Segment Routing TI-LFA FRR Node Protection can't be activated

[tamihirolee]

Hello Experts - just trying to get my feet wet on SR using iosxrv instances (iosxrv-k9-demo.vmdk-6.0.1) laid out in an impromptu topo: 

My goal here is to get fast-reroute node protection working just like this Cisco demo.

Before the node-protecting tiebreaker is included in pe51, each VM is configured as attached. 

pe51 reaches pe52(10.0.0.252) via p53-->p54-->p55. 

RP/0/0/CPU0:pe51#traceroute 10.0.0.252
Tue Jun 28 02:21:49.247 JST

Type escape sequence to abort.
Tracing the route to 10.0.0.252

1 10.0.13.22 [MPLS: Label 16252 Exp 0] 19 msec 9 msec 0 msec
2 10.0.34.2 [MPLS: Label 16252 Exp 0] 0 msec 0 msec 0 msec
3 10.0.45.2 [MPLS: Label 16252 Exp 0] 9 msec 0 msec 0 msec
4 10.0.25.21 9 msec * 9 msec

And the backup path is via p51-->p53-->p54-->p55.

RP/0/0/CPU0:pe51#show isis fast-reroute 10.0.0.252/32 detai$
Tue Jun 28 02:55:29.500 JST

L1 10.0.0.252/32 [50/115] medium priority
via 10.0.13.22, GigabitEthernet0/0/0/1, p53, SRGB Base: 16000, Weight: 0
FRR backup via 10.0.11.22, GigabitEthernet0/0/0/0, p51, SRGB Base: 16000, Weight: 0, Metric: 60
P: No, TM: 60, LC: No, NP: No, D: No, SRLG: Yes
src pe52.00-00, 10.0.0.252, prefix-SID index 252, R:0 N:1 P:0 E:0 V:0 L:0
L2 adv [50] native, propagated, prefix-SID index 252, R:0 N:1 P:0 E:0 V:0 L:0

Now I want to do what is done in the demo clip at 5:50 /12:07. 

But the fast-reroute tiebreaker can't be configured per isis interface address-family level.

RP/0/0/CPU0:pe51(config)#router isis 5
RP/0/0/CPU0:pe51(config-isis)#interface gigabitEthernet 0/0/0/0
RP/0/0/CPU0:pe51(config-isis-if)#address-family ipv4 unicast
RP/0/0/CPU0:pe51(config-isis-if-af)#fast-reroute per-prefix ?
exclude FRR exclusion information
level Configure FRR for one level only
lfa-candidate FRR LFA candidate information
remote-lfa Enable remote LFA computation
ti-lfa Enable TI LFA computation
<cr>

It can be configured as default though, which should be fine. 

router isis 5
address-family ipv4 unicast
fast-reroute per-prefix tiebreaker node-protecting index 100

However, the backup path doesn't shift to p51-->p52-->pe52 as expected.

RP/0/0/CPU0:pe51#show isis fast-reroute 10.0.0.252/32 detail
Tue Jun 28 02:58:52.116 JST

L1 10.0.0.252/32 [50/115] medium priority
via 10.0.13.22, GigabitEthernet0/0/0/1, p53, SRGB Base: 16000, Weight: 0
FRR backup via 10.0.11.22, GigabitEthernet0/0/0/0, p51, SRGB Base: 16000, Weight: 0, Metric: 60
P: No, TM: 60, LC: No, NP: No, D: No, SRLG: Yes
src pe52.00-00, 10.0.0.252, prefix-SID index 252, R:0 N:1 P:0 E:0 V:0 L:0
L2 adv [50] native, propagated, prefix-SID index 252, R:0 N:1 P:0 E:0 V:0 L:0

I guess I'm missing something obvious. Please advise, it'd be much appreciated! 

[tamihirolee]

After posting the original question I've set isis on each router to level-2-only, following the example config in the demo movie. Also isis Lo0 interface has been set to passive, which I simply forgot in the previously posted configurations. Still no luck..

RP/0/0/CPU0:pe51#sh isis fast-reroute 10.0.0.252/32 detail
Tue Jun 28 09:58:32.446 JST

L2 10.0.0.252/32 [40/115] medium priority
via 10.0.13.22, GigabitEthernet0/0/0/1, p53, SRGB Base: 16000, Weight: 0
FRR backup via 10.0.11.22, GigabitEthernet0/0/0/0, p51, SRGB Base: 16000, Weight: 0, Metric: 50
P: No, TM: 50, LC: No, NP: No, D: No, SRLG: Yes
src pe52.00-00, 10.0.0.252, prefix-SID index 252, R:0 N:1 P:0 E:0 V:0 L:0

I'd be grateful for any advice and/or suggestions! 

[tamihirolee]

Whether no response means nobody cares or not, let me share some updates.

