We have two sites, A (central office) and B (remote), connected via a transparent LAN connection with ethernet handoff to our 2800 routers. They are EIGRP neighbors. I upgraded the IOS for site A to a newer T release and reloaded, but an adjacency never formed between the two sites.
Upon further investigation, it appears that the line protocol for site B on that interface actually went down. In the sh log for site B, I see the following at the time I reloaded site A:
Sep 17 07:20:15: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet0/1, changed state to down
Sep 17 07:20:15: %DUAL-5-NBRCHANGE: IP-EIGRP(0) 1: Neighbor 192.168.X.X (GigabitEthernet0/1) is down: interface down
This seems strange to me, as I would only expect the %DUAL-5-NBRCHANGE and not the %LINEPROTO-5-UPDOWN at site B. What could have caused this to happen?
A sh int g0/1 on site B gives "GigabitEthernet0/1 is up, line protocol is down." Site A's int is up/up.
My first thought when I read the suggestion from Reza was yes when EIGRP experiences a problem in establishing neighbor relationships then authentication might be a cause.
But my second though was how would EIGRP authentication have an impact on the line protocol?
The problem here is that one router has the line protocol marked as down. If the interface line protocol is down then the router will not send any packets out that interface. And that is why the EIGRP neighbor relationship is failing. So we need to focus on why the line protocol went down.
Perhaps a good starting place would be if the original poster will post the output of show interface and post the configuration of at least that interface (perhaps the entire router might be better).
I agree completely with your analysis. The interface up/line protocol down is a state typically displayed on routers when an Ethernet cable is disconnected from an interface, or if the port was unable to link with its opposite device. Something must have happened to the physical connectivity of the remote router to the Ethernet handoff. The EIGRP neighborship flap is just a direct consequence.
I would suggest the usual troubleshooting approach for physical layer problems - verifying the connectors, cables, replacing patch cords, temporarily connecting a different device (a laptop perhaps) to the handoff to see if the link comes up. From the information currently available, it is difficult to suggest what exactly might have happened.