xconnect

[vin.marco]

Hi guys
I have to configure an xconnect between a PE with ios XR version and a PE with ios XE version.

The scenario is as follows:

I have 2 switches that have vlan, in this case 100, which must communicate through an MPLS network.
The MPLS network is configured in LDP and fully functional.

I have configured the PE interfaces, but something is not working.

interface GigabitEthernet4
 no ip address
 negotiation auto
 no keepalive
 no mop enabled
 no mop sysid
 xconnect 10.86.1.1 100 encapsulation l2tpv3 pw-class TEST
 service instance 2 ethernet
  encapsulation dot1q 100
  rewrite ingress tag pop 1 symmetric
  bridge-domain 2
 !
end
!
pseudowire-class TEST
encapsulation l2tpv3
interworking ethernet
ip local interface Loopback86
!

l2vpn
 pw-class TEST
 !
 xconnect group TEST
  p2p TEST
   interface GigabitEthernet0/0/0/3.100
  !
 !
!
interface GigabitEthernet0/0/0/3.100 l2transport
 encapsulation dot1q 100
 rewrite ingress tag pop 1 symmetric
!

I know how to tell where it is wrong.
Thank you

[nkarpysh]

Hello,

I dont see l2tpv2 encap and neighbor on your ASR9k config.

Should be smth like this:

l2vpn
pw-class XXXXX
encapsulation l2tpv3
protocol l2tpv3
ipv4 source 10.86.1.1
!
!
xconnect group XXXXX
p2p XXXXX
interface GigabitEthernet0/0/0/3.100
neighbor ipv4 Y.Y.Y.Y pw-id 1

   pw-class XXXXX

HTH

Niko

[vin.marco]

Hi and thanks for the support
I tried to implement this configuration but it doesn't work.
I am attaching the configuration:

ASR900

!
!
pseudowire-class XXX-XXX
 encapsulation l2tpv3
 interworking ethernet
 ip local interface Loopback86
!
!         
interface GigabitEthernet0/2/1
 no ip address
 negotiation auto
 no keepalive
 xconnect Loopback ASR9k 500 encapsulation l2tpv3 pw-class XXX-XXX
 service instance 10 ethernet
  encapsulation dot1q 100
  rewrite ingress tag pop 1 symmetric
 ! 

ASR9K

l2vpn
 pw-class XX-XX-XXXXXXX
  encapsulation l2tpv3
  !
 !
 xconnect group XX-XX-XXXXXXX
  p2p XX-XX-XXXXXXX
   interface GigabitEthernet0/1/0/10.100
   neighbor ipv4 Loopback ASR900 pw-id 500
    pw-class XX-XX-XXXXXXX
   !
  !
 !
interface GigabitEthernet0/1/0/10.100 l2transport
 encapsulation dot1q 100
!

Am I wrong something?

[vin.marco]

xconnect is UP,
the configuration I have implemented is:

l2vpn
 pw-class XX-XX-XXXXXXX
  encapsulation l2tpv3
  protocol l2tpv3
  ipv4 source Loopback ASR9k
  !
 !
 xconnect group XX-XX-XXXXXXX
  p2p XX-XX-XXXXXXX
   interface GigabitEthernet0/1/0/10.100
   neighbor ipv4 Loopback ASR900 pw-id 500
    pw-class XX-XX-XXXXXXX
   !
  !
 !
interface GigabitEthernet0/1/0/10.100 l2transport
 encapsulation dot1q 100
 rewrite ingress tag pop 1 symmetric
!

So now what I get in the GigabitEthernet0 / 1/0 / 10.100 interface in vlan 100 is untagged and rotated in the xconnect tunnel.
Quite right ?

But if I want to add other vlans in the same interface, should I create other sub-interfaces or add the vlans in encapsulatipon dot1q?

[nkarpysh]

Hey,

For the interface - glad you figured it out. It needs to be symmetric between your IOS-XE and IOS-Xr. So if you pop VLAn on XE side, so you should do it on the XR side as well:

XE:

 service instance 10 ethernet
  encapsulation dot1q 100
  rewrite ingress tag pop 1 symmetric

XR:

interface GigabitEthernet0/1/0/10.100 l2transport
 encapsulation dot1q 100
 rewrite ingress tag pop 1 symmetric

FOr additional VLANs - you will need to create separate sub-interfaces and tag those with corresponding dot1q accordingly. Next you will need to create separate xconnect for new interface. You can use same neighbor statements, just interface will be different:

 xconnect group ZZZ
  p2p ZZZ
   interface GigabitEthernet0/1/0/10.200
   neighbor ipv4 Loopback ASR900 pw-id 600
    pw-class XX-XX-XXXXXXX

And accordingly on IOS-XE side.

Niko

[vin.marco]

OK thanks,
so it is not possible to carry other vlna in the same xconnect?
The switch has several vlans configured, the 100 is one of the 4 that I have to carry.

Couldn't I create sub-interfaces for each vlan and transport it to the same pw-id 500?

[nkarpysh]

In xconnect - no, you cant, Xconnect will need to know to which VLAN incoming packet belongs and it wont be able to distinguish.

Partially you can achieve this with regular VPLS (multipoint L2 service). but I would NOT recommend it.

If those are different VLANs, mainly different L2 domains then you should isolate each into different L2 transport on the routers side, thus those all need be carried on separate bridge-domains at least. Otherwise you'll create a potential for L2 loops.

Niko

[vin.marco]

I confirm that tunnel l2 is UP but I cannot ping from the switches.
The switch ports are configured in dot1q with vlan allowed 100.

what's wrong with it?

[nkarpysh]

Hey,

Its hard to guess. Few assumptions you can check:

- Verify that switches are connected to routers with trunk port (no with access port of VLAN 100).

- Also you can ping e.g. with count of 1000 and time 0 and then check interface counters on IOS-XE, IOS-XR and counters related to l2vpn (e.g. on XR command will be "show l2vpn forwarding" e.g.:

RP/0/RSP0/CPU0:router# show l2vpn forwarding	bridge detail location 0/0/cpu0
Bridge-domain name: bg1:bd1, id: 0, state: up
 MAC learning: enabled
 Flooding:
   Broadcast & Multicast: enabled
   Unknown unicast: enabled
 MAC aging time: 300 s, Type: inactivity
 MAC limit: 4000, Action: none, Notification: syslog
 MAC limit reached: no
 Security: disabled
 DHCPv4 snooping: profile not known on this node
 IGMP snooping: disabled, flooding: disabled
 Bridge MTU: 1500 bytes
 Number of bridge ports: 1
 Number of MAC addresses: 0
 Multi-spanning tree instance: 0

  GigabitEthernet0/0/0/1.2, state: oper up
    Number of MAC: 0
    Statistics:
      packets: received 0, sent 0
      bytes: received 0, sent 0

This way you can see how many packets received on local AC and sent to xconnect and simetricly on the other side to find where the drops occurred.

- Or you can simply can open TAC case and engineer can do quickyl webex with you to verify it.

Niko