Binding error/bug ?

[Simon Hobson]

Lars Jacobsen wrote:

>According to previous information on the list, IP leases are bound to
>either Client ID and/or MAC. But witch MAC ?
>
>I have noticed that if the discover packet (like the one at the bottom)
>witch is coming though a relay agent, the IP is being bound to the relay
>agent's MAC, and not the MAC listed in the option 61. The information
>from option 12 (Hostname) is also recorded in the lease file.
>
>But shouldn't the lease file (DHCPd) bind the IP to the "real Client
>MAC" contained in option 61, not the "Client MAC" that's sending the
>packet to the DHCP server ?

The client mac address field in the discover packet SHOULD be that of 
the client, not the relay agent. You dhow a packet trace and 
complaint that information recorded in the lease is different, but 
you don't show us the lease so we can't see if it's wright or wrong.

Option 61 is NOT, I repeat NOT the client Mac address. It MAY be the 
client mac address, but may actually be anything the client is 
configured to use - it's just that Microsoft chose to fix it at the 
Mac address with all the problems that's been causing users ever 
since !


Do you have a packet trace for the same packet before and after it 
went though the relay agent ?


One last thought, the packet trace shows the source of the packet as 
an RFC1918 address, whilst the stated relay agent Ip address in the 
DHCP payload is a public address.

>Internet Protocol, Src Addr: 192.168.101.1 (192.168.101.1), Dst Addr:
>192.168.101.20 (192.168.101.20)

>      Relay agent IP address: 85.27.148.193 (85.27.148.193)

Is it possible that you have a proxy that is deliberately mangling 
packets, perhaps to deal with NAT or something ?