Gerardo
The situation you describe is a bit unusual. The way to find the best answer would be to run debug ARP on RTR1.
But I have a theory about what may be happening that may explain the behavior. RTR1 will ARP for address 206.24.35.11 and in the ARP packet the source address will be the primary address of RTR1. In this situation both RTR2 and RTR3 will hear the ARP request and I believe that both RTR2 and RTR3 will send responses.
It is clear why RTR3 sends a response but why does RTR2 send a response? I believe that this is because the feature of proxy ARP is enabled on Cisco router interfaces by default. The principle of proxy ARP is that if a router receives an ARP request for a "remote" destination and the router has information about how to get to the remote subnet the router will send an ARP response with its own MAC address. I believe that RTR2 is responding to the ARP request for 206.24.35.11 with its own MAC address based on proxy ARP.
If RTR1 sends the ARP and receives two responses it will choose one to use. In this case it has chosen to use the response from RTR2 where you would prefer it to choose the one from RTR3.
Part of what complicates this situation is that you have one segment with two subnets. But only one router is aware of both subnets. If RTR2 knew of both subnets it would not send an ARP response for 206.24.35.11. It is Best Practice in dealing with multiple subnets on a segment that all routers on that segment should have all subnets defined using secondary addressing.
HTH
Rick
HTH
Rick
