OSPF help. Looping log messages?

dj_oberry · ‎07-31-2008

Hello all. I have a setup with a Cisco router and 6 dialup access units (portmaster and ascend max units). I run OSPF. All the MTU sizes are the same. However, I am seeing the following error messages (ip information removed)

Jul 31 11:26:00: %OSPF-5-ADJCHG: Process 1, Nbr xx.xx.xx.xx on GigabitEthernet0/1 from FULL to DOWN, Neighbor Down: Too many retransmissions

I see this message for three different units, all portmasters. Then I get this message for all three:

Jul 31 11:54:53: %OSPF-5-ADJCHG: Process 1, Nbr xx.xx.xx.xx on GigabitEthernet0/1 from DOWN to DOWN, Neighbor Down: Ignore timer expired

Then the following message:

Jul 31 11:56:23: %OSPF-5-ADJCHG: Process 1, Nbr xx.xx.xx.xx on GigabitEthernet0/1 from LOADING to FULL, Loading Done

Now so far my dial-up customers havent called in complaining. But this just keeps looping repeatedly on those three units. It worries me. Any ideas? As I put earlier, I thought maybe some sort of mtu issue but that hasnt changed and they are set to same number.

any information would be much appreciated.

rkhalil · ‎07-31-2008

It sounds to me like there is a connectivity problem that is dropping packets, making it necessary to re-transmit. Check layer 2 for possible duplex mismatch and check interfaces for error statistics.

As first, can you ping all these routers from cisco? Do you have lost packets? Can you do some debugging (debug ip ospf adj/events/retr)?

dj_oberry · ‎07-31-2008

Hmm, you may be right. I'll check the switch and compare with the devices. I'm able to ping the devices. With this still going on, i did ping tests to each unit (100 ping setup) and see no packet loss. The router interface is clean. Debugging only seems to show me this:

Jul 31 11:08:25: OSPF: Rcv DBD from xx.xx.xx.xx on GigabitEthernet0/1 seq 0x1AF4 opt 0x2 flag 0x2 len 552 mtu 0 state EXCHANGE

several times for each of the three with issues. Then the next one completes and it shows up again. The first three messages, as well as this one, came from sho ip ospf adj. I'll try events and retr now.

Wait... I wonder if its the mtu shown in that message. It shows 0. Why is that I wonder...

rkhalil · ‎07-31-2008

If you see several times Rcv BDB EXCHANGE the router A never gets an ACK from neighbor router B because the DBD packet from router A is too large for router A MTU. and router A repeatedly retransmits the DBD packet.

You are running mpls ? try to see mtu..

show mpls interfaces Gi0/1 detail

also try...

put the command "ip ospf mtu-ignore" in the interface gi0/1

--

Raul

dj_oberry · ‎07-31-2008

Interface GigabitEthernet0/1:

IP labeling not enabled

LSP Tunnel labeling not enabled

BGP tagging not enabled

Tagging not operational

Optimum Switching Vectors:

IP to MPLS Disabled

MPLS Disabled

Fast Switching Vectors:

IP to MPLS Disabled

MPLS Disabled

Not running mpls so not sure. Dont know why the packet shows 0 mtu like that. The router is set to default, as is the portmaster. Will try the command you recommended and see how that works.