Hi Satya,

I believe the address Michael is using in the route-map to set nh is one of
the local addresses on the very ASBR.

Thx,
R.

On Thu, Dec 12, 2019 at 8:25 PM Satya Mohanty (satyamoh) <[email protected]>
wrote:

> Hi,
>
>
>
> If the next-hop is set to some third-party device at the option-B ASBR and
> a new label is advertised, then we have to make sure that at that
> third-party device, the particular prefix is in fact bound to that
> particular label. Therefore, in addition to setting the next-hop at the
> ASBR there needs to be a way to specify the particular label also.
>
>
>
> Otherwise just using any dynamic label will not work.
>
>
>
> Thanks,
>
> --Satya
>
>
>
>
>
> *From: *BESS <[email protected]> on behalf of Gyan Mishra <
> [email protected]>
> *Date: *Thursday, December 12, 2019 at 11:01 AM
> *To: *michael walden <[email protected]>
> *Cc: *"[email protected]" <[email protected]>, Robert Raszuk <[email protected]>
> *Subject: *Re: [bess] RFC 8277
>
>
>
>
>
>
>
> I can try it out in VIRL.
>
>
>
> What version of XR or XE are you running?
>
>
>
> On Thu, Dec 12, 2019 at 1:52 PM michael walden <[email protected]>
> wrote:
>
> Thank you for the responses.  This is specifically not using next hop self
> to change the next hop, but using a route map or RPL to set next hop to a
> specific IP value such as below.  I realize this probably isn't a common
> configuration.
>
>
>
> route-map NH permit 10
>  match ip address prefix-list ONE
>  set ip next-hop x.x.x.x
>
>
>
> route-policy NH
>   if destination in (1.1.1.1/32) then
>     set next-hop x.x.x.x
>
>
>
> On Thu, Dec 12, 2019 at 12:48 PM Gyan Mishra <[email protected]>
> wrote:
>
>
>
>
>
> With option b since the LSP is segmented into 3 segments
>
>
>
> Left side Rx to R1 -ibgp
>
> Middle R1 to R2 - ebgp vpnv4
>
> Right side R2 to R3 - ibgp
>
>
>
> So for the LSP to be segmented in opt b for it to work properly what is
> required is either next-hop-self on the ibgp peer so the rewrite happens
> forcing the label to change so the lsp is segmented on left side and right
> side from the middle.  The reason also for the segmentation in opt B is we
> are not importing the  SP loops between each other as done with opt c so
> the segmentation of the lsp has to happen for opt b to work.
>
>
>
> If you are doing this on Cisco XR it requires next hop static /32  route
> for MPLS forwarding to happen on the ebgp vpnv4 session where with XE
> automatically installs “MPLS forwarding” when ebgp vpnv4 is configured.
>
>
>
> If the rewrite is not happening to segment the lsp generate a new label
> then sounds like you are hitting a big.
>
>
>
> The inter as opt a b c ab are all IETF standards and the functionality and
> behavior should be the same across all vendors.
>
>
>
> Cheers,
>
>
>
> Gyan
>
>
>
> On Thu, Dec 12, 2019 at 11:55 AM Robert Raszuk <[email protected]> wrote:
>
> /* replacing IETF mailer with BESS */
>
>
>
> Hi Michael,
>
>
>
> Clearly you have discovered a bug in first implementation. Second
> implementation "other platform" works correctly.
>
>
>
> I assume you are doing next hop self on R2. Advertising labels have only
> local significance to a box which is listed as next hop in the NLRIs.
>
>
>
> Btw this behaviour did not really change from original RFC3107 so it is
> quite surprising you would still be running into such issues.
>
>
>
> Thx,
> R.
>
>
>
> On Thu, Dec 12, 2019 at 5:27 PM michael walden <[email protected]>
> wrote:
>
>
> Regarding RFC 8277 Section 3.2.2
> 3.2.2.  When the Next Hop Field Is Changed
>
>    If the Network Address of Next Hop field is changed before a SAFI-4
>    or SAFI-128 route is propagated, the Label field(s) of the propagated
>    route MUST contain the label(s) that is (are) bound to the prefix at
>    the new next hop.
>
>
> "the Label field(s) of the propagated route MUST contain the label(s) that
> is (are) bound to the prefix at the new next hop."
>
> This specification should apply when the next hop field of the route is
> changed by any means such as route map, route policies, or next-hop-self
> and propagated?
>
> I've experienced differing behaviors between platforms as to whether or
> not a the label is changed when the next hop field is changed with a route
> map or route policy.
>
> For example below inter-as option b R2 receives VPNv4 route 1.1.1.1/32
> with label 1 from R1.  R2 changes the next hop field with an outbound route
> policy but doesn't replace the label before propagating the route to R3.
> R3 receives VPNv4 route 1.1.1.1/32 and original label 1 breaking the LSP.
>
> R1-----eBGP- VPNv4-----R2-----iBGP VPNv4-----R3
>
>
> However on other platforms inter-as option b R2 receives VPNv4 route
> 1.1.1.1/32 with label 1 from R1.  R2 changes the next hop field with an
> outbound route policy and does replace the label before propagating the
> route to R3.  R3 receives VPNv4 route 1.1..1.1/32 <http://1.1.1.1/32> and
> new label 2.
>
> R1-----eBGP- VPNv4-----R2-----iBGP VPNv4-----R3
>
>
>
>
>
>
> _______________________________________________
> BESS mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/bess
>
> --
>
> Gyan S. Mishra
>
> IT Network Engineering & Technology
>
> Verizon Communications Inc. (VZ)
>
> 13101 Columbia Pike
> <https://www.google.com/maps/search/13101+Columbia+Pike?entry=gmail&source=g>
> FDC1 3rd Floor
>
> Silver Spring, MD 20904
>
> United States
>
> Phone: 301 502-1347
>
> Email: [email protected]
>
> www.linkedin.com/in/networking-technologies-consultant
>
>
>
> --
>
> Gyan S. Mishra
>
> IT Network Engineering & Technology
>
> Verizon Communications Inc. (VZ)
>
> 13101 Columbia Pike FDC1 3rd Floor
>
> Silver Spring, MD 20904
>
> United States
>
> Phone: 301 502-1347
>
> Email: [email protected]
>
> www.linkedin.com/in/networking-technologies-consultant
>
>
>
_______________________________________________
BESS mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/bess

Reply via email to