ospf flapping

redboynono · ‎07-21-2008

we find some ospf notification log on our core-switch like below:

Jul 21 07:15:56.167: %OSPF-5-ADJCHG: Process 1, Nbr 202.165.96.216 on Port-channel3 from FULL to DOWN, Neighbor Down: Dead timer expired

Jul 21 07:16:03.227: %OSPF-5-ADJCHG: Process 1, Nbr 202.165.96.216 on Port-channel3 from LOADING to FULL, Loading Done

anyone had seen it before ??

Ryan Carretta · ‎07-21-2008

Dead timer expired suggests that there were lost hello packets between the two switches. Check to see if the link(s) between these OSPF neighbors is congested and/or if there is (or was) a high-cpu condition on either device.

shiva_ial · ‎07-21-2008

hi,

from the output what i could see

is preobelm occured in ospf relationship,

which makes the portchannel down at 15:56

means this neighbor is not reachable from

the router you have taken log.

it was again reachable only by 16:03

in simple, 7 mins no ospf relationship between the router you have taken log and its

neighbor 202.165.96.216

rgds/shiva

Ryan Carretta · ‎07-21-2008

Please upload the output of 'show process cpu history' from both devices.

Giuseppe Larosa · ‎07-22-2008

Hello shiva,

are only 7 seconds just the time to rebuild the OSPF adjacency.

Down doesn't mean that the portchannel went down, but refers to ospf neighbor state machine from Full -> down because DEAD timer expired means router was not able to receive ospf hello during last 40 seconds. (with default timer values)

Jul 21 07:15:56.167: %OSPF-5-ADJCHG: Proces(s 1, Nbr 202.165.96.216 on Port-channel3 from FULL to DOWN, Neighbor Down: Dead timer expired

Jul 21 07:16:03.227: %OSPF-5-ADJCHG: Process 1, Nbr 202.165.96.216 on Port-channel3 from LOADING to FULL, Loading Done

The trouble comes if the event happens many times.

Try to investigate on how much the port-channel is used. A QoS scheduler to protect ospf traffic could be needed if the link is very busy.

hope to help

Giuseppe

redboynono · ‎07-22-2008

thank you all for your replying .