WAN Dynamics received a message from a customer late one evening that no network services company wants to get: “We performed a code upgrade and broke connectivity to our disaster recovery site”. This customer had a datacenter interconnect (DCI) with BGP routing and VXLAN between the datacenters but after the EOS upgrade there was reachability issues between them. I jumped in to assist and at first glance, things looked good. The switch interfaces and BGP sessions were up and there was full connectivity from VTEP to VTEP in the datacenter interconnect. That said, MAC addresses were not being learned across VXLAN. Hmmm, wonder why that would be? I decided to check the switch logs for any potential clues as to what could be happening. Aha! in the log buffer there are messages about VXLAN being disabled: