[c-nsp] weird OSPF behavior

[Oliver Boehmer (oboehmer)]

Steven Fischer <> wrote on Thursday, February 26, 2009 23:26:

> I am seeing weird behavior on OSPF between a 2811 Router, and a 4510R
> switch.  A number of google searches on this came up empty.  It
> appears as if OSPF is dropping with the following message on hourly
> intervals, sometimes one hour, sometime two hours, sometimes three
> hours. 005840: Feb 26 15:28:05.415 EST: %OSPF-5-ADJCHG: Process 300,
> Nbr 192.168.208.1 on Vlan208 from FULL to DOWN, Neighbor Down: Dead
> timer expired
> 
> This message came from the 4510R
> 
> 4510R
> interface Vlan208
>  ip ospf dead-interval minimal hello-multiplier 4
[...]

you are running aggressive hellos against a relatively "small" 2811
router which uses the CPU for both control and forwarding plane, so
chances are the 2811 is too busy switching packets or doing something
else than generating OSPF hellos at that pace.
With BFD not being available on the 4500, you can only 

a) incresae the interval to be less aggressive (recommended)
b) try to tune the scheduler on the 2811 (for example "scheduler
allocate 3000 1000") and see if this helps. This will allocate more CPU
time to non-interrupt tasks (i.e. control plane), this could have minor
effects on forwarding..

(b) might help to reduce the number of flaps, you will likely never be
able to get rid of them completely in such a setup.

	oli