[nsp] OSPF neighbour troubles

[Ryan O'Connell]

At 15/07/2003 13:42, Xavier wrote:
>I've a problem with an OSPF area between a few Cisco routers and 
>Livingston PM3 devices.
>Since a few days, I've the following errors (every 5 mins) on a Cisco:
>
>Jul 15 14:39:31 [a.b.c.d.229.213] 12980: Jul 15 13:39:30 CET: 
>%OSPF-5-ADJCHG: Process 1, Nbr w.x.y.z on Vlan13 from EXSTART to DOWN, 
>Neighbor Down: Too many DBD retransmitions
>Jul 15 14:40:31 [a.b.c.d.229.213] 12981: Jul 15 13:40:30 CET: 
>%OSPF-5-ADJCHG: Process 1, Nbr w.x.y.z on Vlan13 from DOWN to DOWN, 
>Neighbor Down: Ignore timer expired
>
>I reseted the OSPF process on all devices, same problem. When I shut down 
>OSPF on w.x.y.z, I get the same error but coming from another one!

There's a few likely problems. Basically, the DBD packets are being 
filtered/dropped for some reason such as:
1) MTU set incorrectly on the interface. All packets exchanged before the 
DBD are quite small, the DBDs can be larger. Try a 1500 byte ping across 
the interface to test.

2) Broken hardware, causing (1) above. (Try ping, as before)

3) Access-list. (This one bit me two nights ago) If you put an ACL in place 
to filter traffic, you can end up keeping an existing session up (Hellos 
are exchanged via multicast) but if you accidentally block unicast OSPF 
between the boxes you'll get exactly the symptoms described next time the 
OSPF session tries to come up. (So you think "Nothing's changed" - but it 
has since OSPF last came up)