CEF Adjacency for DMVPN Tunnels Phase 3

Hi All,  for dmvpn tunnel interfaces on hub, I'm seeing punt adjacencies towards spokes.  Spokes, with their static nhrp mappings, don't have punt adjacencies.  Is this normal? 

show adjacency -- has "incomplete"

show ip cef adjacency punt -- showing tunnel interfaces

Why do the NHS's have punt adjacencies while the spokes don't?

It appears that "ip nhrp map" results in an "IP adjacency"; while dynamic nhrp mapping result in punt adjacencies.

Is this the cause and effect with regard to IOS commands?


Edit note:  using Version 15.3(1.3)T


  • Hi Martinl and All,

    Your results using iou 15.4.2T ?

    With iol 15.3.(1.3)T, Phase 3 with ipsec, I'm seeing punt adjacencies on hub, but not on spokes. 

    IOS                                  Spoke-to-Hub                             Hub-to-Spoke

    iou 15.4.2.T                      punt                                           punt

    iol 15.3(1.3)T                   attached                                     punt          <will recheck>


  • Hi Martinl and All

    I dont have a wide exporure to DMVPN with the different IOS.  Doing a sanity check on what the tables should look like when configured.  Guess it is normal that hub uses punt adjacencies.  //RandyB

  • Hi,

       The punt CEF adjacency and the %-next-hop-override are not related and irrelevant for this question. It's normal to have a punt adjacency over DMVPN for the DMVPN/tunnel subnet, as all traffic destined to those IP addresses should be process-switched (for example routing protocol traffic and NHRP traffic for that subnet).



  • Hi Martinl and All,

    Using command:  show ip route -- do see %; easy for me to overlook

    show ip route next-hop-override

    D   % [90/28288000] via, 00:27:25, Tunnel903

                          [NHO][90/1] via, 00:00:38, Tunnel903

  • Hi Cristian and All,

    Thank you all for your explanations on CEF.//RandyB

  • I will be off until September 14th


Sign In or Register to comment.