Re: [dhcwg] "client identifier" in server replies

On Thursday 10 July 2003 09:56, Bud Millwood wrote:
> If your wire protocol has absolutely no way of identifying nodes on that
> wire, the only way to use DHCP is for each node to be able to uniquely
> identify itself. Whether your node or your wire-level protocol generates
> the unique identifier, that identifier should serve as your hw address,
> IMHO. Is there anything wrong with this line of thought?

Yes, the chaddr field in the packet is used to set the address on the frame 
being sent by the DHCP server or relay agent to the client.   So you can't 
just stuff some random value in there.

I don't know why the restriction on sending client-identifier back to the 
client is in place in the protocol - can anyone explain this?   I have 
difficulty believing that a client would reject a packet containing a 
client-identifier option - clients generally just discard options they aren't 
expecting.   A way to resolve this would be to send the client identifier 
only if it appears in the parameter-request-list, though - then a client that 
wasn't expecting it would never see it.

In response to the original question, I think the way to address this is to 
write up a draft and submit it to the working group.

_______________________________________________
dhcwg mailing list
dhcwg@ietf.org
https://www1.ietf.org/mailman/listinfo/dhcwg