Mailing List Archive

RSVP-TE (MPLS-TE) and LDP question
Seems that when I try to use RSVP in place of LDP for label distribution, I
cannot completely remove mpls ldp configs from IOS XR, but I can from IOS XE



On an RSVP-TE Tunnel headend, I have .



IOS XR (XRv9000)



mpls ldp

router-id 10.0.0.11



.and if I remove that with "no mpls ldp" I loose connectivity to the MPLS
L3VPN that is also on that PE



But.in IOS XE (csr1000v) I have.



mpls ldp router-id lo0 force



.and if I remove that with "no mpls ldp router-id Loopback0" (and also
remove "mpls ip" from the pe---p uplink) I am still good to the MPLS L3VPN
that is also on that PE



I don't understand what is going on with this minimal ldp config in IOS XR
that causes L3VPN to no longer work after I remove that small config shown
above.



As a side note, I can remove that ldp config from XR p core nodes.. Just not
XR pe nodes

.furthermore, I think since I have that ldp config in my PE's, I have LFIB
"Unlabelled" entries in my PE, I guess since I have no LDP config in the
transit P nodes. But in XE since I can remove that ldp config I no longer
have Unlabelled lfib entries and a nice clean lfib with only the L3vpn
aggregate label





-Aaron









_______________________________________________
cisco-nsp mailing list cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: RSVP-TE (MPLS-TE) and LDP question [ In reply to ]
On Mon, May 11, 2020 at 01:02:23PM -0500, Aaron Gould wrote:
> Seems that when I try to use RSVP in place of LDP for label distribution, I
> cannot completely remove mpls ldp configs from IOS XR, but I can from IOS XE

It's an implementation 'bug' on IOS XR.

If you have L3VPN type service (also affects labeled-ucast, including 6PE), you *must*
have 'mpls ldp' and router-id configured at minimum, even if you are not using any LDP
adjacency whatsoever. I believe ldp process needs to run to allocate labels for l3vpn,
even if you do not use LDP transport.

So, just leave 'mpls ldp' and router-id configured below it. As long as you don't
have LDP adjacencies defined, and there are no LDP tunnels configured, you won't have
any LDP in use.

P routers are not affected, as they do not need to allocate labels for VPN services.


James
_______________________________________________
cisco-nsp mailing list cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: RSVP-TE (MPLS-TE) and LDP question [ In reply to ]
Thanks James for the confirmation as that's precisely what I'm seeing.
Would be nice to see a link to a cisco document or someone out there online
that speaks to this

-Aaron

-----Original Message-----
From: James Jun [mailto:james@towardex.com]
Sent: Monday, May 11, 2020 3:26 PM
To: Aaron Gould
Cc: cisco-nsp@puck.nether.net
Subject: Re: [c-nsp] RSVP-TE (MPLS-TE) and LDP question

On Mon, May 11, 2020 at 01:02:23PM -0500, Aaron Gould wrote:
> Seems that when I try to use RSVP in place of LDP for label distribution,
I
> cannot completely remove mpls ldp configs from IOS XR, but I can from IOS
XE

It's an implementation 'bug' on IOS XR.

If you have L3VPN type service (also affects labeled-ucast, including 6PE),
you *must*
have 'mpls ldp' and router-id configured at minimum, even if you are not
using any LDP
adjacency whatsoever. I believe ldp process needs to run to allocate labels
for l3vpn,
even if you do not use LDP transport.

So, just leave 'mpls ldp' and router-id configured below it. As long as you
don't
have LDP adjacencies defined, and there are no LDP tunnels configured, you
won't have
any LDP in use.

P routers are not affected, as they do not need to allocate labels for VPN
services.


James

_______________________________________________
cisco-nsp mailing list cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/