OSPF Neighbor up/down

Unanswered Question
Dec 23rd, 2009
User Badges:

Hello,


I have a DMVPN network. There are two DMVPN Hubs, I am having problems on a hub which is a 2801/SecK9 router. There are two DMVPN tunnels on this routers. A few spokes are periodically getting up and down. I Believe the router is not able to handle the load. There are around 40 spokes on this router. What should i do to further investigate this issue


Thanks


Zeeshan

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
Peter Paluch Wed, 12/23/2009 - 00:37
User Badges:
  • Cisco Employee,

Hello Zeeshan,


If you suspect that the CPU load is the issue, verify it using the show processes cpu sorted command if there is any current process exploiting the CPU. Also note the average CPU load during the last 1 minute and last 5 minutes. If the average loads go over 50% then there indeed may be a problem with the router being too busy. The command show processes cpu history is also useful - it displays the history of your CPU load in an ASCII-art graph. Watch for the level of load depicted by the # sign - that is the average value determined over a longer period and it can be called the normal operations baseline. If this operational baseline is over 50% then it is a sign of a potentially overloaded router. Also, spikes of high load (above 50%) that have a longer duration are indications of a possible cause of your problem. This document may be interesting for you:


http://www.cisco.com/en/US/docs/switches/lan/catalyst3750/software/troubleshooting/cpu_util.html


However, you shouldn't automatically assume that if the load is high, it is the certain cause of your problem. You are saying that your OSPF adjacencies are flapping. Is it possible to confirm using the show ip ospf neighbor command that, say, the hello packets have not arrived in time? What is the cause of the flap according to the console/logging message? Also, do only the same routers flap again and again, or does the flapping occur randomly on all DMVPN spokes? If the CPU load is high indeed on the hub router then it should influence all adjacencies, not just a selected and stable group of them. If only some neighbors flap then I would suggest having a closer look at the spoke router - perhaps it is his load that is causing problems, not the hub router.


Let's try to narrow down the possible causes of the problem.


Best regards,

Peter

Muhammad Zeesha... Wed, 12/23/2009 - 02:16
User Badges:

Peter,


Thanks for the reply. The average CPU is around 20-30%.

OSPF adjacencies are flapping randomly. One more thing to add, the physical interfaces of the two DMVPN tunnels are from an HWIC-4ESW card. Can it be problem?. Its a layer 2 card and I have created logical vlan interfaces to be used as DMVPN tunnel source.



Zeeshan


Message was edited by: sr1405268

Peter Paluch Wed, 12/23/2009 - 14:25
User Badges:
  • Cisco Employee,

Hello Zeeshan,


Hmm, the 20% to 30% CPU load should not be a definitive problem, nor should be using the SVI interfaces to provide the tunnel source address. Can you at least identify spikes of high CPU load (over 50%) in the history, even short-termed, or are there none?


Can you perhaps post here the configuration of your tunnel interfaces both on the hub router and some of the spokes that experience the flapping? I would also be interested in seeing the logging messages that accompany the OSPF adjacency flap.


Best regards,

Peter

Actions

This Discussion