Wired Intelligent Edge

 View Only
Expand all | Collapse all

VXLAN Route reachability from within the switch itself

This thread has been viewed 18 times
  • 1.  VXLAN Route reachability from within the switch itself

    Posted Aug 13, 2021 11:05 AM
    Hi all,

    I've noticed some strange behaviour in my Aruba-CX environment. Since I didn't find anything about it in documentation I am sharing it here, to see what you guys think.  It could end up as a TAC case, If it does I'll share the results.

    I have an EVPN-VXLAN setup with  a couple of 8325 as spines , two VSX sets of 8325's as Leaves and a set of 6405 as Leaves.

    Some vlans are shared between all vteps, and everything works as expected there. On the 6405 VSX  VTEP' s I have a BGP connection to my firewall, from which I get a default route. Both 6405 have their own /31 to the firewall.. The default route is learned by the other VTEP's. All connected hosts can use the default route to reach anything beyond the firewall.

    The only thing that does not work, is traffic from the leaves themselves to VXLAN L3 only routes (such as the default route).
    So If I try to ping  google from one of the 8325 leaves, I get a ' network unreachable' message, even though a default route is present in the VRF. All hosts connected in that VRF can do that ping without issues.

    Another behaviour is with traceroute. If I traceroute from a host connected to the 8325' s, the trace stops Immediately because the switch reports: Destination net unreacheble.


    It seems to me that the underlying Linux OS cannot ' see'  the VXLAN routes.

    Is this expected behaviour or did I make a mistake in my configuration somewhere?



    ------------------------------
    Jelmer Hartman
    ------------------------------


  • 2.  RE: VXLAN Route reachability from within the switch itself

    Posted 9 hours ago

    Hi Jalmer !

    I'm currently facing the same issue. Were you able to find a solution? If so, I'd really appreciate it if you could share what worked for you. Thanks!