[c-nsp] SERVICES DOWN DUE FAILURE TO SWITCHING

[emmanuel manoni]

Hi all,

There was this failure in SP network I work for which I can't make sense of

Seamless MPLS has been deployed Networkwide, There is a cluster of access
layer routers,let me take one of them as 'A'.Router 'A' is in Ospf Area
5,It has three bgp RRs,these RRs are in the same cluster ID,and they've
peered to the network main RRs.All network RRs are in the same ISIS-L2
domain.One of Area 5 RRs is connected to another Ospf Area(Area 1) with
Router 'B' which is also an RR for area 1 access routers,with both ISIS and
Ospf connectivity.Router A has some MPLS L3VPN services available in Router
B. It was working fine.Recently,there was a new direct link introduced
between Router A and B.We decided to extend Area 5 to Router B via new
direct link, without adding Router B as an RR.Traffic was flowing via that
link with no any issue. Yesterday we had a failure in the new link between
Router A and B and Traffic from A to B failed to switch to old path.Weird
thing is some services switched to old path,many others didn't,and even
weirder Router A hosting same services for different sites with same vrf
for service (Routing Table) also had other sites ok,and others not ok. I
analyzed routing at the time, nothing seemed to be wrong except some sites
were up others were not in the same area/router.And situation autorestored
itself after almost an hour.

Anyone with more insight, kindly assist.Thanks in advance.