Solved: configuring a PW on an SVI, 7609s and ES+

aacole · ‎01-26-2010

I'm trying to get a PW to interconnect 2 vlan interfaces. The routers are 7609-s with ES+ cards:

Mod Ports Card Type                              Model
--- ----- -------------------------------------- ------------------
1    4 7600 ES+                               7600-ES+4TG3C
2    4 7600 ES+                               7600-ES+4TG3C
5    2 Route Switch Processor 720 (Active)    RSP720-3C-GE
6    2 Route Switch Processor 720 (Hot)       RSP720-3C-GE
7    6 Firewall Module                        WS-SVC-FWM-1

I configure the vlan with an xconnect, no error reported during the configuration. But the xconnect dosent come up.

XC ST Segment 1                         S1 Segment 2                         S2
------+---------------------------------+--+---------------------------------+--
DN     ac   Vl10:10(Eth VLAN)            UP mpls 10.255.255.1:101             DN
DN     ac   Vl11:11(Eth VLAN)            UP mpls 10.255.255.1:203             DN

The vlan is up and active, and reading through all the relevant comments on this support forum say this should work. I have the correct hardware (ES+ cards) I also have found exampls on cisoc.com which show this working, it couldnt get much simpler!

If I configure the xconnect between active gigabit interfaces it comes straight up.

The feature navigator indicates this is supported, and I'm told it should work, but it dosent. Any one got any ideas, cos I'm struggling.

This is a debug from the router, it seems to indicate a session error, but not sure why.

Jan 26 18:21:18.371: MPLS peer 10.255.255.1 vcid 101, VC UP, VC state UP
*Jan 26 18:21:18.371: XC L2TP: Failed to find session for peer 10.255.255.1, vcid 101
*Jan 26 18:21:18.471: MPLS peer 10.255.255.1 vcid 101, VC DOWN, VC state DOWN
*Jan 26 18:21:18.471: XC L2TP: Failed to find session for peer 10.255.255.1, vcid 101

Sergei Vasilenko · ‎01-29-2010

Hello,

Which IOS version used on c7600s?

Potentially the root cause is that Port-chan is your core facing MPLS int.

If core-facing is not a bandled int (not a port-chan) than should work just fine in SRD.

AFAIK the MPLS core-facing port-channel interface is not supported

with svi/vfi xconnect configuration before 12.2(33)SRE.

The restriction was that you can not have VPLS or SVI EoMPLS VCs if

you have mpls ip under port-channel interface.

The support was added for ES, ES+ in SRE, while implementing feature

“H-VPLS with Port-Channel Core Interface”:

http://www.cisco.com/en/US/docs/routers/7600/install_config/ES40_config_guide/es40_chap6.html#wp1458749

This feature is available in 12.2(33)SRE, as shown in the following link:

http://www.cisco.com/en/US/docs/routers/7600/install_config/ES40_config_guide/es40_pref.html

So if poosible, consider SRE to test.

Thanks,

Sergey

View solution in original post

Pavel Bykov · ‎01-28-2010

How does your configuration of VC looks like on ES+ ?

aacole · ‎01-28-2010

Hi, thanks for your reply. The configuration I've used is the same as I would use on a physical interface,

int vl 10

xconnect 10.255.255.1 1010 encapsulation mpls

If I configure the same on a physical port and no shut the interfaces the xconnect comes straight up, so I'm sure the underlying MPLS is ok. This is done on the same nodes where I configured the vlan xconnect.

I am thinking this may be a code issue as I've seen examples of this type of configuration, and I'm told it should work.

Pavel Bykov · ‎01-29-2010

Well, if it's PW, then enable detailed status messages:

pseudowire-class PWCLASS
encapsulation mpls
status

int vl 10

xconnect 10.255.255.1 1010 pw PWCLASS

After that, look at the detailed status:

show mpls l2transport vc 1010 detail

Usually i come into contact with LFI or Service Instance constructs

E.g.:

(define service instance under ES interface)

service instance 1 ethernet
encapsulation dot1q 10
rewrite ingress tag pop 1 symmetric
xconnect IP_ADDRESS 50 pw-class eompls

aacole · ‎01-29-2010

Hi, Just tried your suggestion using the PWCLASS, I had tried this before but not included the status keyword.

However, when I look at the output of show mpls l2transport vc 1010 detail I see that there was nothing under the outgoing interface,

and no labels.

Checking my config, the link between the 2 nodes is a port channel, I had mpls enabled on the channel, but not on the physical ports, so tried this, didn’t make any difference.

In the network I also have just built a redundant path via a 3rd node. As part of the debugging work I shut down the port channel.

The xconnect established over the redundant path. This is really odd, as I already had a vrf working across the port-channel, before I built the redundant path, so I thought the MPLS was running ok, now I'm not so sure.

It now looks that the xconnect won't work across a port channel, although I'm not aware of any restrictions regarding this. Am doing more work on this.

Do I need to include the service instance as well? I'm just reading up on this.

Pavel Bykov · ‎01-29-2010

There are many constructs available on how to deploy a L2 service. You can have point to point, point to multipoint, or other topologies. You can also create PWs differently.

The way you are deploying them I come across when working with 3750ME switches. Otherwise on ES linecards nearly all of the deployments are using LFI or service instances. They are just different way of doing things and achieving functionality that you need (or your services require).

Sergei Vasilenko · ‎01-29-2010