3), 30 hops max, 60 byte packets
 1 20.20.20.3 (20.20.20.3) 0.831 ms 0.793 ms 0.833 ms
root@nodej4:~#

Attached is a picture of the topology.

[Ganesha HV (ganeshahv)]

Attached pic of the topo.

[Rishabh Tulsian (rtulsian)]

Fix committed, available in latest build