In the above topology, I have shut down the interface on p53 connected to p51. Now the backup path from pe51 to pe52 is via p51-->p52.

RP/0/0/CPU0:pe51#show isis ipv4 fast-reroute 10.0.0.252/32 detail 
Sun Jul 24 02:27:41.672 JST

L2 10.0.0.252/32 [40/115] medium priority
     via 10.0.13.22, GigabitEthernet0/0/0/1, p53, SRGB Base: 16000, Weight: 0
       TI-LFA backup via p52 (PQ) [10.0.0.52]
       via 10.0.11.22, GigabitEthernet0/0/0/0 p51, SRGB Base: 16000, Weight: 0, Metric: 60
       Label stack [16052, 16252]
       P: No, TM: 60, LC: No, NP: No, D: No, SRLG: Yes
     src pe52.00-00, 10.0.0.252, prefix-SID index 252, R:0 N:1 P:0 E:0 V:0 L:0

But as far as p52 is used as a PQ node, isn't it what is called rLFA instead of TI-LFA?

My mind boggles...

[kmichiel · Accepted solution]

First, TILFA node and SRLG protection will be available in IOS XR 6.1.1, released around end of summer.

Second, even though TILFA re-uses some of the concepts of LFA and rLFA, significant enhancements have been introduced with TILFA.

TILFA provides a guaranteed protection in all topologies for link, node, and SRLG failures (node and SRLG in IOS XR 6.1.1), therefore the name "Topology Independent". TILFA does not use any additional signalling to achieve this. This means that e.g. no targeted LDP session is used between the Point of Local Repair (PLR) and the PQ-node.

TILFA also automatically provides the most optimal repair path: the post-convergence path. The TILFA repair path is tailored over the path that the traffic will follow after IGP eventually converged. In the example the selected PQ-node is located on the post-convergence path.

[tamihirolee]

OK kmichiel I'll wait a few more months for the release of 6.1.1, and give it another shot. Thanks a lot for the clarification!

[tamihirolee]

Much belated, but let me share the positive results of my follow-up testing using iosxrv-k9-demo-6.1.2.vmdk. 

On pe51, the backup path to pe52 follows p51 -> p52 as expected.

RP/0/0/CPU0:pe51#show isis fast-reroute 10.0.0.252/32 detail
Mon Jan 30 13:25:13.213 JST

L1 10.0.0.252/32 [50/115] medium priority
     via 10.0.13.22, GigabitEthernet0/0/0/1, p53, SRGB Base: 16000, Weight: 0
         Backup path: TI-LFA (node), via 10.0.11.22, GigabitEthernet0/0/0/0 p51, SRGB Base: 16000, Weight: 0
           P node: p51.00 [10.0.0.51], Label: ImpNull
           Q node: p52.00 [10.0.0.52], Label: 24004
           Prefix label: 16252
       P: No, TM: 60, LC: No, NP: No, D: No, SRLG: Yes
     src pe52.00-00, 10.0.0.252, prefix-SID index 252, R:0 N:1 P:0 E:0 V:0 L:0
   L2 adv [50] native, propagated, prefix-SID index 252, R:0 N:1 P:0 E:0 V:0 L:0

Forwarding label 24004 on p51:

RP/0/0/CPU0:p51#show mpls forwarding labels 24004 detail
Mon Jan 30 13:26:12.453 JST
Local  Outgoing    Prefix             Outgoing     Next Hop        Bytes
Label  Label       or ID              Interface                    Switched
------ ----------- ------------------ ------------ --------------- ------------
24004  Pop         SR Adj (idx 0)     Gi0/0/0/1    10.0.12.2       0
     Updated: Jan 30 12:07:58.595
     Path Flags: 0x400 [  BKUP-IDX:1 (0xa0e86554) ]
     Version: 23, Priority: 1
     Label Stack (Top -> Bottom): { Imp-Null }
     NHID: 0x0, Encap-ID: N/A, Path idx: 0, Backup path idx: 1, Weight: 40
     MAC/Encaps: 14/14, MTU: 1500
     Packets Switched: 0

       16052       SR Adj (idx 0)     Gi0/0/0/2    10.0.13.2       0            (!)
     Updated: Jan 30 12:07:58.595
     Path Flags: 0x100 [  BKUP, NoFwd ]
     Version: 23, Priority: 1
     Label Stack (Top -> Bottom): { 16052 }
     NHID: 0x0, Encap-ID: N/A, Path idx: 1, Backup path idx: 0, Weight: 10
     MAC/Encaps: 14/18, MTU: 1500
     Packets Switched: 0
     (!): FRR pure backup

This is pretty cool.. at least me thinks. 

[smailmilak]

Hi,

TI-LFA supports IPv6 as well, not only IPv4. Correct?

I see that I can enable it on xrv, but in the config guide for 6.3, even 7.1.x it says TI-LFA support IPv4 only