Posted on March 12, 2015 2:24 pm
 |  Asked by Vivek Dasgupta
 |  1537 views
Tags:
RESOLVED
1
0
Print Friendly, PDF & Email

I am trying to setup a basic VXLAN configuration using vEOS on KVM VMs. Topology is as follows :

[Linux VM1][edge1][Eth2][core1][Eth1][Eth1][core2][Eth2][edge2][Linux VM2]

Each switch has two ports apart from the mgmt port. Both edge switches are configured to allow VLAN 42. I am trying to setup VTEPs on core1 and core2, hoping to see VXLAN traffic between them.

Basic connectivity is there. Without any VLANs I am able to ping between Linux VM1 and VM2. Even with trunk ports on core switches allowing VLAN 42, I am able to connect.

Now onto the VXLAN config

--------------------------------------
 core1 :
 -------
Interface vxlan1
 vxlan multicast group 227.10.1.1
 vxlan source-interface Loopback15
 vxlan udp-port 4789
 vxlan vlan 42 vni 5555
 vxlan flood vtep 10.25.25.4
Interface eth1 :
Interface eth2 :
 switchport trunk allowed vlan 42
 switchport mode trunk
Interface Loopback15
 ip address 10.25.25.3/24
core2 :
 -------
Interface vxlan1
 vxlan multicast group 227.10.1.1
 vxlan source-interface Loopback15
 vxlan udp-port 4789
 vxlan vlan 42 vni 5555
 vxlan flood vtep 10.25.25.3
Interface eth1 :
Interface eth2 :
 switchport trunk allowed vlan 42
 switchport mode trunk
Interface Loopback15
 ip address 10.25.25.4/24
On both switches :
 #sh int vxlan 1
 Vxlan1 is up, line protocol is up (connected)
 Hardware is Vxlan
 Source interface is Loopback15 and is active with
 Remote MAC Learning via DataPath
 Static vlan to vni mapping is
 [42, 5555]
 Headend replication flood vtep list is
 42
 VTEP learning restricted to
 42 Any
-------------------------------------------

With this I am unable to ping between Linux VM1 and Linux VM2. I am also not able to see any VXLAN traffic between core1 and core2.

Any pointers would be helpful …

Thanks
V

1
Posted by Alex
Answered on March 16, 2015 9:39 pm

Hi Vivek  in your configuration you are using a multicast control plane, you need to therefore run PIM on the routed interfaces (ethernet1) connecting the Core1 and Core2 switches . Also you would need to advertise the loopback addresses of both VTEPs across the routed link, to provide IP connectivity between the VTEPs.

Alex

 

If you are going with head-end replication (flood lists on each VTEP), you can clean up the config by removing the vxlan multicast group configuration. As Alex said, confirm the routes to each other loopback are valid.

If you want to stick with multicast, then remove the flood-list and confirm that this group is carried through the routed network.

(John Gill at May 5, 2015 1:32 am)

Post your Answer

You must be logged in to post an answer.