OSPF external routes churn

[Akhtar Samo]

Hello all,

I am observing some strange behaviour related to the routing table, almost all external routes and some inter-area routes are getting refreshed every 10 seconds.

I am getting more than 1000 entries after running 'sh ip route | i 00:00:0', these external routes are being advertised by a neighbor 6500 which redistributing these static routes.

Platform is 6500 with SUP-720

Regards,

Akhtar

[Raju Sekharan]

let us concentrate one prefix to troubleshoot

Can you check if the LSA asge also getting refreshed

"Show ip ospf database external < prefix>" should give the LS Age

Then paste the output of

1. show ip route

2. Show ip cef

3. Show log

[Akhtar Samo]

Finally we were able to figure out the problem.

'show ip ospf statistics' showed that area 107 is having increasing number of SPF

  Area 0: SPF algorithm executed 79 times

  Area 100: SPF algorithm executed 668 times

  Area 106: SPF algorithm executed 120 times

  Area 107: SPF algorithm executed 136926 times

'show ip ospf statistics detail' showed which LSAs is causing SPFs to trigger.

Area 107: SPF algorithm executed 136926 times

SPF 136917 executed 00:10:00 ago, SPF type Incremental

  SPF calculation time (in msec):

SPT    Intra  D-Intr Summ   D-Summ Ext7   D-Ext7 Total

  0      0      8      0      0      0      0      8

  RIB manipulation time (in msec):

  RIB Update    RIB Delete

  5             0            

  LSIDs processed R:0 N:1 Stub:84 SN:0 SA:0 X7:0

  Change record

  LSIDs changed 6

  Changed LSAs. Recorded is LS ID and LS type:

  x.x.x.x(N)x.x.x.x(N)x.x.x.x(N)x.x.x.x(N)

Finally tracked down those LSAs and found that flood war messages were appearing on an internal router in area 107, which was due to overlapping IPs between two internal routers. So that was fixed out. Now SPF runs in area 107 have normalized and 10 sec route refresh for external routes is not happening on the ABR.

Thing which i am not able to relate is why SPF runs in area 107 was causing 10 sec route refresh in area 0 for external routes (External routes are redistributed in area 0)

Regards,

Akhtar

[Giuseppe Larosa]

Hello Akhtar,

>> Thing which i am not able to relate is why SPF runs in area 107 was causing 10 sec route refresh in area 0 for external routes

external routes are generated by an ASBR node. The validity of the external routes is checked looking at the LSA type 4 ASBR summary LSA.

It is likey that the flood war involved the ASBR or the ABR of area 107.

With the LSA type 4 describing the ASBR flapping the OSPF external routes originated by the same ASBR were continously installed and removed from the IP routing table.

Edit:

if a forwarding address is set in the OSPF external routes routing to the forwarding address is checked too. So the affected device could also be the forwarding address used by those flapping OSPF external routes.

Hope to help

Giuseppe