DHCP support for Infiniband

[Eli Cohen]

On Thu, 2006-03-16 at 08:34 -0800, David W. Hankins wrote:
> http://tools.ietf.org/html/draft-ietf-ipoib-dhcp-over-infiniband-10.txt
> 
> I'm not seeing anything in there that relies on the server to echo the
> client id back.
> 
> You're right that it's possible that the XID may collide.  It's only
> a 16 bit field, there are only 65536 possible values, and worse, some
> implementations may fill xid with pseudo-static information (ISC DHCP
> dhclient fills 'xid' with a hash of the address of a memory structure
> the client is using at the time) - but in its defense, it is also in
> the universe where chaddr is always verifiably itself, we don't support
> infiniband, USB, or Firewire yet.
> 
> 2 hits in a field of 65536 where you probably have no more than a few
> hundred clients on the same broadcast domain is a pretty hard thing to
> reach in the general case.
> 
> 
> If you have any thoughts on the subject, I encourage you to take it
> up with the DHC or IPOIB workgroups.
> 
> I'm sorry I don't know a lot about Infiniband (or even the related
> problems on Firewire and USB).
> 

Thanks for your response.